Remote Sensing PDF

REMOTE SENSING
TUTORIAL
By:
Francisco Eugenio González
Marcello Javier Ruiz
Ferran Marqués Acosta
PREFACE
The project "Programme for the Development of Technological Networks

and Application of Remote Sensing Data in West Africa" -with acronym
TELECAN and code MAC/3/C181- was approved under the Second Call
for Transnational Cooperation Programme Madeira-Azores-Canaries
(MAC) 2007/2013, within Axis 3 - Cooperation with Third Countries and the
Wider Neighborhood Joint. It has been promoted by the University of Las
Palmas of Gran Canaria (ULPGC) and the Society of Economic Promotion
of Gran Canaria (SPEGC), and 85% of its budget is co-financed by the
European Regional Development Fund.
TELECAN is a three-year project led by the University of Las Palmas of

Gran Canaria which aims at developing remote sensing products and
services in strategic sectors in the region of the Canary Islands and West
Africa through collaborative work between academic institutions in the
Canary Islands, Morocco, Mauritania, Senegal and Cape Verde.
This Tutorial for Education and Training is one of the strategic lines of
action that the project TELECAN wants to make available to all
organizations, businesses or home users interested in remote sensing in
order to encourage its use to improve the quality and competitiveness of
products generated and to promote consolidation or creation of
technology-based companies.
TELECAN Tutorial
Introduction
2
This Remote Sensing Tutorial has a technical and business orientation,
with the overall objective of stimulating the use of remote sensing and
creation of products and value added services. In addition, it aims at
improving the training and skills of the project partners and Canary
business & research in remote sensing. Further specific objectives are:
 Provide knowledge of space remote sensing, physical fundamentals,

sensors and existing missions and land, atmospheric and marine
products that could be obtained.
 Acquire knowledge on the techniques of digital image processing

remote sensing aimed at improving and extracting relevant
information for each application.
 Know the various data sources available for imaging.
 Know and properly handle the major existing software tools.
 Acquire the skills and abilities to generate remote sensing products

and services and to extract as much information as possible from
Earth observation satellite images.
TELECAN Tutorial
Introduction
3
Our needs may be grouped up focusing on training needs and breaking
down the fundamental objective of this tutorial:
 Encourage the use of remote sensing.

 Facilitate the exchange of resources and knowledge of the subject.
 Meeting the training needs of users (continuos education).
 Extend the use to new users (companies and corporations).
 Establish common learning objectives.
Finally, note that the TELECAN TUTORIAL presented below contains the
efforts and contributions of the universities of the Canary Islands, Morocco,
Mauritania, Senegal and Cape Verde Islands achieved in different project
meetings that have been held in Gran Canaria (2010), in Senegal (2011)
and in Agadir (2012), as well as those of public institutions and companies
related to the remote sensing of the Canary Islands.
TELECAN Tutorial
Introduction
4
STRUCTURE
This document entitled "REMOTE SENSING TUTORIAL TELECAN"

prepared jointly by the chief scientists of the Telecan project (Francisco
Eugenio / Javier Marcello, ULPGC and Ferran Marqués, UPC) by
agreement of the Telecan Committee, is divided into two parts: Formation
(theoretical) and Training (practical fundamentals).
FORMATION:
1. - Fundamentals of Remote Sensing
2. - Sensors and Remote Sensing Space Missions
3. - Remote Sensing Applications
4. - Radiometric, Atmospheric and Geometric Modeling
5. - Remote Sensing Images Processing
TRAINING:
1. - Collection and Analysis of Remote Sensing Images
2. - Tools Image Processing Software
3. - Modeling and Remote Sensing Image Processing
TELECAN Tutorial
Introduction
5
FORMATION PLAN
1. Remote Sensing Fundamentals
Introduction
Remote sensing is a scientific discipline that integrates a broad set of knowledge and
technologies used for observation, analysis and interpretation of terrestrial and
atmospheric phenomena. Its main information sources are the measures and the
images obtained with the help of aerial and space platforms.
As the name suggests, remote sensing implies a distance information acquisition
without direct contact with the object being studied. As you are reading this text, you are
performing yourself an act of remote perception: the light coming out a source is the
physical entity, in this case it is the screen of your PC. This light travels through a
certain distance until it is captured by a sensor, which are your eyes that send this light
to a processor, your brain.
Information acquisition at distance implies the existence of a flow of information
between the object being observed and the sensor. The bearer of this information is the
electromagnetic radiation that may be emitted by the object or come from another body
and could have been reflected by it. All bodies -planets, living beings, inanimate objects-
emit electromagnetic radiation and the amount and type of radiation emitted depends
largely on their temperature.
The current remote sensing systems, unlike the very first ones, have experienced a
quick advance, especially in the last decade, with a technology essential in monitoring
multiple processes that affect the Earth's surface and its surrounding atmosphere, large
impact, especially, for our planet, such as climate change, deforestation, desertification,
etc.
Thus, there are operational satellite systems that sample virtually every region of the
electromagnetic spectrum, with spatial resolutions from 0.5 to 5,000 m. The great
interest of the scientific community in spatio-temporal studies of global change,
environmental monitoring and human impacts on it necessarily involves the use of
remote sensing data.
Tutorial TELECAN
Fundamentals of Remote Sensing
1.2
Remote sensing systems, particularly those located on satellites, provide a repetitive
and synoptic vision of Earth which is of great interest in monitoring and analysing the
effects of human activities on our planet such as the evaluation and monitoring of the
environment -urban growth, hazardous waste-, detection and monitoring of global
changes -atmospheric ozone depletion, deforestation, global warming-, exploration,
both non-renewable resources -minerals, oil, natural gas- and renewable natural
resources -oceans, forests, land-, meteorology -weather forecasting, atmospheric
dynamic processes-, mapping -topography, land use, civil engineering, etc.
The objectives of this chapter are:
 Study of basic principles, advantages, limitations and elements of a spatial
remote sensing system.
 Know the evolution of the Earth's space observation.
 Know and understand the remote sensing physical principles.
 Discriminate the spectral information from the different land cover.
 Study the orbits of the satellites and the physical laws that govern them.
 Observe and identify satellites visible from Earth.
CONTENTS
Remote Sensing Fundamentals
A. Fundamentals of Remote Sensing
1.1. BASIC CONCEPTS: DEFINITION AND OBJECTIVES.
1.2. HISTORICAL EVOLUTION.
1.3. PHYSICAL PRINCIPLES OF REMOTE SENSING: GROUND COVERS
SPECTRAL RESPONSE.
1.4. REMOTE SENSING AGENCIES AND PUBLIC ORGANIZATIONS.
Tutorial TELECAN
1.3
1.1. BASICS: DEFINITION AND OBJECTIVES
In a globalized world citizens are informed daily on the presence

and effects of natural phenomena such as earthquakes, tsunamis,
volcanic eruptions, forest fires, floods, etc.
In addition, humans have always had a special interest in

contemplating nature and observe features of the territory.
Remote sensing is an essential tool to provide Earth

observation satellite images, suitable for multiple
desired applications.
Tutorial TELECAN
1.4
"REMOTE SENSING is an aerospace technology that
uses electromagnetic energy to capture data from
the Earth's surface and its surrounding atmosphere
by remote sensing systems"
"ISPRS: measurement or acquisition of information from

certain properties of an object or phenomenon, by a
recording system that is not in physical contact
with the object or phenomenon under study"
PASSIVE REMOTE SENSING ACTIVE REMOTE SENSING
Tutorial TELECAN
1.5
 REMOTE SENSING MAIN OBJECTIVE:
"Considerably improve our knowledge of our

environment, facilitating the interpretation of the
multiple processes affecting the planet"
Tutorial TELECAN
1.6
 REMOTE SENSING: GENERAL SCHEME OPERATION
1. - Energy Source.
2. - Land cover.
3. - Atmosphere.
4. - Sensor system.
5. - Reception System.
6. – Images Processing.
7. - Interpreter and / or user.
Tutorial TELECAN
1.7
 REMOTE SENSING: SPATIAL INFORMATION TOOL
Advantage
 Global coverage and regular large areas of the Earth.
 Observation multiscale.
Tutorial TELECAN
1.8
 Information on non-visible regions of the spectrum.
 Digital processing of the received images.
Tutorial TELECAN
1.9
Limitations
 Models are NOT obtained with high accuracy in comparison with in-situ
observations or aerial photography.
 REMOTE SENSING: COMPLEMENTARY TECHNOLOGIES
 In-situ observations
 Aerial photography
Daedalus 1268 (INTA)
Tutorial TELECAN
1.10
1.2. HISTORICAL EVOLUTION
 RELEVANT EVENTS
 The invention of photography made remote sensing possible.
 Remote sensing started in 1860 with a photograph of the Earth's surface,

taken from a balloon by Tournachin.
 First Earth observation satellite in 1960 (TIROS-I).
 There are currently multiple public and private agencies and Research and
Education Centers actively working in the field of remote sensing.
Tutorial TELECAN
1.11
Tutorial TELECAN
1.12
Tutorial TELECAN
1.13
1.3. FUNDAMENTALS OF PHYSICS: GROUND COVERS
SPECTRAL RESPONSE
 ELECTROMAGNETIC SPECTRUM: SOURCES AND USES OF THE

FREQUENCY BANDS
Tutorial TELECAN
1.14
 SPECTRAL REGIONS USED FOR REMOTE EARTH OBSERVATION
 Visible spectrum (0.4-0.7 m): It is the frequency range of the human eye.
Maximum solar radiation. Subdivided into three bands: R, G, B.
 Near-infrared (0.7-1.1 m): Also called photographic reflected IR. It is the solar
energy reflected by any body. Its behavior is similar to the visible spectrum.
 Middle infrared (1.1-8 m): Solar radiation and emission mixing. The atmosphere
is significantly affected. It is exploited to measure concentrations of water vapor,
ozone, aerosols, etc.
 Thermal infrared (8-14 m): Radiation emitted by the bodies themselves. Tp can
be determined by a body (thermal IR). Images may be available at any time of
the day.
 Microwave (1mm-1m): There is a growing interest of Remote Sensing in this

band. Atmospheric disturbances are minor and it is transparent to clouds. Active
sensors are typically used.
Tutorial TELECAN
1.15
 SPECTRAL REGIONS USED FOR REMOTE OBSERVATION OF EARTH:
ATMOSPHERIC TRANSMISSIVITY
 EMISSION RATES USED BY REMOTE SENSING
1. Radiation reflected on Earth's surface emitted by the Sun. It depends on the

characteristics of the surface or object (i.e., human vision and the environment).
2. Any body whose temperature is above absolute zero (0 K: -273 ° C) emits energy
in the form of radiation (i.e., a camp fire emission).
Tutorial TELECAN
1.16
 ELECTROMAGNETIC RADIATION LAWS
 Planck's Law
Provides the spectral radiance of a black body as a function of temperature.
Any object with T> 0K radiates energy.
T Energy (Radiation in )
 Wien's Displacement Law

For a given temperature, there is a for which the electromagnetic energy is the peak.
We can select the appropriate band for the detection of a phenomenon, if we know its
temperature range.
Tutorial TELECAN
1.17
 Stefan-Boltzmann Law
Provides the total energy emitted by a black body throughout the spectrum.
 Direct relationship between temperature and the amount of electromagnetic

energy emitted.
 The hotter is the object, the more energy is generated.
 Small variations in temperature Large variations in the energy emitted (T4).
 Real bodies are not black bodies.
Emissivity
Tutorial TELECAN
1.18
 Spectral Reflectivity
A real body has an emissivity and reflectivity that vary with wavelength  SPECTRAL
SIGNATURE
 Interactions of the Atmosphere

 The atmosphere is a mixture of gases at different layers.
 The first 80 kilometers contains more than 99% of the total mass of the Earth's
atmosphere.
Tutorial TELECAN
1.19
 When EM radiation travels through the atmosphere, it is absorbed or scattered
by particles in the atmosphere.
 The atmosphere also emitsThermal IR.
Tutorial TELECAN
1.20
1.4. REMOTE SENSING AGENCIES AND PUBLIC
ORGANIZATIONS
 MAIN AGENCIES AND ORGANIZATIONS
At regional (PET), national (PNOT-PNT), European (ESA) and global (NASA) level, the
different space agencies continue to rely on remote sensing with new missions, allowing
its application in many strategic sectors.
Tutorial TELECAN
1.21
INTERNATIONAL
1. ESA: European Space Agency
From past to present
Tutorial TELECAN
1.22
Missions for Earth Observation controlled by ESA
Scientific missions: "Earth Explorer"
Tutorial TELECAN
1.23
 GMES: The European Earth monitoring
 GMES, Global Monitoring for Environment and Security, is a joint initiative of the
European Commission and the European Space Agency.
 GMES is a network for the collection and dissemination of data on the environment and
security from space and in situ observations of the Earth.
 This system will support decision-making by public and private authorities in Europe
and back up research.
 Following the example of meteorology, GMES develops operational services, but in

other areas such as:
1. Emergency management.
2. Monitoring air quality.
3. Monitoring soil.
4. Monitoring the ocean, etc.
General Scheme
Tutorial TELECAN
1.24
Space Segment
Missions that may contribute to GMES
Tutorial TELECAN
1.25
Missions dedicated to GMES
Sentinel-1: Mission SAR band C

 Applications:
 Monitoring of sea ice and the Arctic environment.
 Marine Surveillance.
 Monitoring risks due to ground displacement.
 Maps to organize humanitarian aid in crisis
situations..
 Four nominal operating modes:

 strip map (80 km swath, 5x5 m res.).
 interferometric wide swath (250 km swath,
5x20mm.
 extra wide swath (400 km swath, 20x40 m res).
 wave (5X5 m res, images of 20x20 km).
 Helio-synchronous orbit at 693 km altitude
 Nominal 12-day revisit period
 Expected to last 7 years, having fuel for 12 years
Tutorial TELECAN
1.26
Sentinel-2: Mission super-spectral
 Applications:
 Land cover maps.

 Risk and fast hedge maps in case of emergencies.
 Vegetation and chlorophyll maps.
 Super-spectrometer with 13 bands (VNIR & SWIR)
 Spatial resolution: 10, 20 and 60 m
 Swath: 290 km
 Satellite mass: 1200 kg
 Revisit period: 5 days (without clouds) with 2 satellites
 helio-synchronous orbit at 786 km altitude
Tutorial TELECAN
1.27
Sentinel-3 Mission: Global oceanic and terrestrial soil
 Applications:
 Color and temperature of the sea surface and ground.

 Topography of the sea surface and glaciers.
 Topography of coastal areas, lakes and sea ice.
 Vegetation maps.
 Satellite mass: 1250 kg
 Helio-synchronous orbit at 814.5 km altitude
 Revisit period: 27 days
Tutorial TELECAN
1.28
Spatial data access
Tutorial TELECAN
1.29
2. NASA
Tutorial TELECAN
1.30
NATIONAL
 PNOT: National Earth Observation Satellite
Tutorial TELECAN
1.31
 INGENIO Satellite Services
 Optical high resolution multispectral images.
 1 panchromatic channel and 4 multispectral channels (R, G, B, NIR).

 Sweeping width: 60 Km
 PAN resolution: 2.5 m
 MX resolution: 10m
 > 600 images / day (2.5 million km2/día).
 Minimum 6 full annual coverage of the Spanish territory (max. 5º).
 Side entry: ± 40°.
 Min revisit time: 3 days (40° angle).
 Life cycle: 7 years.
 Mass: ± 750 Kg.
 Synchronous Helium Orbit. Height: ~ 670 km.
 LTDN ± 10:30 AM.
Tutorial TELECAN
1.32
Ground Segment
Difference between optical and radar images
Tutorial TELECAN
1.33
CANARY ISLANDS
PET’s main objective in the Canary Islands is to stimulate

development of services based on the use of remote sensing data and
to promote operational applications and services integrating all
existing capabilities.
 Integrate all Canary research groups, agencies and companies related to remote
sensing.
 Encouraging the sharing and optimization of remote sensing resources available

in the Canary Islands.
 Increase competitiveness and excellence of R & D related to remote sensing.
 Serve as a driving force for the development of new operational applications and
services based on the use of remote sensing images.
 Promote the development and creation of technology-based companies.
All these objectives aim at consolidating scientific, technological and

industrial capacity in the field of Remote Sensing in the Canary Islands
Tutorial TELECAN
1.34
 Priority areas
Tutorial TELECAN
1.35
Tutorial TELECAN
1.36
 Difusion Activities
El Hierro underwater volcano
Tutorial TELECAN
1.37
NASA MODIS RGB images multitemporal monitoring El Hierro submarine volcano
(October, 2011)
Monitoring and diffusion of information via web submarine volcano:

http://www.teledeteccioncanarias.es
Tutorial TELECAN
1.38
 Institutions:
 American Society of Photogrammetry and Remote Sensing.
 NASA Centers: Gooddard Space Flight Center, Ames Research Center,
Jet Propulsion Laboratory (JPL).
 NOAA-National Oceanographic Administration Agency.
 Canadian Center for Remote Sensing.
 Remote Sensing Society: NRSC, UK.
 Centre National d'Estudes Spatiales CNES, France.
 Institute for Aerospace Survey and Earth Sciencies: ITC, Netherlands.
 Institute for Remote Sensing Applications, Ispra, EU.
 Spanish Association of Remote Sensing, …
 University Departments:
 Environmental Research Institute of Michigan.
 Laboratory for applicatios of Remote Sensing, Purdue University, Indiana.
 Remote Sensing Unit, University of California, …
 Specialized Magazines:
 IEEE Transactions on Geoscience and Remote Sensing.
 International Journal of Remote Sensing.
 Photogrammetric Engineering and Remote Sensing.
 Remote Sensing of Environment.
 Remote Sensing Reviews, …
 International Conferences:
 IGARSS.
 American Society of Photogrammetry and Remote Sensing.
 Remote Sensing Society.
 European Signal Processing.
 IASTED, …
 Websites:
 http://www.itn.nl/ ~ bakker / noaa.html (NOAA).
 http://www.gsfc.nasa.gov/SEAWIFS.html (NASA-SEAWIFS).
 http://www.sat.dundee.ac.nk/ (Est. of Dundee, Scotland).
 http://www.eumetsat.de/eu/ (EUMETSAT).
 http:// www.eurimage.it/ (Eurimage), …
Tutorial TELECAN
1.39
2. Sensors and Space Remote
Sensing Missions
Introduction
In order to meet the needs of different data users, there are many remote sensing
systems providing a wide range of spatial, spectral and temporal parameters. Some
users may require frequent coverage with relatively low spatial resolutions
(meteorology). Other users may want the highest possible spatial resolution with
coverage repeated only infrequently (mapping), while some other users may need very
high spatial resolution and frequent coverage in addition to a quick access to the
images (military surveillance). Remote sensing data also can be used to initialize and
validate large computational models, such as global climate models (GCMs), in order to
simulate and predict changes in the Earth's environment. In this case, it may not be
necessary to achieve a high spatial resolution due to computational requirements and it
would be essential to accurately and consistently calibrate the sensor in space and
time.
Wide range of spatial, spectral and temporal parameters

The unavoidable and unstoppable technological progress occurred over the last few
years has led to the emergence and continuous improvement of on-board sensors on
space platforms. Their spatial, spectral, radiometric and time resolutions have been
improved resulting in a higher level of detail, accuracy and amount of information that
enables the use of these images in new application areas.
Thus, each day the Earth is observed by a constellation of satellites. They are our eyes
in space. These satellites are built and put into orbit by Space Agencies and
International Organizations and they provide valuable information to help us to achieve
a better understanding of our planet and to study the effects of human action on the
environment. The conditions of life on Earth may be improved with these data.
TELECAN Tutorial
Sensors and Space Remote Sensing Missions
2.2
This chapter examines the operation of devices that capture remote sensing images
and the platforms that support them. It pays special attention to the different types of
remote sensing satellites. The last part of this chapter is dedicated to review the
characteristics of remote sensing space missions more interesting at this point to the
context of this project TELECAN. Spatial remote sensing missions are designed and put
into orbit so quickly that this TUTORIAL TELECAN must necessarily be updated
regularly, using the information available from different space agencies.
 Distinguish between sensors and space platforms;
 Know the main types of sensors used in remote sensing;
 Discriminate spatial, spectral, radiometric and temporal information from
spatial-temporal characteristics of the sensors and space platforms;
 Understand the operation of satellites;
 Study the orbits of the satellites and the physical laws that govern them;
 Know the characteristics of the main space missions to observe the Earth.
CONTENTS
A. Fundamentals
2.1. SENSORS AND PLATFORMS
2.2. SENSORS RESOLUTIONS
2.3. REMOTE SENSING SPACE SYSTEMS
2.4. ORBITAL PARAMETERS: TYPES OF ORBITS
2.5. SPACE MISSIONS:
 METEOSAT SECOND GENERATION
 NOAA-METOP
 LANDSAT / SPOT
 GEOEYE
 WORLDVIEW
TELECAN Tutorial
2.3
2.1. SENSORS AND PLATFORMS
 SENSORS
 Systems used for the acquisition of the radiation emitted or

reflected.
 Payload in space-based remote sensing systems.

 Conversion of radiance in digital levels.
 Classification
1. According to the radiation source:
 Passive sensors: No light at the target, limited to receiving the

energy emitted or scattered by the target.
 Active sensors: Provide their own source of electromagnetic
radiation.
TELECAN Tutorial
2.4
 Remote Sensing forms Methods of acquiring information by a
remote sensor:
 Reflection (a).
 Emission (b).
 Emission-reflection (c).
2. Depending on the region of the spectrum used:
TELECAN Tutorial
2.5
 Electro-optical sensors: Visible and IR
Scanning Technologies AVHRR
 Line scanner
 Wiskbroom scanner
 Pushbroom scanner
MODIS
TM / ETM
+
MSS
SPOT
IKONOS
QUICKBIRD
TELECAN Tutorial
2.6
 Microwave sensors
 Passive: Microwave Radiometer
 Active:
 SAR
 Radar Altimeter
 Wind scatterometer
TELECAN Tutorial
2.7
 PLATFORMS
 Platforms at short distances
 Instrumentation radars: measure the reflectivity of objects and

surfaces.
 Tomographic systems: have biomedical and industrial
applications.
 Platforms at large distances: satellites, planes / helicopters.
 Considerations:
 Planes achieve a higher spatial resolution.
 Satellites have a higher temporal resolution and coverage.
 Factors for the selection of a platform:

 The extension of the area to be covered.
 The speed of development of the phenomena observed.
 The functional characteristics of the instruments and
sensors.
 The availability and cost of data.
TELECAN Tutorial
2.8
 Parallelism
Satellites describing a polar orbit (500-1000 km) or low altitude and

geostationary satellites (36,000 km).
Images of different types of remote sensing platforms
TELECAN Tutorial
2.9
2.2. RESOLUTION SENSORS
 SPATIAL RESOLUTION
 It specifies the size of the pixel of the remote sensing image.

 Lowest distance between objects that can be solved by the
sensor.
High spatial resolution: 0.5 - 4 m
Medium spatial resolution: 4-30 m
Low spatial resolution: 30-1000 m
TELECAN Tutorial
2.10
 Spatial resolution scales: Decreasing in resolution with the height
of the scanning sensor.
TELECAN Tutorial
2.11
 Spatial resolution variation with the angle of observation of the
sensor: Effective size of the pixel on the Earth (GIFOV) is bigger at
the extremities of the field of view than in the nadir.
TELECAN Tutorial
2.12
 SPECTRAL RESOLUTION
 Specifies the number of spectral bands in which the sensor can

capture radiation.
 The number of bands is not the only important aspect of spectral

resolution: it is also essential the position of the bands in the
electromagnetic spectrum.
TELECAN Tutorial
2.13
 RADIOMETRIC RESOLUTION
 It is the sensitivity of a detector to variations in the intensity of the

emitted, reflected or scattered electromagnetic energy that is being
detected.
 Different levels of intensity that may be discriminated by the sensor

within a band.
TELECAN Tutorial
2.14
 TEMPORAL RESOLUTION
 It specifies satellite revisit frequency to a specific location (It depends

on the latitude and the width of the swath).
 Time is important when:

 Persistent clouds offer limited direct view of the surface of the
Earth.
 Short-term phenomena (floods, oil spills, ...).
 Rapidly changing phenomena (atmospheric).
High temporal resolution: <1 day - 3 days

Medium temporal resolution: 4 - 16 days
Low temporal resolution:> 16 days
TELECAN Tutorial
2.15
Relationship between all types of resolutions: Commitment
TELECAN Tutorial
2.16
2.3. REMOTE SENSING SPACE SYSTEMS
Remote Sensing Space System Elements
Segment
spatial
Ground segment
TELECAN Tutorial
2.17
 SPACE SEGMENT
 Satellite: System placed in orbit for providing a particular application.
 Space Platform:
- Attitude control.
- Propulsion.
- Electric Power.
- TT & C (Tracking, Telemetry & Command).
- Thermal Control.
- Mechanical.
Remote Sensing Satellite Subsystems
TELECAN Tutorial
2.18
Different Space Platform Subsystems
Structure
TELECAN Tutorial
2.19
 Payload: Equipment designed to meet the specific application.
TELECAN Tutorial
2.20
 GROUND SEGMENT
 Architectures
 Flight Operation Segment Satellite

- TT & C earth stations.
- Control Centers.
 Payload Data Segment Data

- Reception Stations.
- Processing and Archive Centers.
- Topic Centres.
TELECAN Tutorial
2.21
Block diagram of Earth station and TTC
ESTACIÓN TERRENA
SEGUIMIENTO
ESTACIÓN PROGRAMADO
TERRENA
BANDA
BASE TRANSMISIÓN
CENTRO SISTEMA DE D
DE TRACKING
COMUNICACIONES
CONTROL
BANDA
BASE RECEPCIÓN
ESTACIÓN SEGUIMIENTO
TERRENA AUTOMATICO
MONITOREO FRECUENCIA
Y CONTROL Y TIEMPO
TEST E. ELÉCTRICA AIRE ACONDICIONADO
Landsat 5 & 7 Space Program Stations
TELECAN Tutorial
2.22
Examples of Earth Stations
TELECAN Tutorial
2.23
2.4. ORBITAL PARAMETERS: TYPES OF ORBITS
 ORBITAL MECHANICS
 An orbit is the trajectory described by a spatial body around the

Earth.
 The motion of the satellite around the Earth is mainly

determined by the force of attraction between two bodies.
 Newton postulated that the attractive force was proportional to

the masses and inversely proportional to the square of the
distance between them.
m
F GM
r2
where:
G = 6.672 10-11 m3Kg-1s-2 (Universal Gravitational Constante).

M = 5.974 x 1024 kg (mass of the Earth).
TELECAN Tutorial
2.24
 Cartesian Parameters
They describe the movement of the orbit using position and velocity
vectors at a given time (epoch). It is fully defined with 6 parameters -3
position and 3 speed parameters.
Space Shuttle STS66
EPOCH 11/09/94 13:00:06.210 GMT
X : -410.1438489632090 Km Vx: -6.009935717301647 Km/s

Y : 333.6855095962104 Km Vy: -1.961230487490145 Km/s
Z : -406.9395999999989 Km Vz: 4.448585999999987 Km/s
 Keplerian parameters
Information on the size, shape and orientation of the orbit in a

specified point in time.
TELECAN Tutorial
2.25
 Types of Orbits
 Each space MISSION requires a specific orbit.
Mission Type of orbit Altitude Period Tilt
Communications
Geostationary 35,786 Km (GEO) 24 hours 0°
Meteorological
Earth Resources Polar-synchronous 150-900 Km (LEO) 90 minutes 95 º
Navigation (GPS) Semi-synchronous 20,230 Km (MEO) 12 hours 55 °
28.5 º or 57
Space shuttle Low orbit 300 Km 90 minutes
º
Communication Perigee: 7971 Km

Molniya 12 hours 63.4 º
Intelligence Apogee: 45,170 km
TELECAN Tutorial
2.26
 Circular LEO, MEO, GEO
 Elliptical
TELECAN Tutorial
2.27
2.5. SPACE MISSIONS
 METEOSAT SECOND GENERATION
 MSG Program
 The MSG program includes 4 satellites, ensuring services until

2018.
 MSG-1 (Meteosat-8) was launched in 2002 and MSG-2
(Meteosat-9) in 2005.
 MSG-3 was launched in July 2012.
TELECAN Tutorial
2.28
Space Segment
Channels Central wavelength (um) Spectral band (um)

VIS 0.6 0.635 0.56 to 0.71
VIS 0.8 0.81 0.74 to 0.88
IR 1.6 1.64 1.50 - 1.78
IR 3.9 3.92 3.48 - 4.36
IR 8.7 8.70 8.30 - 9.10
IR 10.8 10.8 9.80 - 11.80
IR 12.0 12.0 11.00 - 13.00
WV 6.2 6.25 5.35 - 7.15
WV 7.3 7.35 6.85 - 7.85
IR 9.7 9.66 9.38 - 9.94
IR 13.4 13.40 12.40 - 14.40
HRV 0.5 to 0.9
TELECAN Tutorial
2.29
Ground Segment
TELECAN Tutorial
2.30
TELECAN Tutorial
2.31
Products
TELECAN Tutorial
2.32
 NOAA-METOP
EUMETSAT Polar System (EPS) provides METOP and NOAA runs

missions data
Space Segment
TELECAN Tutorial
2.33
Ground Segment
TELECAN Tutorial
2.34
Products
TELECAN Tutorial
2.35
 LANDSAT / SPOT
 NASA Satellites USGCS.
 LANDSAT 1 (1972)
 LANDSAT 2 (1975)
 LANDSAT 3 (1978)
 LANDSAT 4 (1982)
 LANDSAT 5 (1985)
 LANDSAT 6 (1993)
 LANDSAT 7 (1999)
 LDCM (Landsat 8) 2013
TELECAN Tutorial
2.36
 Landsat 8
 Spacecraft: Landsat Data Continuity Mission (LDCM).

 Launch Vehicle: Atlas V-401
 Launch Date: Feb. 11, 2013
TELECAN Tutorial
2.37
 SPOT
 SPOT (Satellite Pour l'Observation de la Terre). It is a medium /

high resolution optical satellite.
 Operated by Spot Image (Toulouse, France). Started by CNES

(Centre National d'Etudes Satiales) in the 70s.
 Launched with Ariane 2, 3 and 4.

 SPOT 1 (1986) with 10 m and 20 m MS PAN.
 SPOT 2 (1990)
 SPOT 3 (1993)
 SPOT 4 (1998)
 SPOT 5 (2002) with 2.5 m / 5 m and 10 m MS PAN.
TELECAN Tutorial
2.38
LANDSAT / SPOT Imagery
TELECAN Tutorial
2.39
 GEOEYE / WORLDVIEW
High resolution missions: High resolution satellites are very complex

instruments, with growing demand for its numerous applications in fields as
diverse as mapping, natural resource identification, risk management and
defense. The most important satellites are:
TELECAN Tutorial
2.40
 GeoEye
Space Segment
TELECAN Tutorial
2.41
Sensor Exploration Features
Imagery
TELECAN Tutorial
2.42
 Worldview 2
Space Segment
TELECAN Tutorial
2.43
 Sensor Exploration Characteristics
 Helio synchronous orbit, 770 km altitude.
 1 panchromatic channel (0.5 m), 8 multispectral channels (2m).
 Revisit period of 1.1 days (Max viewing angle 45°).
 Scene width 16.4 Km.
Imagery
TELECAN Tutorial
2.44
3. Remote
4. Radiometric, Sensing and
Atmospheric
Applications
Geometric Modeling
Introduction
Remote sensing techniques have changed dramatically since the beginning of this
science. Its spectacular development since 1858 -when the French photographer
Gaspard-Félix Tournachon raised his balloon over Paris to take the first aerial
photography- has led to the modern sophistication of sensors on satellites around the
Earth. These current sensors sample almost all possible regions of the electromagnetic
spectrum, achieving spatial resolutions below one meter. Thus, spatial remote sensing
currently allows the study of our planet by providing a synoptic and repetitive view that
allows us to obtain data in a short period of time. In return, remote sensing images can
only provide information about the most superficial layer.
In this context, remote sensing is applicable in the marine, terrestrial and atmospheric
sector, allowing a systematic analysis of many geophysical parameters of great interest
to researchers, businesses, governments and general public. The following chapter in
this tutorial focuses on the description of the main applications of remote sensing
grouped up into the three sectors mentioned above that are more interesting for the
Canary Islands and Northwestern Africa.
Regarding marine applications, the ocean is the most influential element in controlling
long-term stability of the Earth's climate. Not surprisingly it occupies 71% of the Earth
area and it involves a complex web of relationships between physical, chemical,
biological and geological processes. That is why its study should be approached from a
multidisciplinary perspective. This tutorial will explore issues related to oceanography,
coastal monitoring, marine productivity, water quality, etc. Specifically, parameters such
as sea surface temperature, sea level, salinity, winds on the surface of the oceans,
ocean currents, ocean color, water quality, bathymetry and coastal benthic classification
will be described.
When referring to the land area we will focus on continental crust, which is formed by
the continents and continental shelves. This crust has an average thickness of 35 km,
although it is only possible to extract information from the upper layers by using remote
sensing. With regard to the most important terrestrial applications analyzed by remote
sensing, and included in this tutorial, management and monitoring of soil, plant and
forest spaces, mapping, urban planning, agriculture, environment, prevention risk, as
well as defense-related applications and security can be highlighted.
TELECAN Tutorial
Remote Sensing Applications
3.2
The Earth's atmosphere is the gaseous part of the Earth. It is the most external and less
dense layer in the planet. It is comprised of one or more gases which vary in amount
depending on pressure at different heights, being oxygen (21%) and nitrogen (78%) its
main gases. Remote sensing has been a fundamental tool in the study and analysis of
local and regional atmospheric conditions for many years. In this tutorial the main
parameters associated with weather forecasting, study of gases in the atmosphere and,
specially, greenhouse disaster prevention, renewable energy, air quality, aerosol
concentration, etc. will be presented.
 Show marine applications of remote sensing space, identifying the most
important physical, biological, chemical and geological parameters and analyze
the main sensors and products available for them.
 Describe terrestrial applications of space remote sensing, identifying their main
areas of application and describing the main available products and their key
features.
 Study the atmospheric applications of remote sensing, and present operational
products for weather forecasting as well as products related to scientific research
of the atmosphere.
CONTENTS
3.1 MARINE REMOTE SENSING APPLICATIONS.
3.2 TERRESTRIAL REMOTE SENSING APPLICATIONS.
3.3 ATMOSPHERIC REMOTE SENSING APPLICATIONS.
TELECAN Tutorial
3.3
3.1. MARINE REMOTE SENSING APPLICATIONS
 INTRODUCTION
Main parameters to study with remote sensing:
Biological Oceanography Biological Oceanography
• Sea surface temperature

• Ocean color
(SST)
• Water Quality
• Sea height (SSH)
• Algae blooms
• Salinity (SSS)
• Sea Grass
• Surface Winds
• Ocean Currents
Chemical Oceanography Geological Oceanography
• Water quality • Coastal management

• Salinity (SSS) • Bathymetry (erosion,
• pH, calcite, etc. sedimentation, etc.)
TELECAN Tutorial
3.4
 SEA SURFACE TEMPERATURE (SST)
SST is the water temperature close to the ocean’s surface.
Before 1980 measures were obtained from instruments installed on the coast, boats or
buoys. From 1980 on, most global SST information comes from satellite estimates.
 The ocean emits radiation in the thermal IR and microwave band. The level
received varies with sea temperature.
 There are radiometers working in these bands to measure the SST.
 Thermal IR radiation comes from the surface layer (first 10 microns). The
microwave radiation comes from the first superficial millimeter (1 mm).
 IR sensors have a better resolution (1 km) than microwave sensors (25 km)
but they are affected by clouds that absorb the radiation emitted.
TELECAN Tutorial
3.5
 Radiometers in the thermal IR
 Methods for estimating SST multiband
The linear combination of radiances measured in two bands in the same atmospheric
window provides a good estimate of the emitted radiation.
Getting the coefficients (ai).
- Theory: radiative transfer model

- Regression: in-situ data and brightness temperatures
The great variability of the observation angle is an additional item due to difference in
the optical path to cross through the atmosphere.
TELECAN Tutorial
3.6
Sample images of sea surface temperature from thermal IR
images
 Microwave radiometers
 Advantages: The radiation is not affected by clouds and it is easier to correct

atmospheric effects.
 Disadvantages: Due to the lower signal intensity in the Planck radiation curve of
the Earth in the microwave region, accuracy and resolution are poorer for the
SST estimated in the passive microwave measurements compared to the SST
obtained from measurements using the thermal infrared. The roughness of the
sea surface generated by wind and precipitation also affect the microwave signal.
SST measurements are usually taken using the channel close to 7 GHz and a water
vapor correction thanks to observations at 21 GHz. Other frequencies used for the
correction of the roughness of the sea (including foam), precipitation and minimal
effect provoked by clouds on microwave radiation are 11, 18, and 37 GHz.
TELECAN Tutorial
3.7
 Passive microwave instruments used to obtain SST:
- Scanning Multichannel Microwave Radiometer (SMMR on Nimbus-7 and
Seasat)
- Tropical Microwave Imager (TMI on the Tropical Rainfall Measuring Mission)
- Advanced Microwave Scanning Radiometer (AMSR on Aqua and ADEOS II)
- WindSat in the Coriolis mission
Example image of sea surface temperature from microwave

radiometers
TELECAN Tutorial
3.8
 SEA HEIGHT
Sea surface topography is the height of the ocean surface relative to a level of no
motion defined by the geoid which is the shape that the surface of the oceans would
take under the influence of Earth's gravity and rotation alone, in the absence of other
influences such as winds and tides. Variations in sea surface topography can be up to 2
meters and are caused by ocean circulation, temperature and salinity.
Topography provides information on tides, circulation and distribution of heat and mass
in the Earth's global ocean.
Sea surface height (SSH) is measured using gauges or altimeters on board satellites.
Altimetry combines the precise determination of the orbit with measuring the distance to
the ocean surface by using microwave pulses. Reliable measurement is not an absolute
parameter, but Sea Surface Height Anomaly (SSHA).
 Altimeters
Altimetry satellites basically determine the distance from the satellite to the target
surface by measuring the round-trip time elapsed in a radar pulse delivered to the
surface.
Radar altimeters permanently transmit signals to Earth, and receive the echo from
the surface. The orbit of the satellites has to be known accurately (Doris system) and
their position is determined in relation to an arbitrary reference surface, i.e. an
ellipsoid.
Ku (13.6 GHz) is the frequency band used (Topex/Poseidon, Jason-1, Envisat, ERS,
etc.).
TELECAN Tutorial
3.9
Geos-3 and Seasat laid the foundation for a new generation of ocean satellites.
In the 1980s, only Geosat was launched into orbit, whose data were ranked at first.
In the 1990s, Topex/Poseidon, with a precise orbit determination and location system
as Doris allowed for better accuracy of satellite positioning and monitoring of height
sea surface variations.
Altimetry began providing vital information for a larger user community with ERS-1
and later with ERS-2.
Jason-1 and Jason-2, with a relatively short repeat cycle (10 days), allowed to
observe more often the same point in the ocean. Both satellites have a 5 day time
span.
Envisat had a longer cycle (30 days from November 2010), but closer spacing
between exploration bands (90 kilometers in the equator).
TELECAN Tutorial
3.10
TELECAN Tutorial
3.11
 Temperature and height anomalies
Application combining data from temperature and sea level: El Niño
El Niño is caused by the occasional burst of warm surface waters in the Pacific close
to Peru’s and Ecuador’s coast. El Niño brings severe weather patterns such as
droughts, floods and cyclones.
Nowadays, it is possible to predict El Niño from ocean data.
TELECAN Tutorial
3.12
El Niño - La Niña
• El Niño / La Niña are quasi-periodic patterns that occur across the tropical Pacific
Ocean roughly every five years.
• La Niña is an ocean-atmosphere phenomenon that is the counterpart of El Niño.

During La Niña, the sea surface temperature in the equatorial area of the Eastern
Central Pacific Ocean is lower than normal by 3-5° C.
TELECAN Tutorial
3.13
 SALINITY
• It is defined in the ocean as the grams of salt per 1000 grams of water.
• One gram of salt per 1000 grams of water is defined as 1 psu (practical salinity
unit). In the open sea the salinity range is usually 32-37 psu.
• Salinity varies due to evaporation and precipitation over the ocean, as well as
river runoff and snow melting.
• Along with temperature, salinity is an important factor in contributing to the

changes in the density of seawater and, therefore, in the circulation of the ocean.
• While sea surface temperature has been measured from space for more than
three decades, the technology to measure sea surface salinity from space has
only recently emerged. As oceans have 1,100 times the heat capacity of the
atmosphere, ocean circulation becomes critical for the understanding of heat
transfer through the Earth and therefore for the understanding of climate change.
• Sea Surface Salinity (SSS) can be measured by passive microwave radiometers

working around 1.4 GHz.
• The power received by the radiometer is proportional to the microwave emissivity

and the temperature of the ocean surface. Salt increases reflectivity and
decreases emissivity of water. Thus, if the water temperature can be obtained by
other means, the salinity may be deducted from the received radiation.
• Satellite remote sensing was not able to measure salinity up to the year 2009,
when the platform SMOS (ESA) was launched. SMOS measures SSS by using
24 low noise radiometers in each arm. These arms are 4.5 m long and have an
Y-shaped antenna. Its accuracy is 0.1-0.2 psu and its resolution is about 35 km.
• Aquarius (NASA) was launched in 2011 and includes a radiometer plus a

scatterometer to measure salinity and to consider the effects of sea surface
roughness in the signal reaching the radiometer.
TELECAN Tutorial
3.14
Missions to measure salinity
1st Global map salinity (SMOS)
TELECAN Tutorial
3.15
Salinity Maps (AQUARIUS)
 SURFACE WINDS
• Ocean wind is the motion of the atmosphere relative to the surface of the ocean.
• Typically, winds over the ocean are measured with anemometers close to the
surface and attached to buoys, platforms or ships. Winds can also be remotely
measured by using Doppler radars that can capture the sea wind (the scope is
usually limited to several hundred kilometers due to drop of the signal).
• More recently, advances in remote sensing satellites have allowed near surface
wind measurements using passive and active instruments.
TELECAN Tutorial
3.16
 Remote Sensing Instruments
• Passive Radiometer: Ocean surface rapidly responds to the movement of air

above it, which provides a different roughness pattern depending on the relative
velocity and the wind direction in relation to the ocean surface. The roughness of
the ocean surface provides a specific "brightness" that can be detected by using
passive microwave radiometers and can be accurately translated to the wind
velocity near the surface.
• Active radars: Specific microwave wavelengths are sensitive to Bragg

scattering, which is a characteristic of centimeter-scale ocean surface waves,
known as capillary waves. They are directly influenced by changes in the
near-surface winds, which enable specialized radars, known as scatterometer,
observe these changes. These radars transmit microwave pulses to the ocean
surface, which immediately scatters a portion of the reflected energy back to the
radar. Once the radar cross section is normalized, the near-surface wind speed
can be calculated as a function of the backscattered energy. Unlike passive
microwave radiometers, the active radar system can combine measurements
from different azimuth angles to obtain the approximate direction of the wind.
TELECAN Tutorial
3.17
Example of surface winds
TELECAN Tutorial
3.18
TELECAN Tutorial
3.19
 MARINE CURRENTS
Sea currents are continuous and direct movement of ocean waters, generated by
forces acting on this medium. Ocean currents can flow great distances, and play a
key role in determining the climate of many regions of the Earth.
Currents can be caused by wind pressure, thermohaline gradients or tidal force.

These currents are also influenced by the Earth's rotation through the Coriolis
deflection.
Ocean currents can be divided into surface streams (10%) and thermohaline currents
(90%). Surface currents are primarily driven by the wind, while thermohaline
circulation is due to variations in temperature and salinity (these streams are slow
compared to the movement of surface).
 Types of currents
Large scale circulation
In the Northern hemisphere, currents flow around hills in clockwise direction and
counterclockwise around valleys: The opposite phenomenon occurs in the Southern
hemisphere. These currents form whirls around both sides of the equator. Another
large scale circulatory phenomenon is planetary waves.
Mesoscale circulation
Mesoscale circulation is defined as an energetic phenomenon that at spatial scales

ranges from ten to several hundred kilometers and at temporal scales ranges from a
few days to several months. Mechanisms of genesis are mainly large scale
circulation instabilities, interactions between currents and bathymetry, and pressure
exerted by the wind.
TELECAN Tutorial
3.20
Currents can be estimated by using remote sensing altimeters (SSHA) or by
calculating the optical flow in sequences of sea surface temperature or chlorophyll
concentration images. However, in this case clouds could cause problems and the
vector field can only be estimated in areas that are not completely uniform, i.e.
mesoscalar structure areas.
 Altimeter
Ocean currents can raise the height of the sea surface up to one meter above the
surrounding area. Therefore, currents can be calculated by measuring height
variations with on-board satellite altimeters.
TELECAN Tutorial
3.21
 SST Sequences
It is the motion estimation in image sequences, typically using region matching

techniques.
TELECAN Tutorial
3.22
 Eddies: ocean storms
These 50-300 km wide structures with a circular or elongated shape are similar to
storms, but they occur in the ocean. They can be clearly seen in maps of sea surface
anomaly from altimeters or in temperature images.
Eddies are important for ocean circulation and climate -for the transport of heat and
movement-, and for marine biology and fisheries, as they cause the mixture of
different types of ocean waters that influence conditions that allow the marine food
chain to develop.
Over 10 years of satellite data, Analysis of Sea Level Anomalies, reveals the high
activity areas twists or eddies during those years. Anticyclonic (see above) and
cyclonic (see below) eddies are shown with lifetimes longer than 18 weeks. This
information has been provided from altimetry data.
Both types of eddies move Westward, and with a slight tendency (less than 10°) to
the equator or toward the poles, respectively.
TELECAN Tutorial
3.23
 OCEAN COLOR
It is the spectral radiation measurement obtained from the reflectance in the visible
band.
Main objective: it is to extract concentrations of marine phytoplankton.
Phytoplankton is the set of aquatic autotrophic plankton, which has photosynthetic

capacity and live dispersed in water. It is important to monitor phytoplankton because:
 It fixes carbon dioxide into organic matter.

 It plays a key role in the global carbon cycle and climate.
 It is responsible for about half of the Earth’s net primary production.
 It is the basics of the marine food chain.
 It affects several industries, mainly fishing.
 It helps to monitor coastal waters and their quality.
TELECAN Tutorial
3.24
 Remote Sensing: Top Products
The main products obtained by satellite are:
Normalized water-leaving radiances (nLw)
Radiance emerging from marine subsurface in each spectral band analyzed and
spreads through interface sea-air.
Units: W cm-2 sr-1 nm-1
Chlorophyll-a concentration (Chl)
Phytoplankton primary photosynthetic pigment used as an index of phytoplankton

biomass.
Units: mg m-3
TELECAN Tutorial
3.25
 CHLOROPHYLL CONCENTRATION
 Open Oceans OC algorithms are the most widely used.
 There are variations depending on the number of bands in the range of green
and blue.
 OC3 and OC4 algorithms are the most current and implemented for MODIS
(NASA) and MERIS (ESA) sensors.
General expression of the Bands used

algorithm OC4 = 443> 490> 510/555
log10 (Ca) = (c0 R c1 + c2 + c3 + R2 R3 R4 + OC3 = 443> 490/555
c4) OC2 = 490/555
where R is log10 (max (Rrs 443, 489 Rrs, Rrs Clark = 490/555
510) / 555 Rrs) Carder = 490/555
TELECAN Tutorial
3.26
Monthly Map for February 2013 (MODIS-AQUA)
 Regional Scale (Case 2 waters)
The algorithms shown above cannot generally be used for coastal or inland waters,
where the difficulty is greatly enhanced by:
 Temporal and spatial variability

 Pollution from land or seabed
 Aerosols (dust, pollution)
 Suspended sediments and CDOM
 Atmospheric effects and anthropogenic emissions
TELECAN Tutorial
3.27
 WATER QUALITY
Monitoring water quality by remote sensing is normally obtained by several key

parameters: turbidity, temperature, chlorophyll, CDOM, etc.
The direct relationship between the variation in the reflectivity of the channels
and the variation of water quality parameters was studied by low and medium
spatial resolution sensors (SeaWiFS, Modis MERIS, etc.) for open waters.
Recently, the high-resolution WorldView2 satellite with 8 spectral bands has

been launched and provides measurements of water quality in coastal or inland
water with a resolution of 2 m.
TELECAN Tutorial
3.28
Spectral characteristics
 Suspended Matter
• Turbidity is defined as the lack of transparency of a liquid due to the presence

of suspended particles. The more suspended solids are there in the water, the
dirtier the water look and the higher its turbidity is. Turbidity is considered a
good measure of water quality. The more turbid the water gets, the lower its
quality is. Turbidity affects the degree to which light is scattered and absorbed
by molecules and particles.
• Shallow coastal waters are often characterized by high turbidity caused by

resuspension of bottom sediments or human-induced sediments (works,
dredging, etc.).
• Algorithms that estimate suspended matter in the water (Total Suspended

Mater - TSM) are based on the increased reflectivity of water in the red and
green bands.
TELECAN Tutorial
3.29
There is currently no accurate and reliable generic algorithm for
quantification of TSM in coastal areas
 There are a variety of algorithms that calculate turbidity mainly using several
bands near the red.
 Algorithms directly using a band reflectivity (i.e., polynomial) are more

exposed to noise.
 MODIS provides 2 products of suspended particulate matter which are

inorganic (PIC) and organic (POC) matter.
 Dissolved Matter
 Colored dissolved organic matter (CDOM, yellow substance, gelbstoff) is
the optically visible component of organic matter dissolved in the water. It has
an important effect on the biological activity of aquatic ecosystems.
 CDOM absorbs short wavelengths (UV, blue), while clean water absorbs the
higher wavelengths of the visible spectrum. That is why water turns into a
greenish-yellow tone.
TELECAN Tutorial
3.30
 Algorithms for estimating CDOM are normally based on the variation of the
reflectivity of the yellow band. They are not robust or noisy.
 CDOM interferes with the remote detection of chlorophyll concentrations.

Therefore, algorithms for the detection of CDOM are used for discriminating
the matter against the photosynthetic pigments and inorganic suspended
matter.
 Here we are an example of an algorithm for calculating CDOM:
 Medium- High resolution
One of the main current challenges is to study water quality by using medium and
high resolution satellite data.
TELECAN Tutorial
3.31
 Oil spills
Another aspect of water quality is the detection of oil spills. This is possible using
data from synthetic aperture radars (SAR) that detect the roughness of the sea. Oil
spills are easily detectable under certain wind conditions, because the backscatter is
much lower in areas contaminated by the spill.
 Oil spills: a case study
TELECAN Tutorial
3.32
• ERGOS (Environmental Response Team for Black Tide)
• WWF/Adena, ESA, INTA, Directorate General of Coasts, Merchant Navy,

SASEMAR and the Canary Islands’ Government.
• Period: June 2000 - October 2002
• Area: Canary Islands (300, 000 km2)
• Detectability basically depends on wind speed (3-6 m / s), sea state and how
old is the stain.
 BATHYMETRY
A very recent application is to obtain the bathymetry of shallow coastal areas -up to 20
or 30 meters- from high resolution satellite data. It is a very complex issue, because
only a few bands manage to penetrate far enough and the effects of seabed reflectivity
have to be avoided.
TELECAN Tutorial
3.33
 BENTHIC CLASSIFICATION
The classification of the seabed is a difficult issue to solve from remote sensing data. It
is only possible a shallow depth classification, since light penetration in coastal waters is
much lower than in the open sea and basically only just blue and green channels can
penetrate.
In this example image, worldview-2 shows that only the first 3 or 4 bands (up to yellow
band) may be used.
TELECAN Tutorial
3.34
Furthermore, only homogenous classes that are separable at those wavelengths can be
distinguished. Turbidity prevents seafloor classification.
Examples of Benthic Classification
TELECAN Tutorial
3.35
3.2 TERRESTRIAL REMOTE SENSING APPLICATIONS.
 VEGETATION AND FORESTS
Terrestrial vegetation systems provide unique reflectance properties of the radiation

received allowing characterization using remote sensing techniques.
 Vegetation Indexes
Spectral indexes are combination of spectral bands to obtain the parameter of
interest (vegetation, water, bare soil, etc.)
Vegetation indexes are calculated by combining reflectivity at different wavelengths

to discriminate and extract vegetation information minimizing the influence of external
factors such as soil, solar irradiance, illumination and observation geometry, etc.)
TELECAN Tutorial
3.36
• There are many indixes for estimating vegetation:
Examples of the most important indexes:
NDVI and EVI (MODIS / Aqua).
TELECAN Tutorial
3.37
 Forests
Remote sensing is useful in forestry applications. Some important applications are
the following:
• Forest cover
• Forest density (crown closure)
• Phenology (seasonality)
• Inventory of forest species
• Biomass estimate
• Deforestation
• Forest Protection
• Prevention, monitoring and evaluating fire
Global deforestation between 1990 and 2005 was annually 14.5 million hectares.
The results of the overall evaluation of forests by remote sensing show that in 2005
the world's forest cover was 3,690 million hectares, or 30% of the Earth's land area.
TELECAN Tutorial
3.38
These new results indicate that the rate of deforestation in the world, mainly the
conversion of tropical forests into agricultural land, was an average of 14.5 million
hectares per year between 1990 and 2005. This finding is consistent with previous
estimates.
TELECAN Tutorial
3.39
 AGRICULTURE
Agriculture is the economic mainstay of many countries. Remote sensing can provide large
information:
• Crop extension
• Stock crops
• Agricultural production
• Forecast of harvests
• Selection and monitoring agricultural areas
• Evaluation of drought or flood damage
• Control of pests and crop diseases
• Detection of metabolic stress (water or nutrition)
• Precision agriculture
TELECAN Tutorial
3.40
 Agricultural production
Example of applications:
• Selection of suitable plots (better soil, moisture, etc.)

• Monitoring the evolution of crops (phenology, fertilization, etc.)
• Detection and damage assessment
Example: progressive collection of wheat in June 2006
 URBAN MONITORING
 Urban applications of remote sensing
Remote sensing applications in urban areas depend mainly on the number of

available bands and the spatial resolution of the sensor.
TELECAN Tutorial
3.41
Bands
Detecting
Hyperspectral urban
materials
Extension Coverage Mapping

Multispectral and and uses structures
applications (city) (Spectral)
(global,
Global-
regional) Built area Mapping of
A band regional urban (City) structures
growing (geometric)
Spatial
Resolution
Moderate High Very high
Time
Different examples are presented below.
 Extension and global/regional urban growth

• Mapping urban areas
• Understanding urbanization
• Analyzing and forecasting growth trends and land use change
• Preventing environmental degradation and pollution
• Global Economic Analysis (population, agricultural waste, industrial,

commercial, etc.)
TELECAN Tutorial
3.42
 Extension and urban/rural growth
• Mapping urban city level
• Estimating population
• Understanding urbanization (city blocks, etc.)
• Planning (ports, airports, roads, etc.)
• Detecting clandestine constructions
• Analyzing vulnerabilities and risks
• Monitoring natural disasters
• Analysing regional economics (population, decrease of agricultural

activity, industrial and comercial activities, etc.)
TELECAN Tutorial
3.43
 Land use and environmental indicators
• Mapping land cover at regional, local or city level
• Mapping urban green areas (parks, forests, etc.)
• Characterizing and estimating drainage areas sealed (floods, etc.)
• Studying urban heat islands: the comfort of cities depends on the 3D

structure of buildings, their geometry, building materials, landscaping, etc.
TELECAN Tutorial
3.44
 Vulnerability to disasters
• Vulnerability maps at local level
• Hyperspectral optical sensors (can be combined with Lidar (height), SAR,

etc.)
 Damage assessment after disasters
Peru: Earthquake (7.9 degrees)

Study of change detection using SPOT-5
TELECAN Tutorial
3.45
 Urban remote sensing at very high resolution
• Level analysis at object level (building, road, bridge, etc.)
• Detection and modeling of structures
• 3D modeling of buildings.
• Catastrophes, etc.
 LAND USE AND LAND COVER
• Economic development, population growth and climate variability have caused

rapid changes in recent decades.
• Our life is linked to different land cover -water, forests, deserts, etc.- around us.
When they change, our health, economy and environment may be affected.
• Remote sensing allows obtaining land use and land cover maps (LULC) that are
becoming more and more reliable and are achieving a better spatial resolution.
TELECAN Tutorial
3.46
Class Selection
• There is no single standard for global LULC maps. The most commonly used
are: Corine, LCCS (GlobCover, GLC2000), Anderson, IGBP, etc.
• For regional analysis, it makes sense to choose the right classes for the study
area.
TELECAN Tutorial
3.47
 CORINE LAND COVER
CORINE (Coordination of Information on the Environment) involved the creation of a

database of land use in the European Union. It is run by the European Environment
Agency.
 ANDERSON
In 1972 Anderson developed a system of multilevel classes of land cover.
It is used primarily by the USGS using Landsat data.
TELECAN Tutorial
3.48
 IGBP (International Geosphere-Biosphere Programme)
International program for coordinating the interactions between biological, chemical

and physical land processes and human systems at global and regional level
(MODIS 500m).
TELECAN Tutorial
3.49
 LCCS (Land Cover Classification System)
Powered by FAO and UNEP (United Nations Environment Programme)
 GLC2000
Global land cover map for the year 2000 generated by more than 30 institutions
using SPOT-Vegetation data (1 km)
 GLOBCOVER
Global land cover map of the ESA made in collaboration with EEA, FAO, GOFC-
GOLD, IGBP, JRC and UNEP using MERIS data (300 m)
TELECAN Tutorial
3.50
 SIOSE
The Spain Information System on Soil Occupation aims at integrating information

from databases and land cover of the Autonomous Communities and the Central
Government. SIOSE uses multiple sources of information, including SPOT and
Landsat data.
TELECAN Tutorial
3.51
 MAPPING AND GIS
Remote sensing systems are very useful in mapping applications and geographic
information systems thanks to the increasing in spatial resolution.
Example of Google Earth.
 EARTH SURFACE TEMPERATURE
Land Surface Temperature (LST) is a parameter that serves as an indicator of energy

balance and water on the Earth's surface for the detection of climate change. It is also
useful to monitor the health of the vegetation or to analyze desertification processes or
to be used as an input data model for weather prediction.
TELECAN Tutorial
3.52
 NATURAL DISASTERS
Remote sensing has enabled humanity to understand more clearly the dangers that
threaten our planet.
It is essential for disaster management from the design of models of risk and
vulnerability analysis, to early warning and damage assessment:
• Disaster prevention (lives, material and natural resources)

• Real-time tracking of a disaster
• Analysis of the effects after a natural disaster
• Tracking recovery activities
Types of Natural Disasters:
• Fire
• Floods
• Earthquakes, faults, etc.
• Eruptions
• Drought
TELECAN Tutorial
3.53
 Fire
Global maps of fires from low resolution sensors in the thermal IR bands
 Fire in the Canary Islands (summer 2007)
The main strategies for detecting the burned area are:
A post-fire image: NBR index (Burnt Normalized Ratio) or Tasseled-Caps

(Landsat)
Images before and after the fire: change detection techniques (NDVI, NBR, PC,
classification).
TELECAN Tutorial
3.54
 Floods and Earthquakes
 DEFENSE
Remote sensing has been used for decades in areas such as:
• Security and Intelligence
• High-resolution mapping
• Verification of international treaties
• Border control
• Emergency Management
• Natural Disaster Monitoring
TELECAN Tutorial
3.55
 OTHER APPLICATIONS
• Geology (minerals, sedimentation, erosion, etc.)
• Soil moisture
• Topography (digital elevation models of terrain)
• Archeology
• Geodesy
TELECAN Tutorial
3.56
3.3 ATMOSPHERIC REMOTE SENSING APPLICATIONS
 INTRODUCTION
• The atmosphere is a mixture of gases.
• It has different layers.
• The first 80 kilometers contain more than 99% of the total mass of the Earth's
atmosphere.
• Air circulation is a large-scale movement of air through the troposphere. It is the

mechanism by which heat is distributed around the Earth.
• The large-scale structure of the atmospheric circulation varies from year to year,
but the basic structure remains fairly constant as it is determined by the speed of
the Earth’s rotation and the solar radiation difference between the equator and
the poles.
TELECAN Tutorial
3.57
 Interactions of the atmosphere
• When EM radiation travels through the atmosphere absorbed or scattered by

particles in the atmosphere.
• The atmosphere also emits Thermal IR
The main applications of remote sensing in the atmospheric field are:
• Weather forecast
• Study of atmospheric gases
• Prevention of disasters (storms, winds, etc.)
• Renewable energies (wind, solar)
• Air quality: pollution, aerosols, calimas, etc.
• Greenhouse gases
• Climate change
TELECAN Tutorial
3.58
The main tasks of remote sensing in the atmospheric field are:
• Geostationary satellites (MSG): Predictions in real time (nowcasting)
• Polar satellites (ESA, NASA): Better spatial resolution
Atmospheric products generated by different agencies or remote sensing centers are

presented below.
 EUMETSAT
EUMETSAT’s main missions is currently Meteosat, a Second Generation Satellite,

and, specifically, SEVIRI sensor, although data from other sensors in polar orbit are
also used.
Some of the products supplied by EUMETSAT are presented and described below.
They can be divided into:
- Near real-time meteorological products

http://oiswww.eumetsat.org/IPPS/html/MSG/PRODUCTS/
- Atmospheric products
http://www.eumetsat.int/Home/Main/DataProducts/Atmosphere/index.htm?l=en
 Near real-time meteorological products
 Winds MSG (SEVIRI)
Atmospheric Motion Vectors (AMV) product consist of wind vectors estimated at different
heights by tracking the movement of clouds and other atmospheric components (for
example, patterns of water vapor and ozone).
TELECAN Tutorial
3.59
• Winds are caused by changes in atmospheric pressure (air mass pushed
downwards by gravity) leading to air movements.
• Pressure gradients propel winds with air that moves from higher pressure areas
to lower pressure areas.
• Friction on the surface causes that very low altitude winds move more slowly and
often in directions other than the high levels of the atmosphere.
• Air circulation is also affected by the heating of the sun.
• Techniques for remote sensing of winds are:
- Winds of Clouds: tracking clouds and water vapor in the VIS or IR bands (MSG).
• Tracking the movement of clouds in the geostationary satellite imagery is the

oldest method to calculate wind vectors.
TELECAN Tutorial
3.60
• The direction and speed is estimated by measuring the difference vector in the
location of a particular cloud in two consecutive images, divided by the interval of
time between two images (typically 15 or 30 min).
• Wind can be accurately estimated by using clouds that move passively with the
winds: high-level cirrus and low level clouds -high cumulonimbus and orographic
clouds are poor candidates.
• Problem: dependence on the presence of clouds in the picture (grid

non-uniform).
• Solution:, images of water vapor (6.5 to 7.5 microns) allow to monitor the wind by
the movement of water vapor in cloud-free areas.
Procedure for obtaining vectors AMV
- Sea Surface Winds: analysis of the roughness of the surface of the oceans, with
active or passive microwave sensors.
 Disaster Prevention: wind alert
AMV wind product correlated with the height of the waves on March 4, 2013.
TELECAN Tutorial
3.61
 Precipitation
Multi-sensor Precipitation Estimate (MPE) product consists of rain rates in real time in
mm/hr for each Meteosat image in original pixel resolution. The algorithm is based on
the combination of microwave measurements from polar satellites and images on
Meteosat’s IR channel.
Applications: operational weather forecasting in areas with little or no radar

coverage, particularly in Africa and Asia.
TELECAN Tutorial
3.62
 Clouds MSG (SEVIRI)
Cloud Analysis (CLA) product provides an identification of cloud layers specifying the
type of cloud coverage, height and temperature.
Applications: weather forecast, numerical weather prediction, climate research and

monitoring.
The Cloud Analysis Image (CLAI) product identifies types of clouds. This is a picture
of the product obtained along with CLA.
Applications: Weather forecast, numerical prediction, climate research and

monitoring.
TELECAN Tutorial
3.63
Cloud Mask (CLM) product is a mask that indicates the presence or absence of cloud
in each pixel. Specifically, each pixel is classified as clear sky over the water, on land
clear sky, cloud or not processed (outside the disk of the Earth).
Applications: The main use is in support of the applications of short-term

forecasting and for remote sensing of continental and oceanic surfaces.
The Cloud Top Height (CTH) product indicates the height of the highest clouds. It is
obtained from the information extracted from the images and data of the cloud’s
analysis. It also makes use of other external weather data.
Applications: aeronautical meteorology.
TELECAN Tutorial
3.64
 Moisture MSG (SEVIRI)
The Tropospheric Humidity (TH) product

provides the relative humidity in both the
middle and the high troposphere.
The upper level is obtained from the

middle layer relative humidity between
approximately 600 hPa to 300 hPa using
WV6.2 channel micrometers, while the
average moisture in the troposphere
represents the average value between
850 hPa and 600 hPa using WV7.3
channel microns.
TELECAN Tutorial
3.65
 Atmospheric products
The available atmospheric data from satellites has benefits beyond its contribution
to weather forecasting. The data accumulated in the EUMETSAT files help to detect
and to understand the processes that affect climate change. Information on the
composition of the atmosphere, obtained from satellite measurements, plays a vital
role in the knowledge of the environment, and the environmental risk assessment,
such as the destruction of the ozone layer in the stratosphere and the accumulation
of pollutants in the atmosphere.
Some EUMETSAT meteorological products are:
- Coastal Winds ASCAT at 12.5 km Swath Grid - Metop

- Aerosol Properties over Sea - MSG - 0 degree
- All Sky Radiances - MSG - 0 degree
- Atmospheric Motion Vectors - MSG - 0 degree
- Cloud Analysis - MSG - 0 degree
- Cloud Analysis Image - MSG - 0 degree
- Cloud Mask - MSG - 0 degree
- Cloud Top Height - MSG - 0 degree
- High Resolution Precipitation Index - MSG - 0 degree
- Total Ozone - MSG - 0 degree
- Tropospheric Humidity - MSG - 0 degree
- Volcanic Ash Detection (CAP) - MSG - 0 degree
http://www.eumetsat.int/Home/Main/DataProducts/Atmosphere/index.htm?l=en
 Aerosols MSG (SEVIRI)
The product AES estimates the aerosols’ optical thickness in pixels of clear sky over
the sea in VIS0.6 channels, VIS0.8 and NIR1.6. Furthermore, it determines the
TELECAN Tutorial
3.66
coefficient of Angstrom. The product is a daily average. Applications include
numerical weather prediction, research and climate monitoring.
• Aerosols are particles suspended in the atmosphere from natural and

anthropogenic sources with sizes between 2 nm and 1 mm.
• Sources of aerosols: soil dust, sea salt, volcanic emissions, pollen, biomass
burning or industrial combustion.
• Geographically one of the main sources that generate this atmospheric dust or
haze is the Sahara desert and the Sahel region.
• Dust affects primarily climate -temperature and precipitation-, marine productivity,

and health.
• Dust also affects directly and indirectly the Earth's radiation balance. On the one
hand, it disperses outward incident solar radiation causing a cooling of the
planet's surface, and on the other, it prevents that the radiation emitted by the
Earth's surface escapes which causes warming.
• It includes nutrients such as iron, phosphates or organic detritus, which have a

positive effect on the ocean by fertilizing regions of low marine productivity.
TELECAN Tutorial
3.67
• Furthermore, dust affects air quality and thus it has a detrimental effect on
human health due to the transport of spores, fungi, bacteria or pesticides.
 Ozone MSG (SEVIRI)
The overall density of the ozone in the atmospheric column for each image segment
is based on 9.7 micron channel (channel SEVIRI ozone) and other IR and WV
channels.
This product is used by NWP centers, ozone monitoring services and research
institutes.
Other space agencies generate atmospheric remote sensing products.
TELECAN Tutorial
3.68
 EUROPEAN SPACE AGENCY (ESA)
ESA has also launched several satellites to study the atmosphere, but since the
creation of EUMETSAT, the objective is geared more to scientific studies than to
operational services.
Some of the atmospheric parameters are described below.
 Ozone
Hole in the ozone layer: area in the stratosphere where abnormal ozone reductions
occur. It is an annual phenomenon observed during spring in the Polar Regions,
followed by a recovery during the summer.
Lately, significant reductions have been found in this layer, especially in Antarctica. It
seems to have been caused by a increasing in the concentration of chemicals,
standing up chlorofluorocarbons (CFCs) used as refrigerants and aerosol
propellants.
TELECAN Tutorial
3.69
 Carbon Dioxide
It is the most important greenhouse gas that causes global warming. Despite the
importance of CO2, our current knowledge (mainly natural) of its sources and sinks is
still insufficient.
TELECAN Tutorial
3.70
SCIAMACHY ENVISAT satellite is the first instrument which can measure the global
distribution of CO2 with high sensitivity over land where the sources and sinks of CO 2
can be found. So far, it provides a global confirmation to based measures in some
places.
 Methane
It is the second most important greenhouse gas that affects global warming.
Despite the importance of this gas, our current knowledge of its sources is still
unsatisfactory. Recently large quantities of methane in rainforests have been found
thanks to SCIAMACHY. This founding points out to the possibility of the existence of
methane sources not yet known or to a significant underestimation of the known
sources.
Important sources of methane are rice paddies, ruminants (cattle and sheep),
wetlands and methane emissions from plants, which is possibly a major new source
of methane discovered by SCIAMACHY that could explain the high level of methane
TELECAN Tutorial
3.71
 Nitrogen dioxide
NO2 is one of the main pollutants. It is brownish-yellow. It is formed in combustion

processes at high temperature (motor vehicles, power plants), being a toxic and
irritating gas prevalent in urban areas. It mainly affects the respiratory system.
TELECAN Tutorial
3.72
 Ultraviolet Index
The UV index is an indicator of the intensity of the Sun’s ultraviolet radiation on the
Earth's surface. UV index also indicates the ability of the solar UV radiation to injure
our skin.
TELECAN Tutorial
3.73
 Water Vapor
Water vapor is a gas obtained by evaporation, boiling liquid water or ice sublimation.
It is odorless and colorless. It is responsible for humidity and at high concentration it
condenses and forms fog or, at higher concentrations, it produces clouds.
 Sensor MIPAS (ENVISAT): allows 3D studies
TELECAN Tutorial
3.74
 REMOTE SENSING SYSTEMS (RSS - NASA)
Remote Sensing Systems is a research center supported primarily by NASA that

focuses on the generation of products from microwave sensors.
 GERMAN AEROSPACE AGENCY (DLR)
World Data Center for Remote Sensing of the Atmosphere

(WDC-RSAT)
Since 2003 the German Remote Sensing Center DLR operates the WDC-RSAT. By
using satellite data, this center offers scientists and the general public data free from
many parameters and atmospheric missions.
TELECAN Tutorial
3.75
TELECAN Tutorial
3.76
4. Radiometric, Atmospheric and
Geometric Modeling
Introduction
The technical complexity of the current remote sensing systems, the volume of the
available data and the different levels of processing involved in obtaining geophysical
parameters, require the establishment of a hierarchy of processes that allow to generate
useful operational products (high level) for specific end users and for the scientific
community. Typically, only a small ratio of the total available data is processed at the
highest level (user level), which generally increases the cost of data at the processing
level.
'Each processing level within the hierarchy requires more auxiliary data and it is
more complex than the previous levels.'
The type and number of hierarchical processing levels obviously depends on the remote
sensing system considered. A proper structuring of hierarchies in the main remote
sensing systems (Landsat, SPOT, NOAA, NASA EOS, ERS-ESA) has enabled the
creation of databases of consistent and reliable images.
The rapidly changing capabilities of computers and high speed computer networks,
allow images to be acquired, processed at different levels and distributed to the
scientific community in near real time, according to their requirements and operational
applications.
As an example, in the next figure we show the flow chart of the hierarchical processing
levels applied to NOAA-AVHRR/SeaStar-SeaWiFS data that are generally used in the
main remote sensing systems.
TELECAN Tutorial
Radiometric, Atmospheric and geometric Modeling
4.2
Fig.1. - Generic hierarchy of processing levels applied to NOAA-AVHRR data.
The first sections of this chapter analyze the sensor characteristics affecting the
radiometric quality of the images and the atmospheric modeling applied to the data
observed by the sensor. This is important to answer the question "What are we
observing?" Another relevant question is "Where are we observing?" The answer to this
question is determined by the images’ geometric features.
TELECAN Tutorial
4.3
The images taken by sensors located on-board of satellites contain geometric
distortions, in addition to the radiometric errors related to the instrument characteristics
and the presence of the atmosphere. Geometric distortions can be caused by many
factors. The relative motion between the satellite, the exploration sensor and the Earth
originate displacement errors of the pixels in the image obtained. The exploration
features of the sensor, the Earth’s curvature and variations, uncontrolled position and
orientation of the geometric platform generate geometric errors with a different origin
and complexity.
Next, the main sources of radiometric, atmospheric and geometric distortion and their
effects on the image will be analyzed, as well as the basic techniques for cloud
detection. Also the procedures used for their compensation will be reviewed and applied
to the correction of images from different sensors.
 Distinguish among the main sources of error produced in Earth observation
satellite images;
 Know the radiometric distortion factors caused by space platforms;
 Understand the atmospheric influence on the spectral radiance detected by the
remote sensing sensor;
 Study the main sources of geometric distortion and the procedures for restoration
of satellites images;
 Know cloud detection techniques.
CONTENTS
Radiometric, Atmospheric and Geometric Modeling
4.1. SOURCES OF ERRORS IN SPACE OBSERVATION
4.2. RADIOMETRIC MODELING
4.3. ATMOSPHERIC MODELING
4.4. GEOMETRIC MODELING
4.5. CLOUD DETECTION
TELECAN Tutorial
4.4
4.1 SOURCES OF ERRORS IN SPACE OBSERVATION
 APPROACHES TO THE USE OF REMOTE SENSING DATA
 Centered Image: Spatial relationship between different features on the Earth

surface  Maps Creation: Photogrammetry
 Centered Data: High precision in the spectra-temporal calibrations.
Integration and comparison of multi-temporal and multi-sensor data
TELECAN Tutorial
4.5
 SOURCES OF ERROR IN A SPATIAL IMAGE
What are we observing?

Radiometric and atmospheric properties.
Where are we observing?

Geometric characteristics of the images.
The images taken by sensors located on satellites contain:
- Radiometric errors: instrument (sensor) and atmosphere presence.
- Geometric errors: Relative motion between the satellite, the exploration

sensor and the Earth, exploration of the sensor, Earth curvature, platform
variation.
TELECAN Tutorial
4.6
 SATELLITE IMAGE PROCESSING HIERARCHY
What are we observing?

Radiometric Modeling
Atmospheric Modeling
Where are we observing?

Geometric Modeling
TELECAN Tutorial
4.7
 SATELLITE IMAGE PROCESSING HIERARCHY: PREPROCESSED
IMAGES TASKS.
 Radiometric Calibration: Convert digital levels to radiance values or

brightness temperature values.
 Atmospheric correction: Take into account the contribution of atmospheric

radiation reaching the sensor (NDVI or SST recovery).
 Geometric correction: Correct distortions in the images received related to

curvature and rotation of the Earth, sensor exploration and variations of the
platform.
 Detection of clouds: Mask correctly cloudy pixels to ensure that the

geophysical parameters obtained are representative of the Earth surface.
 SATELLITE IMAGE PROCESSING HIERARCHY: OBTAINING SST
TELECAN Tutorial
4.8
 SATELLITE IMAGE PROCESSING HIERARCHY: OBTAIN SST
TELECAN Tutorial
4.9
4.2 RADIOMETRIC MODELING
 Convert DN to radiance values: it is necessary to obtain geophysical parameters
or to compare images from different sensors.
 If the sensor were completely stable, pre-launch calibration would be enough.
 Variations in thermal conditions and degradation of the instrument -response

and sensitivity of detectors in the sensor over time- require incorporation of a
dynamic calibration system in flight.
Example: WV2 Radiometric Modeling
TELECAN Tutorial
4.10
Example: AVHRR Radiometric Modeling: Linear Model
TELECAN Tutorial
4.11
4.3 ATMOSPHERIC MODELING
 Disadvantage Remote Sensing: sensor receives data from the Earth's surface
through the atmosphere.
 Absorption and scattering are atmospheric effects, but it is nearly transparent to

electromagnetic radiation at many wavelengths.
Atmospheric windows
Atmospheric transmittance: Opacity of the Atmosphere
TELECAN Tutorial
4.12
 ATMOSPHERIC CORRECTION: EFFECTS ON THE CALCULATION OF
THE REFLECTIVITY
 Radiance received by the satellite :
 Transmissivity in the upward direction
 Actual reflectivity of land cover
 In order to calculate the surface reflectivity (to estimate the atmosphere

trasnmisivity, the diffuse irradiation and radiance due to atmospheric
dispersion)  ATMOSPHERIC CORRECTION
TELECAN Tutorial
4.13
 ATMOSPHERIC CORRECTION: METHODOLOGIES
 Laborious and complex procedures that require multiple inputs on the conditions
of the atmosphere when the sensor capture the image.
 Atmospheric model based on in-situ data and data from other sensors
(multilook method).
 Physical model based on the Radiative Transfer Equation.
Usual: MODTRAN4, 6S, FLAASH (SW ENVI)

and ATCOR (SW ERDAS)
 Atmospheric model based on data from the image itself.
Usual: DOS and COST
TELECAN Tutorial
4.14
 ATMOSPHERIC CORRECTION: MULTI-BAND METHOD TO RECOVER
THE TSM.
 A simple linear combination of radiances at two wavelengths provides a good

estimate of the radiation emitted by the surface.
 Coefficients:
o Theoretical, using a radiative transfer model.

o Regression between in situ temperature data and brightness temperature.
The great variability of the angle of observation -satellite zenith angle-

causes a change in the optical path of the atmosphere:
TELECAN Tutorial
4.15
 ATMOSPHERIC CORRECTION: VEGETATION INDIXES
 Based on the relationship between NIR and R bands.
 Normalized Difference Vegetation Index (NDVI) is widely used. Its values run
between -1 to 1.
 NDVI is a poor indicator of arid or semi-arid regions.
TELECAN Tutorial
4.16
 ATMOSPHERIC CORRECTION: SECOND SIMULATION OF A
SATELLITE SIGNAL IN THE SOLAR SPECTRUM (6S) TO RECOVER
REFLECTIVITY IN WORLDVIEW 2 (AND MODIS) IMAGES
The model is divided into five main parts:
Geometric conditions.
Atmospheric Modeling.
Definition of the heights of the study area and satellite.
Spectral conditions.
Defining the type of soil.
Fixed configuration:
 Atmospheric model.
 Heights of terrain and satellite. Satellite bands.
 Defining of the type of surface.
Settings for each image:

 Geometrical model -angles of the Sun and the satellite.
 Optical depth of the atmosphere -NASA database.
Radiometer Measurements
TELECAN Tutorial
4.17
WorldView-2 Signatures
Atmospherically corrected WV2 Granadilla Imagery
TELECAN Tutorial
4.18
4.4 GEOMETRIC MODELING
 Digital images from space sensors contain geometric errors.
 Sources of geometric distortion:
Earth Rotation.
Panoramic distortion.
Curvature of exploration.
Earth curvature.
Platform variations.
It is essential to have the exact location of any pixel, in order to compare images -
multitemporal or multisensor analysis- or to validate satellite data with in-situ
measurements GEOMETRIC CORRECTION
TELECAN Tutorial
4.19
 GEOMETRIC CORRECTION: EARTH ROTATION.
Scanning sensors require a finite time to acquire a scene from the Earth
surface: The last lines are erroneously displaced East in terms of what they
represent on Earth -rotates from West to East.
 Ejemplo: Imágenes procedentes de los satélites Landsat, tomadas en Sidney.
* Velocidad angular del satélite es w0=1.014 mrad/seg y la longitud de la imagen L=185 Km.,
El tiempo de exploración de los 185 Km es:
ts=L/(Re· w0) = 28.6 seg (Re=6378 Km)
* Velocidad de la superficie de la tierra:
Ve=Re· we ·cos (lat) (we =72.72 rad/seg)
En Sydney, lat=33.8º, tal que: Ve= 385.4 m/seg.
* Durante el tiempo de adquisición la tierra se mueve al E:
Xe = Ve · ts = 11.02 Km en 33.8º de Latitud
(La imagen tendrá un 6% de distorsión al E).
TELECAN Tutorial
4.20
 GEOMETRIC CORRECTION: DISTORTION OVERVIEW.
It is constant as the instantaneous field of view (IFOV) from sensors on
satellites: Effective size of the pixel on the Earth (GIFOV) is greater at the
extremities of the field of view at nadir.
Consequences:
 Compression of the scene, which increases towards the edges of the

scanned area.
 Effective spacing of the pixels on Earth increases with the exploration
angle -error in the position of the pixels in the transverse direction.
TELECAN Tutorial
4.21
 GEOMETRIC CORRECTION: EARTH AND EXPLORATION
CURVATURE.
i. Satellites with high field of view (FOV): the effect of the curvature of the
Earth is important for high scan angles.
TELECAN Tutorial
4.22
ii. The sensor’s rotating mirror requires a finite time to explore a full
line: During this time the satellite continues to move Curvature of the
scan line on Earth.
TELECAN Tutorial
4.23
 GEOMETRIC CORRECTION: PLATFORM VARIATIONS.
An invariable orientation of the platform is essential in relation with the
geometrical accuracy: A small change in the pointing angle causes a large
variation in the points observed on Earth.
Balanceo
Cabeceo
Guiñada
Balanceo
Cabeceo
Guiñada
For a constant IFOV, variations in the height and/or speed of the platform
results in scale changes in longitudinal and transverse directions to the sensor
scanning.
TELECAN Tutorial
4.24
 GEOMETRIC CORRECTION: EXAMPLE OF NOAA-AVHRR
GEOMETRIC DISTORTIONS.
TELECAN Tutorial
4.25
 GEOMETRIC CORRECTION: METHODOLOGIES
 Orbital models: They are orbital parameters used to model the nature and
extent of distortion sources on the basis of the geometry of the satellite orbit
and the angle of view of the sensor.
 Ground control points (GCP) use mapping functions -general polynomial-

obtained between GCP in both images.
GCP are locations on the Earth's surface that can be identified in the
input image and whose position is known in the image or map
(reference).
 Orbital Model + GCP use a small number of control points to relocate the
pixels that have been previously corrected by a geometric orbital model.
 GEOMETRIC CORRECTION: ORBITAL MODEL
 Modeling the nature and magnitude of systematic geometric distortions:
 Geometry of the satellite orbit.

 Scanning geometry of the sensor.
 Geometry of the Earth.
 Good method for low-resolution sensors used in marine applications (hard to

get GCP).
 Commonly used to correct systematic errors: Products that many users get
from EOSAT, USGS or ESA.
TELECAN Tutorial
4.26
S
B
hS
D' z S'
RT
D'
j
O
ecuador
i
D'
A
nodo ascendente
E
traza del nadir
órbita del satélite
TELECAN Tutorial
4.27
Problems
Lack of precision of the magnitudes involved in the satellite's position in space

(accuracy of the orbital elements, altitude or orientation angles, displacements on
the internal clock of the satellite).
TELECAN Tutorial
4.28
 GEOMETRIC CORRECTION: GROUND CONTROL POINTS (GCPS)
 It models directly corrections in the domain of the image, without explicit

identification of the sources of distortion.
 Knowing the correspondence between a number of points (GCPs), the

transformation function (typically related) to map the image 'slave' to the
image or reference map (correspondence between the two images) can be
determined.
TELECAN Tutorial
4.29
Problems
Identify a high and spatially well distributed set of GCPs.

Time consuming operator (area-based or feature-based methods).
In many cases it is impossible due to occlusion by clouds.
Uncertainty in the exact location of a region.
TELECAN Tutorial
4.30
Example of geometric correction based on control points: Low
resolution multisensorial images (MODIS sensor):
(a) Linear polynomial model.

(b) Quadratic polynomial model.
TELECAN Tutorial
4.31
 GEOMETRIC CORRECTION: ORBITAL MODEL + REDUCED SET
(GCPS).
TELECAN Tutorial
4.32
4.5 CLOUD DETECTION
 Objective: Mask properly all the cloudy pixels.
 Clouds are of interest for:

o Studies on climate.
o Forecast.
 Clouds are masking the signal of interest:

o Ocean color, SST, etc.
o Properties of the Earth's surface, etc.
TELECAN Tutorial
4.33
 CLOUD DETECTION: ALGORITHMS
 Multiband threshold methods

 Exploiting the spectral characteristics.
 Brightness, temperature.
 Problems: thresholds variables (area, day/night sensor bands,
railway, etc.)
(Saunders and Kriebel 1988, CLAVR 1991)
 Extraction of features and classification

 Spectral characteristics (clustering).
 Spatial characteristics (textures, DWT).
 Artificial neural networks.
 Problems: calculation times.
 Multitemporal analysis
TELECAN Tutorial
4.34
Saunders & Kriebel Multiumbral Algorithm
AATSR 'Cloud Screening' Algorithm
TELECAN Tutorial
4.35
MODIS 'Cloud Masking' algorithm
TELECAN Tutorial
4.36
 SATELLITE IMAGE PROCESSING HIERARCHY
Example 1: AVHRR
Complete Procedure: Pre-processing of Earth Observation Satellite Images.

 Radiometric Calibration
 Atmospheric Correction
 Geometric Correction
 Cloud Detection
TELECAN Tutorial
4.37
Example 2: WorldView 2
Complete Procedure: Pre-processing of Earth Observation Satellite Images.
 Radiometric Calibration
 Atmospheric Correction
 Solar Reflection Correction
TELECAN Tutorial
4.38
5. Remote Sensing Image Processing
Introduction
Digital image processing is the set of techniques applied to digital images in order to
enhance visual quality or facilitate the search or extraction of information.
The interest of digital image processing methods lies in two main areas of application: i)
the reconstruction or improvement of information that provides an image to be
interpreted by a human being, and ii) information processing of a scene to allow
automatic analysis by a machine.
These techniques have experienced a significant growth, being used currently, for a
variety of problems in various fields such as medicine, geography, archeology, physics,
astronomy, biology and, of course, remote sensing.
Throughout this chapter, most image processing techniques will be described. Thus, the
contrast enhancement or color representation of the different spectral bands will be
highlighted to facilitate visual interpretation. Other processing techniques described for
image improvement are: generating spectral indexes applied mainly to the detection of
vegetation, convolution filtering techniques, and techniques in the Fourier domain or
fusion pixel level. These techniques are very useful to improve the spatial detail of
multispectral high-resolution satellite images. Regarding the analysis of images,
techniques oriented to automatic detection of structures in images will be described, as
is the case of the Hough transform, mathematical morphology and other segmentation
algorithms. Also the classification process to generate thematic maps will be discussed
in detail and, finally, various techniques of motion estimation in image sequences will be
presented.
In addition to the explanation of image processing techniques, representative examples

are also included to assess the effects they produce in the image and thus facilitate the
understanding and use of them.
TELECAN Tutorial
Remote Sensing Image Processing
5.2
 Know the characteristics of a digital image and its main parameters.
 Describe various techniques for improving image quality for furthers analysis.
 Present the main techniques for image analysis, detailing the detection structures
techniques, the thematic classification and the motion estimation.
CONTENTS
5.1. INTRODUCTION TO IMAGE PROCESSING
5.2. THE DIGITAL IMAGE
5.3. IMAGE ENHANCEMENTS
5.4. IMAGE ANALYSIS
TELECAN Tutorial
5.3
5.1. INTRODUCTION TO IMAGE PROCESSING
Digital image processing includes the set of techniques for processing and analysing
images by using computers.
Image Processing System
5.2. THE DIGITAL IMAGE
 A/D CONVERSION: DISCRETE IMAGES
Sampling: Sampling is responsible for integrating in points the information which is

in a given area. These points are the smallest elements that divide an image, called
pixels.
TELECAN Tutorial
5.4
Quantification: Once sampled the image, the value of each pixel must be digitally
encoded. This process of assigning a number of levels or bits to each pixel is called
"quantization" of the image.
TELECAN Tutorial
5.5
The digital image is formed by one or more matrices (e.g., sensor bands) of numbers
(DN: digital levels). That is why they are generally multidimensional functions.
 Types of digital images
Values represented by a digital image
– Intensity: scalar (one band)
– Color: vector R, G, B (3 bands)
– Material properties: (1 or more bands)
• X-ray images: absorption
• Ultrasound images: density
• Infrared images: temperature
• Remote sensing images: reflectivity
TELECAN Tutorial
5.6
5.3. IMAGE ENHANCEMENTS
 CONTRAST ENHANCEMENT
The sensor must be able to detect a high dynamic range of values, but one scene in
particular has a very low contrast to encompass only limited radiance values.
The aim is to improve the visual quality of the image. This will make various
transformations (min-max, saturation, equalization, etc.) to the image histogram.
This involves applying a transformation that maps the original digital levels (ND) in gray
levels (NG):
TELECAN Tutorial
5.7
There are several possibilities of processing levels.
These changes are made for each pixel in the image as a separate element and
independent of its position in the image using the histogram of the image.
TELECAN Tutorial
5.8
• The histogram is the statistical distribution of the pixels of an image in terms of
number of pixels for each (DN) possible value.
• It does not contain information on the spatial distribution of the pixels in the
image.
As an example, next we present some possible transformations.
• Linear expansion
Min-Max Stretch expands the dynamic range of the values to fill the full range of
reproduction. GN: greyscale and DN: digital levels.
TELECAN Tutorial
5.9
Use the minimum and maximum signal value  Sensitive to extreme values
(outliers). Solution: Saturation stretch.
Same value for all the range  Sensitive to lower symmetric histograms. Solution:
Non-linear stretch or histogram equalization.
• Normalization expansion
- Robust algorithm (adapted from saturation stretch).

- It is a linear expansion of a signal with a given mean and variance to cut the
ends of the dynamic range.
- It lets us to control the average value of the output image and to vary the
contrast by adjusting the variance.
The mean is constant and the contrast varies by changing the variance.
• Thresholding
- It is a transformation that classifies the image into two categories (binarized)

based on a threshold on the original values of the image.
- It requires a greater number of thresholds for a greater number of classes.
TELECAN Tutorial
5.10
 COLOR REPRESENTATION
The human eye is more sensitive to colors than gray levels, so it is important the color
representation of images for visual analysis.
This representation can be done in two ways:
- False color representation (Pseudo): when a color table is applied to a grayscale

band (LUT: Look-up table) or when red, green or blue colors are assigned to
several bands that do not correspond to the true color.
- True color representation: when the image is taken by a sensor that captures
these 3 bands of color: red, green and blue.
TELECAN Tutorial
5.11
 Pseudocolor one band
Other examples of LUT
 Pseudocolor multiband
Channels with values in the non-visible spectrum are shifted to the visible spectrum.
TELECAN Tutorial
5.12
 True Color
 SPECTRAL INDEXES
They are combinations of bands for obtaining a parameter of interest for
classification purposes (vegetation, water, minerals, etc.)
They are based on the behavior of the reflectivity parameter to maximize its
discrimination.
TELECAN Tutorial
5.13
 Vegetation indexes
They are mainly based on relations between NIR and R bands.
Normalized Difference Vegetation Index (NDVI) is widely used. It generates values

between -1 and +1. NDVI is an inaccurate indicator in arid regions.
TELECAN Tutorial
5.14
 CONVOLUTION FILTERING
Technique to modify the spatial information of the image using the digital values of
the neighborhood (local filtering).
• A window that moves along rows and columns is used.
• An operation with the input image’s pixels that fall within the window is carried
out and the result is the new pixel of the image in the center position of the
window.
Example with original image and the lowpass and high pass filtered.
TELECAN Tutorial
5.15
 Mask convolution (impulse response)
Averaging filters: reduce noise by means of spatial averaging
TELECAN Tutorial
5.16
Gradient filters: contour detector (approximation of the derivative)
 Edge Detector
TELECAN Tutorial
5.17
 FOURIER TRANSFORM
The Fourier transform applied on images provides information about the structures
present in the image and can be used to filter out unwanted frequencies.
The Fourier transform of an image is composed of two types of information: the

amplitude and the phase.
TELECAN Tutorial
5.18
Examples
TELECAN Tutorial
5.19
The power spectrum (square of spectrum modulus) is used in the global spatial pattern
recognition in the image.
 IMAGE FUSION
It aims at improving the spatial quality of the multispectral image (MS) using the detail
that provides the high-resolution panchromatic band (PAN), while preserving the
spectral information.
Objective:
• Get images with high spatial and spectral resolution, from the PAN image (high
spatial resolution, low spectral resolution) and multispectral bands (high spectral
resolution, low spatial resolution).
TELECAN Tutorial
5.20
• Improve the visual quality to facilitate quality photo interpretation and GIS.
• Improve the detection and extraction of objects and features.
• Improve the thematic classification.
• Improve the detection of changes in the multi-temporal images.
 General framework PAN-MS Fusion
TELECAN Tutorial
5.21
 Method based on Discrete Wavelet Transform:
 Wavelet Transform
It provides a decomposition of the image into its components at different scales by

using filters. In this way, each image is decomposed into its approximation and detail
images.
TELECAN Tutorial
5.22
For its representation, the four outputs are joined into a single image.
It can be done at different levels returning to decompose the image approach
TELECAN Tutorial
5.23
Example of a Quickbird image using the Mallat wavelet.
TELECAN Tutorial
5.24
5.4. IMAGE ANALYSIS
 FEATURE DETECTION
 Segmentation
Segmentation is a first step in most image processing problems. It can be defined as

finding a partition of the space (Image I) in n disjoint regions to get to satisfy a
homogeneity criterion P (.):
The homogeneity criterion can be as simple (or complicated) as desired. Normally,

criteria to approach to the semantics of the scene are defined.
This example of coastal upwelling shows the difficulty of segmentation, because

objects have no sharp borders and each user may only be interested in a given area
and, therefore, each segmentation would be different. As an example, some
segmentations obtained by different algorithms are shown in the following image.
TELECAN Tutorial
5.25
There are two basic segmentation algorithms families:
- Discontinuities detector Detect abrupt changes in the signal.
- Homogeneities Detectors Detect areas with similar characteristics.
The use of additional information such as movement or depth improves the results.
Discontinuities detectors
- Edge detectors filters are commonly used to detect discontinuities from gradient
or Laplacian operators. Then edges are joint together to clearly delimit the
objects.
- It is a complex and not always reliable process that may be affected by noise to a
great extent.
TELECAN Tutorial
5.26
Homogeneities detectors
Techniques based on the homogeneity of the pixels. The most common are:
- Thresholding: Detection using techniques based on manually or automatically

thresholding. A global threshold for the image or local thresholds may be used.
Otsu,Jawahar,Li, Rosenfeld,
Hertz, Huang, Trian.,Riddler
Original Pun, GM-EM Beghdadi Shanbag Yasuda
Abutaleb, Pikaz, Kapur, Sahoo,
Yanowitz Yen
Sezan, Olivio,
Lloyd, Kamel, Brink, O’Gorman,
Bernsen Niblack Tsai, Yanni Ramesh, Kittler Sezgin
Palumbo, Pal
Sauvola, White
- Region growing
It analyzes the Region Adjacency Graph (RAG). It merges similar regions and
updates RAG with the new values. This process is iterated until a certain final
criterion.
- Split & merge

Split step: It divides the image into disjoint regions by analyzing the similarity
criterion. This step is carried out by following a quadtree structure (QuadTree).
Merge step: It merges tree leaves considering the similarity criterion. This step
does not follow any predetermined structure.
TELECAN Tutorial
5.27
Watershed
The watershed algorithm is a segmentation technique based on morphological tools

that combines the concepts of region growing and edge detection:
It groups up pixels around the minimum of the image.

The boundaries of the regions are located at points (peaks) of maximum
gradient of the image.
The process can be understood as the result of flooding a topographic relief:
It is interpreted as a relief image.

The water is introduced by the minimum of the relief.
When two are waterfronts meet together, a dam (contour) is built up.
TELECAN Tutorial
5.28
Advantage: The boundaries of the regions are correctly located.
Disadvantage: It produces an over-segmentation of the image.
Example: segmentation of vegetation with various levels of detail
TELECAN Tutorial
5.29
Hough transform
This transform can be used to locate objects whose shape is known. The simplest
example is the application of the Hough transform for locating lines or circular shaped
figures.
Straight
TELECAN Tutorial
5.30
Examples
TELECAN Tutorial
5.31
Circles
Mathematical Morphology
Powerful signal processing tools for filtering the scene specific elements preserving
others.
Mathematical morphology was originally developed for binary images and further
extended to functions and grayscale images. These nonlinear tools are based on
management criteria and are not easily applicable to vector images.
Example of morphological filters on binary images:

- Erosion
TELECAN Tutorial
5.32
- Dilation
- Opening
- Closure
TELECAN Tutorial
5.33
 THEMATIC CLASSIFICATION
It is the step for image analysis that aims at the generation of thematic maps from the
input information for the end user.
In other words, it tries to transform numerical data into descriptive categories of images
to identify the various elements of the image
The categories or classes selected for the thematic map should be able to be
discriminated from the numerical data of the image.
There are various types of classifiers based on the information used:
• Spectral classifier
It is based on the fact that the different classes of the image have different
combinations of digital values in each band due to its reflectance or emittance.
TELECAN Tutorial
5.34
• Space classifier
It is based on analyzing the relationship between neighboring pixels, considering

aspects such as texture, proximity, size, shape, repetition, etc.
• Temporal classifier
It uses images on different dates to favor the classification process (e.g., crops or
certain types of vegetation have different spectral properties depending on the
season which facilitates identification).
• Object-oriented classifier
It is a spatial and spectral hybrid classifier which typically includes pipeline

stages prior to guide the classification so that each object is assigned entirely to
a possible class.
Next, the spectral classifier and, more specifically, each typical classification step are
described in greater detail.
 Feature Extraction step (optional)
It is based on applying a spatial or spectral transformation to get the most interesting

features for classification.
In this stage:
• The information of the image or original bands can be extracted.
• Unwanted variability in the spectral signatures (spectral indexes. Example: NDVI

in the figure) can be deleted.
TELECAN Tutorial
5.35
• Spectral separation of classes can be improved.
• The number of bands (Principal Component Analysis) can be reduced.
 Training step
Select pixels representative of the desired classes to train the classifier. It can be
done in a supervised or unsupervised way.
Separability
Before jumping to the training phase, it is recommended to analyze the separation of

classes to assess a priori whether the classification is possible or if there will be
some classes that will not be discriminated.
It is insufficient to only use the distance between the means. It also requires the
standard deviation or variance.
Separability measures between classes:
TELECAN Tutorial
5.36
There are several measures of separability from Euclidean and angular
distances. Some do not take into account the variance, that is to say, the spectral
overlap between the classes, and they become less reliable (in the table there
are examples of metric separability).
The Jeffries-Matusita distance is widely used. It is bounded to 2 for large class

separations.
TELECAN Tutorial
5.37
Example of Separability
The classification algorithm must be trained. There are several possibilities:
TELECAN Tutorial
5.38
Supervised Training
Training samples are labeled by the user (e.g., figure above). It involves prior
knowledge of the terrain (in-situ data, maps, photo interpretation).
• All classes in the image must be taken into account to avoid errors (Solution:
apply thresholds in the allocation phase).
• The analyst must select representative areas for each class to get the digital level
range for each category.
• Training areas can be established through field visits, maps, photo interpretation,
etc.
• Each training area (ROI) must include the range of variability of the class (Using
more than one training area for each class).
• Number of pixels in training> 10*N (N: number of bands used). It is

recommended 100 * N.
• There is no guarantee that classes will be distinguishable (one thematic ↔ 1

spectral)
Unsupervised Training
Training samples are not labeled. Classes are automatically determined.
• It does not require prior knowledge of the area.
• The algorithm locates concentrations or groups (clusters) of pixels with similar

characteristics (assuming that classes have similar spectral values).
TELECAN Tutorial
5.39
• The groups represent classes in the image, but then they have to be labeled
according to our classes of interest.
• There is no guarantee that the obtained spectral classes correspond to classes

of interest.
• The number of groups in the image is not known use more than classes of
interest.
• Generally the full image is classified (+ training assignment), instead of using the
image areas for unsupervised training.
• Classical algorithms: K-means and Isodata.
Graphic illustration of operation of K-means algorithm on 2

bands
The iterative process is repeated until the change in the average values reaches a
threshold.
TELECAN Tutorial
5.40
Mixer training
It is performed combining both training types. There are several possibilities:
• First, unsupervised training (find representing spectral classes in sub-images) to

guide the supervised training to select the separable thematic classes.
• First, supervised training for labeling the known classes, to guide unsupervised
initialization of clusters.
• Applying the two methods separately and combine the results.
 Assignment or Labelling step
This stage is to assign each pixel in the image to one of the existing classes. A
thematic image is obtained.
Problem: Establish statistical limits for each class.
TELECAN Tutorial
5.41
Techniques:
Nonparametric:
- Parallelepiped classifier
- Minimum distance
Parametric
- Maximum Likelihood Classifier
Parallelepiped classifier
All image pixels with values within the centered parallelepiped mean value of a
training class are assigned to that spectral class. It is the fastest of all.
There are different methods for determining the boundaries of parallelepipeds.
Problem: Overlapping limits and not all pixels are classified
Minimum distance classifier
The pixels of the scene are classified using the average distances to the training
data. A pixel is assigned to the nearest class.
TELECAN Tutorial
5.42
Decision surfaces are linear and they are determined from the average values (no
variances) of the training data (modeling classes symmetrical like in the spectral
domain It is sometimes a problem!)
It is slower than the parallelepiped classifier.
There is an error when a class has not been considered All pixels are classified,
even if the spectral value is far from the average. Solution: use a threshold so that
the decision surfaces are circles centered in each middle class.
Maximum Likelihood Classifier
It uses the statistics of the training sets (mean and covariance) and the pixels are
assigned to the class with the highest probability.
Each class is considered to have a normal distribution.
As the minimum distance classifier, all pixels of the scene are mapped to any of the
classes (except thresholds are applied as shown in the figure).
It is slower than the previous classifiers.
Theoretically offers the best classification.
TELECAN Tutorial
5.43
• Procedure
- The probability of a pixel is calculated for each class.
- The pixel is assigned to the class with the highest probability.
Spectral angular distance classifier
It uses the spectral angle between classes to assign the pixels.
It is independent of the magnitude of the spectral vectors. It is therefore more robust

to be insensitive to variations in topography, lighting, etc.
A maximum threshold allowed angle can be specified and pixels at greater angle are
not rated.
TELECAN Tutorial
5.44
Decision tree
It is one of the simplest but also more efficient methods.
Setting thresholds is quite laborious.
It consists on discriminating sequentially each category based on the spectral values,

or the texture or the auxiliary information.
Rules are established for each class. These rules allow to discriminate each class
from the variables or bands that best discriminate this class from the rest.
Neural Network
Neural networks can predict from a sample of observed inputs and outputs. The
learning objective is to estimate from known results about input data (training
sample), to calculate later results from the remaining unknown input data.
TELECAN Tutorial
5.45
It consists of a set of simple elements (neurons) arranged in layers. The units in each
layer are connected to the next layer through activing functions (weights are
calculated iteratively during training).
Remote sensing is often used in back-propagation networks consisting of three

layers (input, hidden and output).
Advantages: it can include bands or auxiliary data of all types to improve the
robustness and accuracy.
Problems: lack of general criteria for designing the structure of the network. The
tedious training and the classification depends largely on the amount and quality of
the training results. For this reason, SVM (Support Vector Machines) are more
commonly used.
TELECAN Tutorial
5.46
Example of results for a supervised classifier
Minimum distance classifier
Example of results for an unsupervised classifier
TELECAN Tutorial
5.47
 Classification accuracy
The accuracy of the classification must be verified. The main sources are:
- Soil structure (shape, direction and size of objects, spatial distribution of

classes, degree of mixing, sloping terrain, etc.)
- Use of images with inadequate spatial or spectral resolution.
- Atmospheric influences
- Unsuitable acquisition dates
Common methods are:
Visual
Confusion / fail matrix

- In-situ data -two independent sets, one for the training phase and others for
evaluation- are commonly used.
- Check the percentage of pixels of each image class classified correctly and
incorrectly.
Kappa Coefficient
TELECAN Tutorial
5.48
Confusion Matrix
• Method to quantify the accuracy of a classification.
• Square matrix (rows: actual classes, columns: classification classes)
• Diagonal indicates the number of pixels classified correctly.
• The off-diagonal values correspond to misclassifications.
- Overall accuracy: correct/total pixels

- Error of omission (Pixels of a class that is not detected as such): residue in
rows.
- Error of commission (Pixels incorrectly classified): residue in columns.
• It allows to know the main conflicts between classes redefine classes
Kappa Coefficient
• It measures the correspondence between the classified image and the reality,
and the correspondence that would be obtained by simply making a random
classification.
• It seeks to measure the degree of adjustment due to classification accuracy by

eliminating random factors.
0: agreement due to coincidence
1: total agreement image and reality
Negative: misclassification
• It allows comparing different methods (different confusion matrices).
TELECAN Tutorial
5.49
TRAINING PLAN
CONTENTS
ACQUISITION AND ANALYSIS OF REMOTE SENSING IMAGES
1.1. INTRODUCTION
1.2. MEDIUM AND HIGH RESOLUTION IMAGES
IMAGE CATALOG
COST OF THE IMAGES
1.3. LOW RESOLUTION IMAGES
OCEANCOLOR
LPDAAC
PO.DAAC
LAADS
EARTHEXPLORER
EOLI-SA
GIOVANNI
TELECAN Tutorial
Acquisition and Analysis of Remote Sensing Images
T1.2
1.1. INTRODUCTION
The objective of this practice is to provide the necessary information for obtaining
spatial remote sensing images, for high resolution and low resolution sensors.
We present here the main interfaces to access and download data from different
sensors on board of satellites. There are a variety of image servers but we have
focused on what we consider most significant for the type of data they provide or for
having a more intuitive handling. Most servers present an interface from a web browser,
but they also often offer the possibility of ftp access for downloading data. In some
cases, they allow a direct download, and in others you have to make a request and wait
for it to be processed.
Furthermore, there are other applications which, apart from allowing to download the
data, they provide utilities to perform different types of analyzes.
1.2. MEDIUM AND HIGH RESOLUTION IMAGES

Normally, you have to pay for medium and high spatial resolution images and they are
usually quite expensive (increasing in price with the resolution of the sensor). Below
there is a list with the access to the most used image catalogs and their prices.
 IMAGE CATALOG
The catalogs to access the remote sensing images of the major medium and high
resolution satellites are:
TELECAN Tutorial
T1.3
These catalogs allow you to select a geographic area and view images in the file.
TELECAN Tutorial
T1.4
 COST OF THE IMAGES
In general, the cost of high resolution images is high, and it depends on the selected
processing level. Here you are some links to satellite imagery providers where you can
get prices for most high resolution satellites.
Note that the products of the Landsat satellite series are offered free of charge for
non-commercial applications.
TELECAN Tutorial
T1.5
1.3. LOW RESOLUTION IMAGES
Low spatial resolution products are generally free of charge. There are many servers
that allow their access and to download data. Among the most important servers, the
following can be found:
Giovanni (http://disc.sci.gsfc.nasa.gov/giovanni)
OceanColor (http://oceancolor.gsfc.nasa.gov/)
LPDAAC (https://lpdaac.usgs.gov/)
PODAAC (http://podaac.jpl.nasa.gov/)
Laads (http://ladsweb.nascom.nasa.gov)
USGS (http://earthexplorer.usgs.gov/)
USGS (http://glovis.usgs.gov/)
ESA (http://earth.esa.int/EOLi/EOLi.html)
GMES (http://gmesdata.esa.int/web/gsc/home)
EUMETSAT (http://www.eumetsat.int/home/main/dataaccess/index.htm)
In Spain at national or regional level the most outstanding servers are:
INTA-NPOC (http://www.crepad.rcanaria.es/es/npoc/distribucion.html)
INTA-CREPAD (http://www.crepad.rcanaria.es/es/index.html)
PNT (http://www.ign.es/PNT/)
ACIISI-PET (Http://www.teledeteccioncanarias.es)
Some of the image servers shown above also allow certain analyzes in a way that, by
selecting a specific area and a range of years, different mean maps, anomalies or
graphical representations of data sets can be obtained.
TELECAN Tutorial
T1.6
Some of the previous access server products are described below. We specifically
analyze:
Access and download of marine, atmospheric or terrestrial data generated from

space sensor:
OceanColor (http://oceancolor.gsfc.nasa.gov/)
LPDAAC (https://lpdaac.usgs.gov/)
PODAAC (http://podaac.jpl.nasa.gov/)
Laads (http://ladsweb.nascom.nasa.gov)
Furthermore, most of these data can be accessed centrally making a request

with the following application:
Reverberation (http://reverb.echo.nasa.gov/reverb/)
Access to satellite images, orthophotos and LIDAR:
USGS (http://earthexplorer.usgs.gov/)
Most servers presented here belong to NASA. That is why they are the most accessible
and documented. They also feature a variety of products that meet the expectations
required.
Also, we will show the desktop application for accessing and downloading data stored
and processed by the European Space Agency (ESA).
EOLISA (http://earth.esa.int/EOLi/EOLi.html)
Finally, Giovanni a NASA’s tool for making various types of analysis will be explained.
Giovanni (http://disc.sci.gsfc.nasa.gov/giovanni)
TELECAN Tutorial
T1.7
 OCEANCOLOR
In this web you can view, download and access information on oceanographic
parameter products derived from different sensors’ data. Here we are some screen
shots of this web:
As we can see in the drop down menu, as it is shown in the figure on the right, data
from CZCS sensors, OCTZ, SeaWiFS, MODIS, MERIS, Aquarius and VIIRS are
processed and L1/L2 and L3 level products are generated. Geophysical parameters that
can be downloaded are:
Remote Sensing Reflectance (Rrs)

Chlorophyll Concentration (chlor_a)
Diffuse Attenuation Coefficient at 490nm (Kd_490)
Particulate Organic Carbon Concentration (poc)
Particulate Inorganic Carbon Concentration (pic)
Colored Dissolved Organic Matter Index (cdom_index)
Daily Mean Photosynthetically Available Radiation (pair)
TELECAN Tutorial
T1.8
Instantaneous Photosynthetically Available Radiation (ipar)
Normalized Fluorescence Line Height (nflh)
The interface to access to level 1 and level 2 is shown in the following figure:
The access link is: http://oceancolor.gsfc.nasa.gov/cgi/browse.pl?sen=am
Here you are the Interface to L3 data:
The access link is: http://oceancolor.gsfc.nasa.gov/cgi/l3?per=DAY
TELECAN Tutorial
T1.9
It also allows downloading data via http (http://oceandata.sci.gsfc.nasa.gov/)
 LPDAAC (Land Processes Distributed Active Archive Center)
LPDAAC belongs to the Data Information System (EOSDIS) from NASA's Earth
Observing System (EOS). It is responsible for processing, storing and distributing data
and land information products from ASTER and MODIS sensors onboard the Terra
satellite, and MODIS onboard Aqua satellite. The following figure shows the web and,
on the right side, the products available.
TELECAN Tutorial
T1.10
The most characteristic land parameters that can be downloaded from the web are:
Vegetation Indixes
Thermal Anomalies & Fire
Surface Reflectance Bands
Land Surface Temperature
Land Cover Type
In the following link, you can obtain further details of these products.
https://lpdaac.usgs.gov/products/modis_products_table
Data can be downloaded for free from various media:
Data Pool is a data file that provides direct access to the products via ftp. The next
figure shows two screen shots.
TELECAN Tutorial
T1.11
And the access link is: https://lpdaac.usgs.gov/get_data/data_pool
Reverberation is a tool that provides a web interface to centrally access to a complete

Earth observation product database. It presents an attractive interface which, once
completed the search criteria, a product is requested, which once processed can
access, previous an email notice, to an ftp server for downloading it. The figure below
shows how it looks. Furthermore, on the right image you can see information about the
order placed.
TELECAN Tutorial
T1.12
 PO.DAAC (PHYSICAL OCEANOGRAPHY DISTRIBUTED ACTIVE ARCHIVE
CENTER)
It belongs to the Data and Information System, EOSDIS Observing System from
NASA's Earth Observing System (EOS). It is responsible for processing, storing and
distributing ocean and climatic data from many satellites such as ADEOS, Aqua,
AQUARIUS, Coriolis, Cryosat-2, DMSP, ENVISAT, ERS-1, GEOS-3, GFO, GOES,
GRACE, GTS, ICOADS, JASON, METOP, MSG, MTSAT1R, NIMBUS, NOAA,
Oceansat-2, QUIKSCAT, Terra, Topex / Poseidon and TRMM. Here we are an example
of this web.
The main parameters that are responsible for processing in this center are:
Ocean Surface Topography (OST)
Sea Surface Temperature (SST)
Ocean Winds
Sea Surface Salinity (SSS)
Gravity
Ocean Circulation
Sea ice
TELECAN Tutorial
T1.13
In this case, the data can also be downloaded from different media. The options are set
once the parameter of interest is chosen:
It also allows downloading data via direct access to ftp, as shown in the following figure.
TELECAN Tutorial
T1.14
 LAADS (LEVEL1 AND ATMOSPHERE ARCHIVE AND DISTRIBUTION SYSTEM)
It belongs to the Data and Information System, EOSDIS from NASA's Earth Observing
System (EOS). It is responsible for processing, storing and distributing atmospheric and
terrestrial products from MODIS and VIIRS sensor.
A form indicating the searching criteria has to be filled out previously to download the
data.
TELECAN Tutorial
T1.15
Then the order is placed and the product ordered, as it is shown in the following figure.
TELECAN Tutorial
T1.16
Finally, in order to download the product, we access with the order identification
number, once they have sent it to us via e-mail confirming that is processed.
In any case, we can avoid all the previous process by accessing directly via ftp
as it is shown in the next figure.
TELECAN Tutorial
T1.17
 EARTHEXPLORER
This tool provides an online search of different Earth observational data and to
download data from U.S. Geological Survey (USGS). Most of them are free products,
although sometimes you have to pay, if the product requires a priority processing. A
dollar symbol ($) identifies these products. After selecting the area (used area), select
Use Data Set Prefilter. The data that can be found are:
ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer)

AVHRR (Advanced Very High Resolution Radiometer)
Calibration / Validation Test Sites
Commercial Data Purchases (CDP) Imagery
Declassified Satellite Imagery - January
Global Land Survey (GLS)
Heat Capacity Mapping Mission (HCMM) Digital Source
EO-1 Advanced Land Imager (ALI) and Hyperion
Landsat 8 OLI (Operational Land Imager) and TIRs (Thermal Infrared Sensor) (will start
publishing in May)
Landsat ETM + (Enhanced Thematic Mapper Plus)
Landsat MSS (Multispectral Scanner)
Landsat TM (Thematic Mapper)
Landsat TM (Thematic Mapper) Film Only
MODIS (Moderate Resolution Imaging Spectroradiometer) MRLC2001 (Multi-Resolution
Land Characteristics 2001)
NLDC (NASA Landsat Data Collection)
OrbView-3
It can be accessed through the following link: USGS, http://earthexplorer.usgs.gov/
The interface features are shown in the following figure:
TELECAN Tutorial
T1.18
Finally, only registered users can download data. In the following link you can find a
tutorial on using the interface:http://earthexplorer.usgs.gov/documents/helptutorial.pdf
 EOLI-SA
EOLI-SA (Earth Observation Link) is a desktop tool developed by the European Space
Agency to access Earth Observation data catalog and to make the request for an order.
Let us see this tool in the next figure:
TELECAN Tutorial
T1.19
The data accessible via EOLI-SA are:
ENVISAT (ASAR, MERIS, AATSR, MIPAS, SCIAMACHY, Radar Altimeter / MicroWave

Radiometer)
ERS (Scenes SAR, SAR Wave Mode: FDC, Wind Scatterometer: FDC, Microwave Sounder,
Altimeter,
Gome, Orbit Data)
PROBA (PROBA Chris)
LANDSAT (TM: RAW, SCAS; MSS: RAW, SCAS; RBV)
Terra / Aqua (L1B)
ESRB (SAR: PR1, GEC; VNIR: Level1, 2)
NOAA (AVHRR: SHARP 1B, 2A, 2B)
IRS (MOS: L1B)
SeaStar(SEAWIFS: L1A, L1B, L2A, L2B, L2C)
Nimbus(CZCS: L1, L2)
This tool can be downloaded at: http://earth.esa.int/EOLi/EOLi.html.
TELECAN Tutorial
T1.20
You can find more detailed information at:
http://earth.esa.int/EOLIResources/Manual/html/ChapCatalogueWorkSpace.html
The following link shows the free data offered:
https://earth.esa.int/pi/esa?type=file&table=aotarget&cmd=image&id=520
 GIOVANNI
It is a web application developed by NASA to visualize, analyze and access remote

sensing data from different databases. Giovanni is an acronym for GES-DISC (Goddard
Earth Sciences Data and Information Services Center Interactive Online Visualization
and Analysis Infrastructure). In the center of the main screen, data are classified by web
sites. Each of these web pages has a variety of parameters from different sensors.
TELECAN Tutorial
T1.21
There are many web sites within each of the following categories:
Atmospheric Web sites

Applications and Education Web site
Meteorological Web site
Ocean Web sites
Hydrology Web sites
In the following link, you can see a complete list of available geophysical parameters:
http://disc.sci.gsfc.nasa.gov/giovanni/additional/users-manual/G3_manual_Chapter_2_parameters.shtml
On the other hand, Giovanni has different types of representations:
Animation
Anomaly
Plot Area Lat-Lon, Time-averaged
Lat-Lon Plot, Difference Map
Latitude-Time Hovmøller Plot
Longitude-Time Hovmøller Plot
Correlation Plot
Comparison Plot
Plot Cross-Map (Latitude-Pressure)
Cross-Map Plot (Longitude-Pressure)
Cross-Map Plot (Time-Pressure)
Scatter Plot
Scatter Plot, Time-averaged
Curtain Plot
Time Series, Area-averaged
Time Series Difference
Time Series, Area Statistics
Vertical Profile
Zonal Mean
TELECAN Tutorial
T1.22
Histogram
Histogram, Area-Averaged
Histogram, Time-Averaged
The most common representations are:
Anomaly: it is calculated for a selected time period. It can be represented as a map of

latitude/longitude or as a time series. It measures the difference of each value from the
mean value in normal conditions.
Lat-Lon Area Plot Time-averaged: It obtains the mean value for each cell that belongs
to the area selected by the user and it is represented in a 2D color map.
Lat-Lon Plot Difference Map: It performs a representation of the difference of two

parameters or two different databases in a parameter. It is represented as a time series,
where an axis shows the value in the other geophysical time.
Latitude-Time Hovmøller Plot and Longitude-Time Hovmøller Plot: Hovmøller maps

-latitude versus longitude or longitude versus time- represent a longitudinal variation of
a parameter versus time or a latitudinal variation versus time. They allow 2D
representations of the changes occurred in one particular region during a given interval
of time.
Scatter Plot or Scatter Plot Time-averaged: It is an XY representation of two different

parameters. Temporal range is selected, and for each cell a mean value of X and Y are
represented.
Time Series Area-averaged: It is a unique representation of the mean values of a

parameter in a total area defined along a period of time.
Difference Time-Series: It similar to the previous one, but it represents the difference
value between two parameters.
TELECAN Tutorial
T1.23
Once the web site is selected, the general procedure to follow to obtain the
representation of a chosen parameter is:
1. Select the area of interest, indicating the latitude and longitude or select it
graphically.
2. Start date/end date of the period under study.
3. Choose the parameter of interest.
4. Set the representation preferences.
5. Select the analysis you want to perform.
6. Download the data in the desired format.
Practical examples
Example 1. Representation of a Chlorophyll Concentration map
Ocean web sites -> Water Quality Monthly Data
Example 2. Representation of a time series of chlorophyll concentration between

2001 and 2010
Ocean web sites -> Water Quality Monthly Data
Example 3. Representation of a SST day and night difference map for February
2001-February 2010
Ocean web sites -> Ocean color radiometry Online Visualization and Analysis
Example 4. Hovmøller night SST4 latitude and longitude graphic for 2001-February
2012
Ocean web sites -> Ocean color radiometry Online Visualization and Analysis
For different cases you can select the next area of study:
TELECAN Tutorial
T1.24
Practical Example 1
Chlorophyll concentration map, Lat-Lon map, Time-averaged.
TELECAN Tutorial
T1.25
Practical Example 2
Average value of chlorophyll concentration time series (2001-2010)
Practical Example 3
Lat-lon map of time-averaged diferencias
TELECAN Tutorial
T1.26
Practical Example 4
Hovmøller Graphics latitude and longitude
TELECAN Tutorial
T1.27
CONTENTS
REMOTE SENSING IMAGE PROCESSING TOOLS.

2.1. INTRODUCTION
2.2. MONTEVERDI
MONTEVERDI FUNDAMENTALS
2.3. SEADAS
SEADAS FUNDAMENTALS
2.4. ENVI
ENVI FUNDAMENTALS
TELECAN Tutorial
Remote Sensing Image Processing Tools
T2.2
2.1. INTRODUCTION
The information collected by remote sensing sensors is processed and analyzed using
software programs known as remote sensing processing tools.
There are a number of open source programs and many other payment programs to
process this kind of information. Next, we present the most widely used tools for remote
sensing image processing. In addition, an information summary of every tool’s main
features is included.
 COMMERCIAL TOOLS
ENVI: Program focused on the processing and analysis of images of all kinds. It is quite
flexible reading formats and is very friendly for users. Made with IDL language, allows
incorporating additional modules.
ERDAS: It is probably the largest commercial deployment program, being a standard

for interchange formats, especially with GIS. It is available for a wide variety of
platforms. It can be programmed using the Model Maker.
ESRI: ArcGIS is a suite of GIS products. It is grouped up into several applications for
capturing, editing, analysis, processing, designing, printing and publishing geographic
information. ArcGIS Desktop family of desktop GIS applications is one of the most
widely used, including ArcReader, ArcMap, ArcCatalog, ArcToolbox, ArcScene and
ArcGlobe, and various extensions.
Idrisi: It is a tool that incorporates many digital image analysis utilities. Due to its low
price, it has been used extensively in remote sensing teaching. It has been developed
by Clark University (USA).
TELECAN Tutorial
T2.3
E-Cognition: Program that incorporates multiple aspects of context analysis and
mathematical morphology. It is ideal for spatial pattern recognition.
ER-Mapper: Australian program that collects most of the common features in remote
sensing, being quite efficient in managing bulky images by incorporating innovative
systems and virtual algorithms compression.
PCI-Geomatica: New version of the popular PCI Canadian program, in a scheme that
incorporates more packages integrating its digital image analysis, orthorectification,
map production and GIS. It provides quite powerful analysis tools, thanks to its close
connection to the Remote Sensing Canadian Center. You can use a limited demo
version, called Freeview.
DRAGON: Low Cost Program in Windows environment for digital image analysis. It is
very appropriate for an educational setting.
Miramon: GIS and remote sensing program developed by the CREF of Barcelona
University. It is a low cost program with wide functionalities.
 FREE TOOLS
Grass: Program focused on GIS and image processing. It has been developed by
UNIX, on different platforms, but recently Linux and Windows (WinGRASS) versions
have been created. It was initially developed as a military tool for the U.S. Army. NASA,
NOAA, USGS, etc. are the organizations that use it among others.
Orfeo, Monteverdi: OrfeoToolbox (OTB) is a library of remote sensing image

processing developed by CNES in 2006 with the aim of facilitating the use of images
from different sensors. In particular, Orpheus constellation (Pleiades and
Cosmo-SkyMed). Monteverdi is the most robust and operational version.
TELECAN Tutorial
T2.4
SeaDAS: NASA software tool for visualization, processing and analysis of images of
ocean color sensors. It supports the main missions of NASA and, recently, ESA
missions.
Multispec: Program developed by the University of Purdue, a pioneer in the 70s. It is

very appropriate for the learning environment, since it only has a few hardware
requirements. There is also a new version for 32-bit (32-bit version).
SPRING: Program developed by INPE (Brazil National Institute for Space Research). It
has an extensive list of features, but it is primarily oriented towards the GIS
environment. There are versions in Portuguese, English and Spanish.
Next, an introduction to various remote sensing image processing tools is presented. As

free software applications, Monteverdi [http://www.orfeo-toolbox.org/otb/monteverdi.html]
and SeaDAS [http://seadas.gsfc.nasa.gov/] have been selected. We also describe one of
the most common commercial software in the field of remote sensing, ENVI
[http://www.exelisvis.com/ProductsServices/ENVI/ENVI.aspx]. We will start by describing the
main features of Monteverdi, a non-experts’ software, to finish with the advanced users’
ENVI application. We will also include SeaDAS, software historically linked to the
processing of oceanographic images from NASA, and in its new version 7 it provides a
lot of utilities in a new and intuitive graphical interface available for Windows.
2.2. MONTEVERDI
Monteverdi is a computer application sponsored among others by the French Space

Agency (CNES). It is a graphical program based on ITK and OTB libraries (Orfeo
Toolbox) that provides tools for remote sensing image processing. OTB is distributed as
an open source allowing full access to the algorithms used in the library. In addition to
basic utilities, it allows the processing of high-resolution images such as SPOT,
Quikbird, WorldView, Landsat, Ikonos, etc., and hyperspectral and radar images. This
TELECAN Tutorial
T2.5
software is available for free for Linux, Unix and Windows. The latest stable version is
1.14, available for downloading at the following link:
http://sourceforge.net/projects/orfeo-toolbox/files/Monteverdi/Monteverdi-1.14/Monteverdi-1.14.0-win32.exe/download
 MONTEVERDI BASICS
Monteverdi is a user-friendly graphical program for users with little knowledge in remote
sensing. It allows using the principal image formats, as well as the use of some kinds of
vector data. The figure shows an image loaded in the program. At the top of the
interface menu under "File" "Open Dataset" menu, you can access a selection of
image/vector to be loaded. After selecting the data, Monteverdi recognizes its type, and
by pressing "open", the data is available in the main program interface.
TELECAN Tutorial
T2.6
You can see the image by clicking "Visualization" "Viewer" in the top menu. Then a
"Set inputs" window pops up. There you have to select the image or image channel to
be displayed by pressing (+), adding it as "Input". After selecting the image -in our
example a.tif image-, you press the Ok button. At that point you will see a display
composed of navigation, zoom and high resolution. Also on the bottom left, a window
with the histogram of the channels represented and the pixel information will appear.
TELECAN Tutorial
T2.7
Next to display, there is another window with a menu that allows the management of
data. Under "data" you can select the different data entered on the "set input", and the
way they are represented (slides). "Setup" shows the configuration of the data in RGB
color images or grayscale. It also allows you to configure the display in compact format
or in separate windows, and to set the display method -linear, Gaussian, Square root. In
"Histogram" you can see the histogram of the bands represented in the display. In the
case of the straight-line method, the minimum and maximum parameters in the same
histogram can be changed by shifting the vertical bars. Finally, "Pixel description" lets
TELECAN Tutorial
T2.8
you access information according to a pixel position. These options are shown in the
following figures.
TELECAN Tutorial
T2.9
Continuing with the menu "Visualization", access to the "Spectral Viewer" where the
"Set inputs" window will appear again and select the image previously uploaded in the
program. Once the image is loaded, a viewer with navigation windows turns up and also
bars to select the display channels, and, on the left, a list that allows to enter selected
points with the values of the channels and the spectral angle. Besides, we can see
another screen that represents a graph of the channel values of the selected pixel with
the cursor next to the values of the stored points.
TELECAN Tutorial
T2.10
Right click to place items on the list. The figure shows the graphs window spectral
channels of different selected points in the viewfinder.
The following figure shows the spectral angle option which calculates the angle between
the bands of one of the bands selected and the rest of the image. This provides an
information relationship between the point and the image. In this example as we have
selected the point of seawater into the port and there are no waves, the representation
of this area is almost black, as it represents an angle close to zero, which indicates its
relationship to that point.
TELECAN Tutorial
T2.11
To finish with the "Visualization" menu, we access the "color mapping" option where the
"Set inputs" window will appear again. The utility has as complete coloring using
grayscale palette images, so that only one channel is selected. A typical example is to
color a vegetation index such as NDVI. This index can be generated by a menu option
"Filtering" "Feature extraction"  "Radiometric Indices extraction". Once the channel
is loaded with the index, a "Color Map to Apply window" will appear. There you may
select the desired palette and the minimum and maximum values to be represented.
The figure shows the configuration of the palette Winter [-0.5 0.5], RGB image
processed to obtain the NDVI, NDVI grayscale, and the result of the colored image.
TELECAN Tutorial
T2.12
Another basic utility is to save the images to a disk. For example, we have created a
colored image of NDVI that is stored in memory and we want to save it. To do this, go to
"File" menu "Save dataset", and the window "Set inputs" will appear. There you select
the image you want to save (Winter Color Map image). After pressing OK, a "Save
dataset" window will appear where you will enter the route and type of image. We also
introduce the data type that stores each pixel. In our case is an 8-bit RGB image that
can be viewed by any viewer. Next, select the type "unsigned char", keep the "save
metadata" and save it as tif image to store geographic information. The following figure
shows the configuration of the window "save dataset".
TELECAN Tutorial
T2.13
If we need more configuration options when saving an image in a disk, the "save
dataset (advanced)" option can be used. In this option, you may configure, among other
features, the channels that you want to save and the ones that you do not want to store.
The figure shows the "Writer Application" interface, which performs the advanced store
of data.
To finish with the basic options of this tool, we will use the menu option "File" "Extract
ROI from dataset". This option is very useful, if you want to cut a region of interest of an
image. Let us get a square area of the image either in pixel positions or by using
longitudes and latitudes. Next figure shows the "Select the ROI" interface that pops up
after selecting the image to be trimmed.
TELECAN Tutorial
T2.14
2.3. SEADAS
The SeaWiFS Data Analysis System, SeaDAS is an intuitive tool for oceanographic
image processing and data based on "Ocean Color". The latest version 7 is the result of
collaboration with the developers of the ESA and its package BEAM. Thanks to this
collaboration the display is based entirely on BEAM framework which has introduced
many more features compared to the latest version. Furthermore, this new version is
available not only for Linux / Unix systems, but also available for the Windows platform.
In contrast, BEAM integration has caused that the existing IDL module in previous
versions is not yet available.
The software is available for downloading at the following at:
http://seadas.gsfc.nasa.gov/installers/
TELECAN Tutorial
T2.15
 SEADAS FUNDAMENTALS
SeaDAS is a graphical program with a friendly user interface specialized in the

processing of oceanographic data. It allows the use of most remote sensing data and
vector data. The figure shows the load of an image in the program. At the top menu,
under "File" "Import Raster Data", we can see a multitude of choices of sensors and
image formats. After selecting the type of image to be imported, a selection window
pops up where you enter the file path.
TELECAN Tutorial
T2.16
After loading the image, it will appear in the main interface in "Products View" the
window. The figure shows the structure data of the image -format (HDF)- which includes
metadata information, the flags, the grid information and physical information bands of
L2 MERIS sensor’s marine products. By double clicking on one of the bands or products
-in this example it is the suspended matter in the water "total_susp"- a viewer is opened
with the selected data.
TELECAN Tutorial
T2.17
At the bottom left, we see two tabs, the navigation tab and another one called "Color
Manipulation" which allows coloring the images in an easy way.
Another simple way to load images in SeaDAS is by dragging an image to "Products

View". In our example, we may drag the Maspalomas WV” image reprojected. To view it
in RGB, right click and we will see the "Open RGB image View", where a window for
selecting the RGB bands turns up. After selecting the channel and press OK, a new
viewer with the RGB image pops up.
TELECAN Tutorial
T2.18
A very useful option to access NASA oceanographic data is to access via OPeNDAP.
This allows you to download products directly from the database. Here you are a link to
a database of NASA's Modis L3.
http://opendap.jpl.nasa.gov/opendap/allData/modis/L3/aqua/catalog.xml
By selecting the "File" menu  "OPeNDAP Access", we see the following interface. For
downloading meteorological products, select the file and click download.
TELECAN Tutorial
T2.19
Another interesting use of this tool is to export an image or oceanographic product in a
standard format, such as geotif. To perform this task, select the "File" menu  "Export
Raster Data"  "GeoTiff", which brings up a window where you must enter the path of
the file to be generated. However, you should previously press "Subset..." which will
bring up another window configuration. In "Spatial Subset", select the area you want to
save. In "Band Subset", select bands or products -in this example total_susp. In
"Tie-Point Grid Subset", additional product information may be selected. Finally in the
"Metadata Subset", the Metadata to be stored in the geotiff can be selected.
Once everything is configured, press the OK button, and the image with the areas and
the selected products are generated.
TELECAN Tutorial
T2.20
TELECAN Tutorial
T2.21
2.4. ENVI
The "ENvironment for Visualizing Images" (ENVI) is an advance software for processing
and analyzing space remote sensing images. It is one of the most widely used in the
professional world for remote sensing image processing. It allows processing multiple
types of satellite images and data, performing atmospheric corrections, image fusion,
transformations, filters, geometric processing, classification, etc. The latest version of
the program is 5, which improves its graphical interface, being more friendly than the
previous versions. However, in the next practical examples, we will show version 4,
which although it has a less intuitive interface, it retains the full potential of the tool.
TELECAN Tutorial
T2.22
 ENVI FUNDAMENTALS
The basic use of ENVI program does not differ much from the previous program. The
first menu option "File" provides all kinds of tools to open, save, and export images.
Although you can use the generic option screen opening "Open Image File", it is better
to use the "Open External File" because it has pre-configured the options and formats
from a multitude of satellites and sensors. The figure shows the dropdown "Open Image
File" menu to open an image.
In this example we will open a SPOT image, in this case go to option and select SPOT
"GeoSPOT", since it is stored in a geotiff image. At that moment, a window for selecting
the file path turns up. Once it is selected, another window with the available bands will
appear. Since it is a multispectral image, we will set the RGB image display. In order to
do this, click on "RGB Color" and on the bands that we want to select as a red, green
TELECAN Tutorial
T2.23
and blue (1, 2, 3) Channel. Finally, click the "Load RGB" button. As a result, the ENVI
display will appear. We can observe a strange color setting, where the vegetation zone
has a reddish color. This is because the sensor does not have the Blue channel -green,
red, near infrared and mid-infrared-, so you are representing GR-NIR.
TELECAN Tutorial
T2.24
We can observe the presence of 3 viewers, the main one, a navigation viewer that
displays the entire image, and a zoom [x4] viewer. Furthermore, we observe that once
the image is loaded, the list of images’ menu is linked to the viewers by the bottom
button "Display # 1", which corresponds to the numbering of the Viewer windows. If you
want to load a new image, you need to create a new display in the "Available Bands
List".
In the main viewer, there is also a top menu with multiple options. Under "File" you may
modify display preferences. You may save in the disk or print what is displayed in the
viewfinder. In "Overlay", among other options, you may enter grid lines, areas of interest
and vector data. The menu "Enhance" provides options such as "stretching", "Histogram
Matching" and filtering. "Tools" provides multiple options, such as how to link viewers to
visualize the same areas. You may also create regions of interest ROI or color
grayscale images, window location and value of the pixels, etc. The "Window" menu
manage viewers and perform actions such as creating a new viewer and locking it,
presenting information in the image, etc.
Another ENVI’s basic utility is the "Save file as" (see figure) that allows to save to disk
or open images processed by this program. In this menu we can choose a variety of
image formats such as ENVI, ArcGIS, ERDAS, JPEG2000, GeoTIFF, etc. Once the
format is selected -e.g. geotiff format- a selection window will appear allowing you to
choose the data open in the program. In this interface, you may select the subset of the
image space using "Spatial Subset", where you may select a specific area of the image.
by introducing x-y points or by ROI. By clicking on "Spectral Subset", a window pops up
where you can select the bands that you want to keep.
TELECAN Tutorial
T2.25
The "Basic Tools" menu contains a variety of useful functions for image manipulation,
being the most outstanding the options for resizing images, cropping areas, changing
the resolution of the pixels, image rotation, changing the storage format of the
information -Interleave: BSQ, BIL, BIP. It also has statistics and measurement tools,
image segmentation tools, mosaics, masks, etc. It allows to perform mathematical
operations between bands. At the end of the list you find the option "Preprocessing",
where there are tools for image calibration -obtaining radiance and reflectivity-,
atmospheric correction, and other types of corrections.
TELECAN Tutorial
T2.26
TELECAN Tutorial
T2.27
CONTENTS
IMAGE PROCESSING: CORRECTIONS AND THEMATIC

CLASSIFICATION
A. Remote Sensing Fundamentals
3.1. INTRODUCTION
3.2. REMOTE SENSING IMAGES CORRECTIONS
MONTEVERDI
ENVI
SEADAS
3.3. REMOTE SENSING IMAGES THEMATIC CLASSIFICATION
ENVI
MONTEVERDI
TELECAN Tutorial
Image Processing: Corrections and Thematic Classification
T3.2
3.1. INTRODUCTION
This practice is an introduction to the most common treatments performed with space
remote sensing images. We will use Monteverdi, ENVI and SeaDAS. In the first part of
this practice, we will see the different utilities that allow the realization of radiometric,
atmospheric and geometric corrections, while in the second part, we will carry out
processed oriented to the classification of different covers on an high resolution satellite
image.
3.2. REMOTE SENSING IMAGES CORRECTIONS

Radiometric, atmospheric and geometric correction are the most important
preprocessed remote sensing, designed to get the correct values of the energy reflected
or emitted on any point of the Earth's surface.
Radiometric and atmospheric corrections -also called calibrated images- are a critical
step in the processing of remote sensing images in the optical domain. The calibration
allows obtaining a physical parameter independent of lighting conditions and even
atmospheric conditions, allowing for example to work with images taken in different
dates to detect changes.
The geographic information management of remote sensing data is another point of

great importance. Thanks to the geographical information contained in remote sensing
data, images may be superimposed and their values may be compared. This requires
using the same images projection models. Images must be registered to each other,
which allows the images points to match properly.
TELECAN Tutorial
T3.3
In the next sections, we present practical cases of these corrections for each one of the
software tools described above (Monterverdi, ENVI and SeaDAS).
 MONTEVERDI
"Calibration" "Optical calibration" is Monteverdi’s main menu. It allows the calibration

of high and very high resolution images (IKONOS-2, SPOT4-5, QuickBird, WorldView2
and Formosat2). The calibration results in a luminance or radiance image, another high
reflectivity of the atmosphere TOA image, another surface reflectivity TOC image, and a
difference TOA-TOC image.
After selecting the option "Optical calibration" a "Set input" window pops up where you
have to select the image you want to edit (e.g. WorldView-2 satellite). It is very
important to check that this image is together with the metadata file that contains the
necessary information to make corrections. For this reason, you have to make sure that
the file name has not been modified from the original data and no metadata has been
removed. Once OK is pressed, the calculation of the atmospheric correction using the
6S atmospheric model will be performed (Second Simulation of a Satellite Signal in the
Solar Spectrum), which may last a few seconds. Once the calculation is complete, the
following "Optical calibration module" window will pop up.
TELECAN Tutorial
T3.4
The type of aerosol that exists in this area (coastal, continental, desert, etc.) is
configured in this module, and it is very important to introduce the optical thickness
(Aerosol thickness) of the atmosphere. This data is available at:
http://disc.sci.gsfc.nasa.gov/giovanni/overview/index.html #
The result of the correction value can be improved by entering the ozone value and the
atmospheric profile obtained by Aeronet system weather balloons. This data can be
found at:
http://www.weather.uwyo.edu/upperair/sounding.html, ftp://ftpdatos.aemet.es/ozono/
It is advisable to provide the module with spectral sensitivity data of the satellite bands
to complete the configuration. These spectral sensitivity data are provided by the
companies managing the satellites. Once all this is configured, click "Save / Quit" to
generate luminance calibrated images, TOA reflectivity and TOC.
TELECAN Tutorial
T3.5
The following figure shows the main interface with the optical calibration results
(luminance, TOA, TOC and TOA-TOC), and the TOC surface reflectivity image
atmospherically corrected with data values between 0 and 1.
Monteverdi contains a few geometric correction utilities available in the main menu
"Geometry", among which "Reproyect image", "Superimpose two images" and
"Homologous points extraction" are the most outstanding.
"Reproyect image" allows to modify the type of image projection. Once this option is
selected, a "set input" window will pop up and there the image to be reprojected can be
selected. Then the image is selected and the module "Projection" (see figure) pops up
where you set the options of the new projection. In "Output image" you may configure
the type of projection (UTM, LAMBERT2, WGS84, and EPSG). In "Input image" you can
see the georeferencing format of the input image. In "Settings", you may configure the
interpolation method used (linear, Nearest, BCO). Finally in "Quicklook", you can see a
preview of the reprojection. After pressing the button "Save / Quit", the reprojected
image is generated in the main interface.
TELECAN Tutorial
T3.6
Selecting "Superimpose two images" opens a "Set inputs" window (see figure) where
you have to select the image to be reprojected and the reference image. Pressing the
OK button, makes the "Open dataset" window pop up, where you set interpolation and
select the DEM elevation model, if available. Pressing the OK button generates the
reprojected image in the main interface.
TELECAN Tutorial
T3.7
The result of this simple operation is a reprojected image based on the projection of the
reference image.
Finally, selecting "Homologous points extraction" opens a "Set inputs" window where
you have to select the reference image (Fix) and moving image (Moving). Pressing the
OK button makes the "Homologous point extraction" module window pop up (see figure)
where there will see two groups of viewers with fixed and moving images. In Viewers
"Zoom", control points that are present in both images may be selected. They will be
added to the list of "Point List" by clicking on "Add". Once you have selected the control
points, select the processing method that allows calculating the displacement of the
moving image with respect to the fixed one (Translation, Affine and 2D similarity). We
also have to select the method of grinding the image (Rectify moving image,
Superimpose moving to fix). With the parameters set, press the "Evaluate" button to
obtain the displacement of the moving image with respect to the fixed. Moreover, we get
the error produced between the different control points to perform the transformation. To
finish, just press "Save / Quit" to save the result in the main interface.
TELECAN Tutorial
T3.8
 ENVI
ENVI provides calibration utilities for many optical sensors of remote sensing satellites.
Besides allowing to obtain the radiance or reflectivity TOA of the images, ENVI provides
atmospheric correction modules and allows simple correction using methods based on
extraction of dark pixels. In any case, FLAASH it the most widely used module to
perform atmospheric correction. It is based on the atmospheric model MODTRAN
(MODerate Resolution atmospheric TRANsmission).
To perform the calibration and atmospheric correction, we can use a LADSAT 7 image
called LandsatTM_JasperRidge_HRF.FST.
Then go to "Basic Tools" "Preprocessing" "Utilities Calibration" "Landsat TM",

which displays a selection window of the image. Once we select the Landsat image, a
calibration window pops up (see figure), where you set the satellite type, time of image
acquisition, and solar elevation angle. This data is entered automatically by the
TELECAN Tutorial
T3.9
program, since it get them from Metadata. If they were not available, we would have to
provide them. Finally, we select the radiance or TOA reflectivity calculation and the
output of the processed data. Select radiance, since it is the FLAASH algorithm's input
data, and store the result in memory.
Next you need to make an adjustment of the physical units of the bands, since the
radiance obtained is in [W / (m ^ 2 * sr * nm)], while FLAASH requires data in [μW / (cm
^ 2 * sr * nm)]. Thus we have to apply the bands a correction factor of 10. To do this, go
to "Basic Tools" "Band Math" and a window pops up where you introduce the
expression b1 / 10.0. Press OK and the "Variables to Bands Pairings" window pops up.
Then select the band you want to divide by 10, e.g. band 1. Finally we set the path of
the output image JasperRidgeTM_radiance.img and click OK.
TELECAN Tutorial
T3.10
Next we prepare the image for atmospheric correction. To do this we must change the
interleave, since for spectral processing it is better to use BIP or BIL formats instead of
BSQ, which is usually the default format used in the images. If you go to "Basic
Tools" "Convert Data (BSQ, BIL, BIP)", a window pops up and allows you to select
the image you want to process. Once it is selected, the following window pops up,
where you select the interleave type you wish to select. Select YES in the "Convert In
Place" option to modify it in the same image.
TELECAN Tutorial
T3.11
In this way we have created an image of radiance (band 1) configured correctly for the
input format FLAASH. Then proceed to run the "Basic Tools" "Preprocessing" 
"Utilities Calibration"  "FLAASH" module, where the following window pops up.
TELECAN Tutorial
T3.12
For configurating it:
1. Press the "Input Radiance Image" button and select the image radiance. When
the "Radiance Scale Factors" image dialog pops up, choose "Use Single scale
factor for all bands". Since the units are correctly formatted, enter the value 1. If
you had not done previously, you would have to enter a factor of 10.
2. Press the button "Output Reflectance File" and enter the path of the corrected
image.
3. Press the button "Output Directory for FLAASH Files" to save the files needed to
make the correction.
4. Introduce the root to the generated files in "RootName for FLAASH File".
5. Press the button "Restore..." where you can select the file with the configuration
data required for the atmospheric correction of the image
(JasperRidgeTM_template.txt). This will fill the geometric parameters of the
satellite angle and sun, as well as the atmospheric settings and aerosols.
6. Press "Multispectral Settings" to see the functions of the filters of the bands. The
bands used to calculate water vapor are not set because the satellite does not
have bands that can be used for this purpose. For this reason, the "Water
Retrieval" interface option indicates "No". In the same way, "Kaufman-Tanre
Aerosol Retrieval" can be useful to see which bands have been used.
7. Finally, press "Apply" for atmospheric correction. The process may take a few
seconds as you have to run the MODTRAN atmospheric model. Once the
implementation finishes, the TOC surface reflectivity image in integer format (0-
10000) is generated.
TELECAN Tutorial
T3.13
The following figure shows the corrected image JasperRidgeTM with surface reflectivity
values between (0-10000). See the spectral profile that represents the value of the pixel
channels selected by the cursor.
ENVI provides geometric correction utilities in the "Map" menu. In this menu, you can
find modules for georeferencing multiple remote sensing satellite sensors. You can also
find orthorectification modules, mosaic and registered. While georeferencing and
orthorectification modules are used in the lowest levels of image processing, the
registrant is a useful function when you want to compare images of the same
geographical area.
TELECAN Tutorial
T3.14
To use the registered module of ENVI, first you have to open in the viewers the two
images that you will use for registration -in our example Maspalomas WV2 images.
Then go to "Map" "Registration"  "Select GCPs: Image to Image". A display base
dialog selection (in the example, we selected the image of 2009) pops up and also the
Display to modify the image ("Warp") (image of the year 2013).
Press OK to display the manager to select the ground control points GCPs. Mark the
control points in both viewers with a double click. In the "Options" menu, you can find
the option "Automatically Generate Tie Points..." where the module selects control
points automatically. This system cannot usually provide optimum results. For this
reason, many times, you will have to it manually. Select each control point in both
displays and then press "Add Point" to add the item to the list. After selecting enough
points to make the registered list, select "Save GCPs to ASCII..." to save in the disk the
list of points. The figure shows the viewers with the selection and the list of points that
must be stored on disk.
TELECAN Tutorial
T3.15
Once you have the control points for registration, perform the operation using the menu
"Map" "Registration"  "Warp from GCPs: Image to Image". After selecting the menu
option, a file selection dialog GCPs points pops up. There select the image to be
modified, and then the reference image.
TELECAN Tutorial
T3.16
Once the selection of images is finished, a configuration window for the registered
parameters pops up. You can select the method to deform the image to adjust it to the
control points, such as the "Polynomial". You can also select the interpolation method,
for example, "Bilinear". Once it is configured, press the Ok button to generate the
registered image.
Finally we can compare the result by linking together the reference and the registered
image.
TELECAN Tutorial
T3.17
The mosaic module "Map"  "Mosaicking"  "Georeferenced" provides the utility to
generate mosaics using georeferenced images. We are going to use the images WV2
to show it. Once the module is selected, a window "Map Based Mosaic" pops up. In this
window, the mosaic is generated. Go to "Import" and select the 2009 and 2013 images.
After a few seconds, a mosaic with the two images will be generated. The two regions
are displayed separately by green and red edges. The two images overlap in the center
prevailing the green edge image. Once the mosaic is generated, it can be saved to the
disk.
TELECAN Tutorial
T3.18
 SEADAS
SeaDAS allows reprojection of images accessing the menu "Tools" "Reprojection...".

After the selection, a configuration window pops up to configure the parameters of the
reprojection. In "Projection Parameters" the projection type may be set -geographic
WGS84 by default- and it allows setting the interpolation method. The output data type
and path is set in "I / O Parameter. Once the image is reprojected, which can take
several minutes, the new image is stored on disk and accessible for viewing.
TELECAN Tutorial
T3.19
TELECAN Tutorial
T3.20
3.3. REMOTE SENSING IMAGES THEMATIC CLASSIFICATION
Classification is an important step, since it allows to generate a thematic map with the
classes of interest from the numerical data of the spectral bands.
Next, we use ENVI and Monteverdi softwares to generate a classified image with a
supervised and an unsupervised method. SEADAS does not allow making
classifications.
 ENVI
Supervised Classification Process
In the supervised classification, the user selects training areas. These training areas are
defined as groups of pixels that represent a homogeneous area or material that will be
associated with a class. Regions of interest (ROIs) are used in ENVI to create the
training areas. Also, you can ensure separation of classes to avoid overlap between
them by representing them on a n-dimensional (nD Visualizer) map.
ENVI has implemented a variety of classification algorithms, as we can see in the

following figure. In this practice we will focus on the Maximum Likelihood algorithm,
because it provides one of the best results. Maximum Likelihood classifier assumes that
the statistics for each class in each band are normally distributed and calculates the
probability that a pixel belongs to a specific class. Unless you select a threshold, all
pixels are classified. Each pixel is assigned to the class with the highest probability. If
Maximum Likelihood is less than the specified threshold, the pixel is not classified.
TELECAN Tutorial
T3.21
Before applying the classifier, it is necessary to create the regions of interest that we are
going to associate to the classes that we want to get to the output of the classifier.
Select Tool -> Region of interest -> ROI Tool.. and add the different classes, assigning
a name and color, to be easy to identify them in the classified image.
TELECAN Tutorial
T3.22
Once we have chosen several regions of interest for each class, we can make a
separability analysis, as shown in the following figure. It returns the Jeffries-Matusita
indexes and Transformed Divergence that measure the statistical independence of the
pairs of ROIs. They have a range that goes from [0 2]. Values greater than 1.9 indicate
good separability. For values less than 1,it is recommended to combine both classes.
In the case shown above all classes have a good separability.
In addition, we also have the option of performing a statistical analysis of the classes.
The information of ENVI is shown in the figure. We can obtain the spectral response of
each of the classes created.
TELECAN Tutorial
T3.23
The next step is to save the ROIs to proceed to use them in the classifier. Close the
window and select ROIs Tool on the Overlay-> Classification main menu. The following
window pops up. As we can see, it allows limiting the classification to an area or specific
bands. You can also add a mask to indicate pixels that are not used by the classifier.
TELECAN Tutorial
T3.24
Select the configuration parameters of the classifier:
Single Value: Use a single threshold for all classes. Enter a value between 0 and 1 in
the Probability Threshold field. ENVI does not classify the pixels with values below this
value.
TELECAN Tutorial
T3.25
Data Scale Factor: A division factor used to convert scaled radiance or reflectance
values to integers float values.
Rule Output Images: The rule images are created as an intermediate form before
creating the image with the final allocation of classes. Then you can use the Rule
Classifier to create likelihood images of each class. The white areas represent those
belonging to that class. Also, we can preview before generating the final version.
The next method for improving the classification results is to perform a post-
classification process, where we can apply filters (option Majority / Minority Analysis).
TELECAN Tutorial
T3.26
In that Post Classification menu, there is also the option to evaluate the accuracy of the
classified image by calculating the confusion matrix and kappa coefficient.
TELECAN Tutorial
T3.27
The confusion matrix evaluates the conflicts that occur between classes. To get it, a list
of test points whose actual coverage is known is compared to a list of points deduced
by the classifier. In a confusion matrix, rows correspond to the reference classes and
columns correspond to the classification derived classes. The diagonal represents the
points at which there has been agreement between the actual and derived classes. This
matrix is very useful to measure the accuracy and reliability of the classifier. The
example shows that for the Class Grass, 64516 pixels have been correctly classified.
However 26949 pixels have been Unclassified (not classified), 7277 have been
classified as Forest and 10742 as Swamp.
Unsupervised Classification Process (ISODATA)
Within the unsupervised classification algorithms the most widespread is called

ISODATA.
Here you are the configuration window ISODATA classifier.
TELECAN Tutorial
T3.28
Select the number of classes to be defined. This range is required for the algorithm
groups and it divides the pixels based on thresholds.
Introduce the maximum number of iterations and the threshold of change. The last one
is used to terminate the iterative process, when the number of pixels in each class
changes by less than the threshold.
Introduce the minimum number of pixels required to form a class.
The next parameter defines the maximum standard deviation between classes
(Maximum Class stdv) in digital levels (DN).
Finally, introduce the minimum distance between the means of each class -also in DN-
and the maximum combined couples.
In principle, the unsupervised analysis becomes an iterative process where the user
sets arbitrary values that are modified according to the results.
TELECAN Tutorial
T3.29
Filtering may applied to improve the classifier quality, merge classes, change colors,
etc.
TELECAN Tutorial
T3.30
 MONTEVERDI
Supervised Classification Process
Next, we will make a practical procedure to define the different classes over the image,
generating different regions of interest. Then we will run the classification algorithm to
see the results we have got on the classification process. The process is quite similar
following the options that appear on the screen.
Unsupervised Classification Process
Next, we proceed to run an unsupervised algorithm. Specifically Monteverdi has

implemented KMeans algorithm. Thus, we can compare the results obtained in both
classifications.
TELECAN Tutorial
T3.31
The requested values will be completed by the tool. Then press OK.
TELECAN Tutorial
T3.32

Remote Sensing PDF

Uploaded by

Copyright:

Available Formats

Remote Sensing PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Remote Sensing PDF

Uploaded by

Copyright:

Available Formats

REMOTE SENSING

The project "Programme for the Development of Technological Networks

TELECAN is a three-year project led by the University of Las Palmas of

 Provide knowledge of space remote sensing, physical fundamentals,

 Acquire knowledge on the techniques of digital image processing

 Know the various data sources available for imaging.

 Know and properly handle the major existing software tools.

 Acquire the skills and abilities to generate remote sensing products

 Encourage the use of remote sensing.

This document entitled "REMOTE SENSING TUTORIAL TELECAN"

1. - Fundamentals of Remote Sensing

2. - Sensors and Remote Sensing Space Missions

3. - Remote Sensing Applications

4. - Radiometric, Atmospheric and Geometric Modeling

5. - Remote Sensing Images Processing

1. - Collection and Analysis of Remote Sensing Images

2. - Tools Image Processing Software

3. - Modeling and Remote Sensing Image Processing

In a globalized world citizens are informed daily on the presence

In addition, humans have always had a special interest in

Remote sensing is an essential tool to provide Earth

"ISPRS: measurement or acquisition of information from

PASSIVE REMOTE SENSING ACTIVE REMOTE SENSING

"Considerably improve our knowledge of our

 Global coverage and regular large areas of the Earth.

 Digital processing of the received images.

 REMOTE SENSING: COMPLEMENTARY TECHNOLOGIES

 Remote sensing started in 1860 with a photograph of the Earth's surface,

 First Earth observation satellite in 1960 (TIROS-I).

 ELECTROMAGNETIC SPECTRUM: SOURCES AND USES OF THE

 Microwave (1mm-1m): There is a growing interest of Remote Sensing in this

 EMISSION RATES USED BY REMOTE SENSING

1. Radiation reflected on Earth's surface emitted by the Sun. It depends on the

Any object with T> 0K radiates energy.

 Wien's Displacement Law

 Direct relationship between temperature and the amount of electromagnetic

 The hotter is the object, the more energy is generated.

 Small variations in temperature Large variations in the energy emitted (T4).

 Real bodies are not black bodies.

 Interactions of the Atmosphere

 The atmosphere also emitsThermal IR.

 MAIN AGENCIES AND ORGANIZATIONS

From past to present

Scientific missions: "Earth Explorer"

 Following the example of meteorology, GMES develops operational services, but in

Missions that may contribute to GMES

Sentinel-1: Mission SAR band C

 Four nominal operating modes:

 Helio-synchronous orbit at 693 km altitude

 Nominal 12-day revisit period

 Expected to last 7 years, having fuel for 12 years

 Land cover maps.

 Super-spectrometer with 13 bands (VNIR & SWIR)

 Spatial resolution: 10, 20 and 60 m

 Satellite mass: 1200 kg

 Revisit period: 5 days (without clouds) with 2 satellites

 helio-synchronous orbit at 786 km altitude

 Expected to last 7 years, having fuel for 12 years

 Color and temperature of the sea surface and ground.