See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/283300725

A PYTHON TOOL FOR AUV-BORNE ADCP CURRENTS DATA PROCESSING

Conference Paper · September 2015

CITATIONS READS

2 551

3 authors, including:

Simo Cusi Pablo Rodriguez

Ensenada Center for Scientific Research and Higher Education Spanish National Research Council
22 PUBLICATIONS   274 CITATIONS    15 PUBLICATIONS   28 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

GABS (JERICO-TNA) View project

Consorcio de Investigación del Golfo de México View project

All content following this page was uploaded by Simo Cusi on 29 October 2015.

The user has requested enhancement of the downloaded file.

 A PYTHON TOOL FOR AUV-BORNE ADCP CURRENTS

DATA PROCESSING

S. Cusí, P. Rodríguez and N. Pujol

UTM-CSIC, Marine Technology Unit (CSIC), Barcelona, Spain

Abstract – Most Autonomous Underwater Vehicles measured by AUVs to those measured by

(AUVs) mount Doppler sensors to navigate precisely moored ADCPs [1]-[3], in order to explore the
underwater, where the Global Positioning System
(GPS) is unavailable. These sensors, aside of AUVs’ capabilities as reliable current meters.
providing accurate AUV velocity with respect to the Results show promising differences of about 0.1
ground, can perform currents profiling, measuring ms-1 in horizontal velocity values. A comparison
currents along the water column. It has been shown
study evaluating the OceanServer Iver2 AUV
that currents measurements taken by AUVs are very
close to those taken by bottom-mounted ADCPs and versus a bottom-mounted Nortek AWAC,
that a 3D data processing approach can yield Acoustic Wave and Current Meter, [4], shows
differences between both instruments of about 0.07 that 3D AUV-borne ADCP data processing
ms-1, averaging AUV data in 90 second time
windows. In this paper we present an OceanServer yields differences of about 0.07 ms -1 with
Iver2 AUV 3D water currents processing tool, respect to the AWAC, using data averaging time
developed in Python 2.7. The tool outputs .csv files windows of 90 seconds.
for further data processing/representation as well as OceanServer Iver2 AUVs are one of the most
plots of the main variables, along with water currents
plots. commercialized AUV models due to their
relatively low cost and ease of operation.
Keywords – Autonomous Underwater Vehicle, However, no data processing software is
Acoustic Doppler Current Profiler, OceanServer
Iver2, SonTek ADCP, Python 2.7
provided by the manufacturer. ADCP currents
data present higher processing complexity than
other data as the measurement points or cells are
I. INTRODUCTION away from the vehicle and, the measurements
being vectors, require coordinate transformation
Autonomous Underwater Vehicles (AUVs) to match the North-East-Down directions. Also,
can run for several hours (~10 h) and kilometers Doppler technology measurements contain
(~50 km) collecting biogeochemical/physical Gaussian noise that needs to be removed by
data of water masses, exploring the seafloor averaging data during a certain period of time.
with acoustic or optical technologies, or In this text we present a first version of
executing specific tasks with other commercial Python 2.7 scripts aimed at processing Iver2
or custom sensors. One sensor most AUVs ADCP water currents. It explains how data are
mount is the Doppler Velocity Log (DVL), processed as well as filtered and averaged. It
which provides valuable data on platform also explains what are the graphical outputs and
velocity with respect to the bottom (bottom- generated files. These scripts are to be freely
tracking); used to accurately navigate distributed and modified.
underwater, where the Global Positioning
System (GPS) becomes useless. In many
occasions, the DVL can perform as an Acoustic II. INSTRUMENTS
Doppler Current Profiler (ADCP), turning the
AUV into a potential currents measurement A. Iver2 AUV
tool. This capability, however, has seldom been
exploited, judging by the amount of studies one Iver2 is one of the lightest man-portable
can find in which oceanographers use AUV commercial AUVs (Fig. 1). In spite of its low
currents data to characterize water circulation. weight (~20 kg) and size (~15 cm diameter and
Some studies have compared currents data ~1.5 m long), it can carry a DVL/ADCP, a side

SIXTH INTERNATIONAL WORKSHOP ON MARINE TECHNOLOGY, Martech 2015 Page 56

Cartagena, September 15th, 16th and 17th - ISBN: 978-84-608-1708-6
 scan sonar, optical cameras and a CTD. The kHz, Δz is the cell size in meters, and N is the
Ecomapper version mounts an YSI multi- number of pings in the average. The last 2
parametric sensor capable of measuring variables can be modified when setting up the
different biogeochemical parameters such as PH ADCP in order to reduce the noise. The 4 beam
and Chlorophyll. These AUVs can perform configuration allows for redundant calculation
continuous surveys down to 100 m depth for of vertical water velocity, which is used to
about 10 h at speeds ranging from 1 to 4 knots. evaluate the quality of the measurement. The
AUVs PC logs the standard error velocity
(SEV) for every measured cell, representing the
sum of the difference between the calculated
vertical velocities. The manufacturer
recommends discarding the cells in which the
SEV value is higher than 0.5 ms-1.

C. Python 2.7

Fig. 1. The two OceanServer AUVs owned and operated by

The ADCP currents processing tool
CSIC’s Marine Technology Unit. One is equipped for
seafloor imaging (front) and the other for water quality presented here is coded in Python 2.7, a
monitoring (rear). Both are provided with a SonTek common programming language among
DVL/ADCP sensor. scientists and engineers. It is an interpreted
high-level programming language intended for
B. SonTek ADCP general programming purposes. Python counts
with a large community of users worldwide who
ADCPs installed on Iver2 AUVs can have develop and maintain a wide range of libraries.
two different configurations: 6 and 10 beams It is free and can run on a variety of platforms,
[5]. The former setup, with 4 down-looking including Linux, Windows and Mac, among
transducers working at 1 MHz, allows for others.
downward water profiling up to 40 meters and
bottom-tracking for navigation; while the latter,
with 4 down-looking and 4 up-looking III. DATA PROCESSING
transducers working at 1 MHz, provides water
currents measures below and above the vehicle Currents measures are provided at 1 Hz and
covering up to 80 meters, as well as bottom and can be represented by 3D vectors in an AUV-
surface-tracking velocities. Both setups are based coordinate system (Fig. 2). Each of these
provided with one up and one down-looking vectors represents the currents’ value at one cell
500 KHz vertical transducer used to range the (cells are placed along the vertical axis of the
bottom and the surface at distances up to 80 sensor), the number and size of which can be set
meters. The beams devoted to water profiling before the mission starts. Removing AUV’s
are oriented with a 25º slant angle and have a velocity from the measured values is needed to
width of 3.5º. Even though the ADCP delivers obtain absolute AUV-referenced water currents,
averaged data to the AUVs PC for navigation which then need to be rotated, by using AUV’s
and currents logging at 1Hz, the ADCP’s attitude (pitch , roll  and heading ), to match
internal sampling rate is higher: the 10 beam the Earth’s reference system (i.e. North-East-
ADCP internally samples at 2.3 Hz while the 6 Down):
beam one does it at 3.1 Hz –more if ranging less
distance and/or disabling some functionalities-. (2)
The manufacturer [5] defines the Doppler noise
according to the following formula: where J is the rotation matrix (c represents the
cosine function and s the sine function):
(1)
(3)
where σ is the Doppler noise of horizontal
velocity in ms-1, F is the acoustic frequency in

SIXTH INTERNATIONAL WORKSHOP ON MARINE TECHNOLOGY, Martech 2015 Page 57

Cartagena, September 15th, 16th and 17th - ISBN: 978-84-608-1708-6
 This process is done every second as currents (attitude, velocity, position) as well as data on
measures, bottom-tracking velocities and ADCP configuration and performance (cell size,
attitude values are updated at this rate. The 3D number of cells, bottom-tracking quality, noise).
approach could be simplified to a 2D one It also stores other calculated parameters such as
(correcting only for AUV heading), considering cell volume. Data can be discarded by filtering
navigation underwater is steady, with pitch and (min/max values) by any of the mentioned
roll values close to zero, but would make parameters when generating the second,
surface (unsteady pitch and roll) and yoyo (~ conditioned currents data file.
±15º pitch) navigation measures useless. In The tool, aside of being distributed as open
addition, the 3D approach places the Earth- source code, is packaged into a stand-alone
referenced currents vectors according to AUV’s Windows executable file in order to be easily
attitude; which means, for example, that the run on any Windows machine, without the need
most distant cells of a 10-beam ADCP (one 40 of installing the Python interpreter.
m below and one 40 m above) mounted on an
AUV navigating at 15º pitch and neutral roll,
are separated 21.4 m (instead of coincident) in ADCP
the horizontal plane and 77.3 m (instead of 80 measurements
files
m) in depth. .pfd for downward
DVL beams
data file .pfu for upward
beams
.dvl

Match files’
timestamps &
merge data

Fig. 2. Earth-referenced coordinate system (left) and

AUV-referenced coordinate system, which is coincident to
the cells’ coordinate system. The first cell is represented
underneath the AUV in real proportions.
Calculate absolute Earth-
referenced water currents
The tool presented here processes AUV-
Borne ADCP data applying a full 3D approach,
described above, allowing for multi-purpose
Store raw Raw Data Plots
sampling and, therefore, avoiding the need of data in one XY (time, variable)
currents mapping devoted missions. This way, .csv file XYV (time, depth/cell
currents data can be a by-product of a mission number, variable)
where yoyo and/or surface navigation play an Z (variable)
important role.
User Settings
for filtering, binning and
time averaging
IV. RESULTING SCRIPTS

The script takes Iver2 AUV output files Perform filtering, binning
and time averaging
(.dvl, .pfd and .pfu) and outputs one file
containing raw data at 1 Hz. From this file, it
generates another file with filtered (min/max), Store Conditioned
conditioned Currents Plots
depth binned, time averaged (moving average)
data in one Currents vectors
currents (Fig. 3). Both files adopt CSV format .csv file XYZV (latitude,
in order to be compatible with widespread tools longitude, depth,
such as Ocean Data View or Microsoft Excel. currents)
The raw data file, aside of currents data,
includes data describing AUV navigation Fig. 3. Data processing flow chart.

SIXTH INTERNATIONAL WORKSHOP ON MARINE TECHNOLOGY, Martech 2015 Page 58

Cartagena, September 15th, 16th and 17th - ISBN: 978-84-608-1708-6
 V. CONCLUSIONS AND NEXT STEPS REFERENCES

The use of AUVs to map currents is not yet

widespread; in spite of most AUVs carry ADCP [1] M. R. Dhanak, P. E. An, and K. Holappa, “An AUV
survey in the littoral zone: small-scale subsurface
sensors. Currents measured by AUV-borne variability accompanying synoptic observations of
ADCPs show good agreement with those surface currents,” IEEE J. Ocean. Eng., vol. 26, no. 4,
October 2001.
measured by more established techniques, such
[2] D. A. Fong and N. L. Jones, “Evaluation of AUV-
as moored ADCPs. In this paper we introduce a based ADCP measurements,”, Limnol. Oceanogr.:
first version of a tool developed in Python 2.7, Methods 4, pp. 58-67, 2006.
[3] J. Brown, C.Tuggle, J. MacMahan, and A. Reniers,
aimed at processing Iver2 AUV currents data “The use of autonomous vehicles for spatially
using a 3D approach. The tool outputs measuring mean velocity profiles in rivers and
estuaries”, Intel. Serv. Rob. 4, pp. 233-244, 2011.
processed data files as well as a variety of plots.
[4] S. Cusí, P. Rodríguez, and D. Roque, “Performance of
Once released, the tool will need to go an Autonomous Underwater Vehicle Borne Acoustic
through a validation process where user Doppler Current Profiler,” OCEANS’15 MTS/IEE
Conference, Genova, Italy, May 2015.
feedback will be key to adjust usability,
[5] DVL user manual and data format descriptions,
functionality and other aspects. Parallelly, in version 1.1, SonTek/YSI, San Diego, CA, February
order to settle AUVs as valid current meters, 2011.
more experimental work comparing AUV
currents measures to those of other instruments
will need to be done.

SIXTH INTERNATIONAL WORKSHOP ON MARINE TECHNOLOGY, Martech 2015 Page 59

Cartagena, September 15th, 16th and 17th - ISBN: 978-84-608-1708-6
View publication stats