DEBRE MARKOS UNIVERSITY, BURIE CAMPUS

DEPARTMENT OF COMPUTER SCIENCE

A Research proposal on
AN EMPIRICAL STUDY OF CLIMATE CHANGES IN ETHIOPIA BASED ON
EFFICIENT DATA MINING TECHNIQUES

By
1. kerebih Getnet
2. Dejenie Aynalem

Oct 2017

i
Table of Contents

List of acronyms i

Summary of the proposal ii

1. Introduction 1

2. Statement of the problem 3

3. Objectives of the Thesis 4
3.1 General objective 4
3.2 Specific objectives 4

4. LiteratureReview 5

5.Methodology 7
6. Significance of the study………………………………………………………………………5

7. Work breakdown structure 11

8. Cost breakdown 13

References 14

i
 Introduction

Climate change refers to any significant change in the measures of climate lasting for a long
period of time. Climate change includes changes in temperature, precipitation, or wind patterns,
with other effects, that occur over decades or longer. Over the past, human activities have
released large amounts of CO2 and greenhouse gases into the atmosphere. CO2 and Greenhouse
gases come from burning fossil fuels to produce energy, deforestation, industrial processes, and
agricultural practices. Climate change has impacts on water supply, flood prevention, ecosystem,
contaminant management, etc.
To evaluate climate change and its effects, data mining, machine learning and climate
circulation models are the main techniques.
Data mining is a well known technique used to focus on extracting knowledge from datasets.
This knowledge is represented in the form of model which provides description of the given data
and allows predictions for new data.
This predictive ability makes data mining a valuable candidate for climate change and
biodiversity modeling.
The impacts of present and future climate change will be one of the most important
scientific and societal challenges in the 21st century [2]. Given observed changes in temperature,
sea ice, and sea level, improving our understanding of the climate system is an international
priority. This system is characterized by complex phenomena that are imperfectly observed and
even more imperfectly simulated. With tremendous supply of climate data from satellites and
environmental sensors, the magnitude of data and climate model output is beginning to overcome
the relatively simple tools currently used to analyze them and data mining approach will compute
better these analysis challenges.
Currently ML is used to focus on extracting knowledge from complex datasets. This knowledge
is represented in the form of model which provides description of the given data and allows
predictions for new data [7].

1
 2. Statement of the Problem
In Ethiopia, climate change effects have been experienced over a long period of time through
influences on livelihood activities and changes in the environment [5].
Although the different global and downscaling models have made different climate projections
for Ethiopia, there is a general consensus that Ethiopia will see greater climate variability and
extreme events in coming decades. The report indicated by [10] the National Metrological
Agency (NMA) in the National Adaptation Program of Action (NAPA) of Ethiopia shows that in
the future rainfall will decline in some parts while increasing in others parts of the country.
Other studies using statistical model indicated that there was estimated a different result.
The report argued that the differences in climate estimations are a matter of different data and
also reflect the extent of uncertainty about the situation. It is also indicated that giving attention
to increasing variability and extreme events is important for to Ethiopia.
The climatic change results changes in precipitation patterns, rainfall variability, and temperature
is most probably raise the frequency of droughts and floods in Ethiopia. The country’s highly
dependence on rainfall and survival agriculture increases its vulnerability to undesirable effects
of these changes.
In general, the level of vulnerability of different social groups is determined by both
socioeconomic and environmental factors. The poor rural people who do not have access to
different livelihood options, infrastructure, and institutional setups are known to be the most
vulnerable.
The Intergovernmental Panel on Climate Change (IPCC) [4] indicates that vulnerability to
climate change depends on adaptive capacity, sensitivity, and exposure to changing climatic
patterns. In the Ethiopian context, the farming community is the most vulnerable because of their
high dependence on agriculture for their survival income.
The farming community such as subsistence farmers and pastoralists are more vulnerable to
climate change related risks like drought which results shortage of food and water and diseases,
etc.
Generally, limited assessments have been made on climate change and its consequences on
various sectors such as socially, economically and politically conditions in Ethiopia [4].
Therefore, predicting climate properly and understanding the effects of climate change is very
important in our country.

2
This research will deal with predicting climate properly and understanding the effects of climate
change for Ethiopia focusing on the following major issues:
 Lack of data records and poor meteorological information to study the impact.
 The traditional methods are unable to estimate non-linear weather elements.
 Limited assessment has been made in Ethiopia.
 Climate effects are different in regional level due to topography and human activity.
3. Research Questions
The research questions to be addressed in this study are:
 What is the effective data mining method to predict the climate variables of that has
complex topography?
 How should these variables be represented in order to derive accurate predictive future
conditions?
 Can we indicate climate changes of the past and predict the future in Ethiopia using data
mining?
 Is the correlation is at an acceptable accuracy to use data mining in complex regions of
Ethiopia?
1.3 Objectives
1.3.1. General Objectives
The objective of the study is to develop a model to show the past climate changes and predict the
future of specific regions of Ethiopia using data mining techniques.
1.3.2. Specific Objectives
To achieve the general purpose, the study attempts to address the following specific objectives:
 Literature review related to climate variables.
 Collect appropriate spatial and temporal climate data from sampled areas that can be
representative for Ethiopian climate conditions.
 Preprocessing and transforming the data.
 Investigate the appropriate data mining techniques that can be used predict the past and
the future climate changes in sampled area.
 Process the models using the appropriate data sets.
 Evaluate the results of models .

3
 1.4. Significance of the study
The research is expected to have the following significance.
 Providing climate variable data to environmentalists, natural resource managers and
decision makers on important issues to map and understand the climate behavior of
regions.
 It will help to predict the climate and knowing its effects in order to take adaptation
measures to improve livelihood, the environment and rehabilitation of natural
resources.
 Helps for Ethiopia to manage the climate variability including improved a
management of land resources, including soil, water, forests and agriculture.
 And knowing and understanding the climate of the environment is used to take
actions and make decisions for decision makers to take proactive actions than reactive
active measures.
1.5. Scope and Limitation
The availability of data related to the climate of Ethiopia and geographical factors of the climate
are the major limitation of this study. This problem hinders the full investigation, prediction and
its impacts of climate change in Ethiopia at large. Climate change study is quite complex and
involves addressing a lot of issues and problems. Due to the time limitation imposed on this
research we forced to put some boundaries. With this limitation, the scope of this research work
can be stated as follows:
 Study the climate by using two climate elements only: temperature and rainfall.
 Specific locations will be selected that represent Ethiopian climate conditions
(West Gojjam).
 Relatively short period of time (from two to four decades) climate data used to show
the climate data to investigate the past climate data and to predict the future of
sampled locations because there is no long period of time climate data available in
Ethiopia.

 Lack of well-distributed and long-term records of daily temperature and rainfall

data

4
1.6. Methodology

A brief explanation of the methodology will be elaborated under this section.

a) Literature review and related works

Literatures related to this work will be reviewed to assess the efforts that have been made so far
in climate variations and its characteristics in Ethiopia. In literature review I will discuss the
general field of climate science that this thesis work is involved in, and describes some previous
work on the application of machine learning techniques to climate variables interpolation and
prediction and its impacts. I also point out some issues about previous studies and their
limitations.
By focusing on the techniques of machine learning in interpolation of climate change I

decided to focus on literatures, articles and other publications that share ideas on climate and its

impacts, datasets and machine learning algorithms. Literature sources that I will review are

magazines, books, research articles, conference papers, publications, and blogs.

b) Data Collection Methods

All options of source of data in related to Ethiopia climate will be assessed. The three options of
climate data related to Ethiopia are;
1. National Meteorology Agency (NMA); gather weather station temperature and rainfall data
from Ethiopia National Meteorology Agency found in Addis Ababa main office and Bahir Dar
branch office.
2. Cape Town University. African countries climate data can obtain from Cape Town

University, South Africa.

3. Datasets from NOAA and IPCC DDC: There are data sets available in NOAA (National
oceanic and atmospheric Administration) center for environmental monitoring and IPCC DDC
(Inter governmental panel for climate change distribution data center) institutions.
The datasets available here are collected from U.S satellites since around 1990.The data sets are
collected from the whole world.

5
 d) Study Area
The study is limited to West Gojjam. The selected weather stations will be selected
from areas where there are dry and wet places that have different geographical variables
to represent the weather conditions of the Amhara Region.
e) Tools and technologies used

We will use different software tools to implement identified models such as Weka or Matlab or
any other appropriate software tools.

f) Implementation approaches

The implementation approaches is assessing the different data mining techniques and identifying
the appropriate model and process the data then evaluate the results.

5. Significance of the study

The research is expected to have the following benefits.
 Providing guidance to natural resource managers and decision makers on important
issues related to the impacts of climate change on all components of biodiversity.
 Climate prediction and knowing its effects knowing its effects to the biodiversity is
used to take adaptation measures to improve livelihood, the environment and
rehabilitation of natural resources.
 The research used to take actions and make decisions for decision makers.

 Helps for Ethiopia to manage the climate variability including improved management
of land resources, including soil, water, and forests.

6
 3. Work Break down Structure

Oct Nov Dec Jan F

Literature
Review
Data
Collecti
on
Preprocessing,
Transforming

Implementati
on and

Discussion &
Conclusion
Report Writing and
Editing

7
 4. Cost breakdown

Proposed Financial Requirement

No Item Unit Total Remark

1 Transportation Two person for

5 trips
 Debre Markos 40 800
 Bahir Dar 60 1200

 Addis Ababa (??) ?? ??

2 Stationary
 Flash Disk 250 500
 External Hard Disk 3000 3000

3 Tele

Total

8
 References
[1].Durgalalshrestha, “In rainfall-runoff modeling: Application of machine learning
technique”s, 2009
[2]. Claire Monteleoni, GavinA, et al, “ClimateInformatics”, chapter 4.
[3].Ethiopian Economic Association/Economic Policy Research Institute, “The social dimension
of adaption to climate change in Ethiopia”.
[4]. An assessment of the intergovernmental panel on climate change, “Climate change 2001:
Synthesis report”
[5]. MaumitaBhattacharya, “Machine Learning for Bioclimatic Modeling” , IJACSA, Vol. 4, No. 2,
2013
[6]. Leonardo OrnellaGerardo et al, “Applications of Machine Learning for Maize Breeding”.
[7]. NehakhElwalruchidavey,Climatic Assessment of Rajasthan’s Region for Drought with
Concern of Data Mining Techniques:Vol. 2, Issue 5, September- October 2012.
[8].Abbas Golestaniand Robin Gras, “Using Machine Learning Techniques for Identifying
Important Characteristics to Predict Changes in Species Richness in EcoSim, an Individual-
Based Ecosystem Simulation:May 2012 / Vol. 62 No.
[9].YongyutTrisurat ,Rajendra P. Shrestha , Roger Kjelgren;Plant species vulnerability to
climate change in Peninsular Thailand.
[10]. National meteorological agency; http://www.ethiomet.gov.et.