130 
EROSION RISK MAPPING IN THE TROPICS 
J.D.Flach 1 , S.M.White 2 , W.G.Collins 3 , T.R.E.Chidley 3 
1. Image Technology Systems, Aston Science Park, Aston Triangle, 
Birmingham, UK. 
2. NERC Water Resources Systems Research Unit, Dept. Civil 
Engineering, The University, Newcastle.u.Tyne, UK. 
formerly: ODU, Hydraulics Research Ltd, Wallingford, Oxfordshire, UK. 
3. Dept. Civil Engineering, Aston University, Aston Triangle, 
Birmingham, UK. 
ABSTRACT 
This paper investigates the use of remote 
sensing and Spatial Information Systems as a 
decision support tool in river basin management; 
providing information for input to regional 
hydrological models. The test case highlights a 
study undertaken in the Philipinnes to determine 
areas of a river catchment that would benefit 
most from a re-forestation project. The 
integration of various types of data into the 
model is discussed along with the generation of 
an erosion risk map for the catchment. The use 
of weather satellites for areal rainfall estimation 
as input to the model is also discussed. 
KEY WORDS: Erosion Risk, Hydrological Modelling, 
Spatial Information Systems. 
1. INTRODUCTION 
Soil erosion is a major problem in many third 
world countries. The response to rising food 
demand has often been met by increased 
agricultural production in marginal lands, and 
the world trade in timber has caused the 
destruction of many forested upland areas. In 
many cases this has led to accelerated erosion 
rates, total degradation of the land, high 
sediment loads in rivers and canals, and a 
reduction in water available for irrigation. 
Ultimately sediment deposition in reservoirs can 
mean a drastic reduction in operating life. 
Unfortunately the modelling of hydrological 
processes, in particular the prediction of 
sediment yield, in developing countries is often a 
subjective task, hampered by lack of data. 
This paper describes the basis for a remote 
sensing based hydrological information system 
(Ragan and Fellows, 1979), (Allewijn, 1988), and 
its application in the creation of a database of 
useful information for input to distributed 
hydrologic models. 
1.1. The Study Area 
The study area used as a test case for 
developing a methodology was the 4123 km 2 Magat 
River Catchment in Luzon, Philippines. Problems 
of high localised erosion and rapid sedimentation 
in the reservoir at the catchment outlet have 
resulted in plans for re-forestation of parts of 
the catchment. It was neccesary to identify those 
areas of the catchment, which if re-forested, 
would give most return in terms of reduction of 
soil loss at the site and reduction in sediment 
yield to the reservoir. This study and 
subsequent monitoring of the effects of 
catchment management policies is being 
undertaken by the Overseas Development Unit of 
Hydraulics Research Ltd, Wallingford, UK. 
2. EROSION RISK MODELLING 
The prediction of sediment yield in developing 
countries is often a subjective task, and has in 
the past often proved to be extremely unreliable. 
Numerous examples exist where sediment yields 
predicted at the feasibility stages of an 
irrigation project are found to underestimate the 
real yields by a factor of ten or more. One of 
the contributory factors to these unreliable 
predictions is the preponderance of simplistic or 
inappropriate predictive techniques. Formulae are 
either simple, relying on only one or two factors, 
or extremely complex and data-hungry. Most 
formulae have been developed in climatic and 
geomorphic zones which are very different from 
those found in the tropics where many 
developing countries are located. This does not 
stop people from using them! 
The processes of erosion and sediment transport 
are dependent on many factors, whose 
inter-relationships are complex. Prediction of 
erosion and sediment yield is therefore achieved 
by combining the various causative and resistive 
forces for the area under consideration. 
Prime examples of this type of approach are the 
Universal Soil Loss Equation (Wischmeier and 
Smith, 1978) for individual plots of land, and 
catchment scale models that average data over 
the entire catchment (Fleming, 1969). Development 
of computer based distributed models allow the 
catchment to be modelled on a grid cell or 
sub-catchment basis (Beasley, 1960), (Fleming and 
Walker, 1976). Distributed models of this nature 
require sediment routing algorithms to account 
for movement through and between model cells. 
Data requirements for distributed models are 
clearly high which normally precludes their use 
in developing countries. Remote Sensing and 
Spatial Information Systems provide a possible 
source of data for these models. 
As a basis for modelling erosion risk a modified 
version of the SLEMSA model, originally 
developed and used for a study in the SADCC 
region (Stocking 1987), was investigated. This 
model takes into account a number of factors: 
1. Erosive power of rain 
2. Interception by plant cover 
3. Topography (slope, etc.) 
4. Resistance of soil to erosion. 
Given these as required inputs to the model a 
fine mesh database was required. Remote sensing 
and spatial information systems provided the key 
to creating a database from satellite digital data 
and thematic cartographic data.