APPLICATION OF GIS TO EVALUATE LONG-TERM VARIATION OF
SEDIMENT DISCHARGE TO COASTAL ENVIRONMENT
LE TRUNG TUAN
Vietnam Institute for Water Resources Research
171 Tay Son Str., Dong Da, Hanoi, Vietnam
vnwp@hn.vnn.vn
TOMOYA SHIBAYAMA
Department of Civil Engineering, Yokohama National University
79-5 Tokiwadai, Hodogaya-ku, Yokohama 240-8501, Japan
tomo@ynu.ac.jp
This paper presents a GIS-based method to calculate the total sediment discharge
from river basins to coastal areas. This method uses Revised Universal Soil Loss
Equation (RUSLE) to calculate the rate of soil erosion and Gross Erosion-Sediment
Delivery method (GESD) to calculate the total sediment discharge in a GIS modeling
environment. The model is tested using the data of Abe River then applied to four
large river basins in Asia. Global data sets are used as the input to the current model.
The result shows that there are significant variations of sediment discharges due to the
precipitation change in these river basins.
Keywords: sediment discharge, soil erosion, digital map, river basin, GIS, RUSLE,
GESD
1
Introduction
During last few decades, a number of numerical models have been developed to
simulate sediment transport process. A common aspect of these models is that they
need the sediment boundary condition to exactly simulate the process. Since the major
source of sediment transported to coastal zone is rivers draining continents (Milliman
and Meade, 1983), the boundary conditions of these models are often calculated by
using observed river sediment discharge at or near the river mouths.
Although observed river sediment discharge records can provide an excellent
source of data for model simulation, there are some limitations in using these data.
Firstly, sediment discharge measurement is an expensive task because it needs the
construction of observation station at the measurement locations. In many rivers,
especially ones in developing countries, there is often no data available due to the lack
of observation stations near the river mouths. Secondly, the observed data is not
suitable for predicting the long-term sediment transport because they do not reflect the
possible variation of river sediment discharge in the future. Therefore, it is necessary
to establish a method to estimate the sediment delivery from rivers to coastal zones,
which is capable of predicting future variation.
Predicting the sediment discharge of river requires the knowledge of soil
erosion and sedimentation throughout river basins. A number of factors such as
drainage area size, basin slope, climate, land use, land cover, etc. may affect sediment
delivery processes. Accurate prediction of soil erosion over basins and sediment
delivery ratio is an important and effective approach to predict sediment yield. The
classical approach of hydromechanics has not yet succeeded in modeling the complete
processes of erosion and sediment transport in rivers. The reason is that the properties
of the particles are characteristically all random, and the properties of the riverbed are
extremely irregular. Consequently, theoretical models based on the laws of mechanics
are not satisfactory in representing the motion of a single particle, let alone the
immense amounts of sediment transport.
A common method of estimating total sediment discharge in the absence of in
situ hydraulic measurements is the sediment rating curve. Sediment rating curves are
empirical relations between the total fluid discharge and the sediment discharge in the
form of S = aQb, where S is the suspended sediment concentration and Q is the
discharge (Campbell and Bauder, 1940). This relationship enables us to estimate
roughly the mean monthly and annual sediment rates, however, the result is not
adequate. Stevens (1985) developed MODEIN - a procedure that computes total
sediment discharge at a cross section of a stream from measured hydraulic variables,
the concentration and particle-size distribution of the measured suspended sediment,
and the particle-size distribution of the bed material. The computation involves the
extrapolation of the measured suspended-sediment discharge to represent the total
suspended- sediment discharge and the addition of a computed bedload discharge.
The procedure is applicable only if measured data are available so it cannot be used
directly for design or predictive purposes. It is intended to be used only at sites where
all of the bed material is finer than about 16 millimeters, and it can be used only if a
significant part of the measured suspended sediment is composed of particles of the
same size as particles in the bed material.
Young et al. (1989) developed the AGricultural NonPoint Source (AGNPS)
model that simulates runoff, sediment, and nutrient transport from agricultural
watersheds. The model has separate hydrological erosion, sediment transport, and
chemical transport modules which route water, sediments, and other contaminants
through cells from the catchment boundary to the outlet in a stepwise fashion. In the
model, sediment transport is estimated by equations proposed by Foster et al. (1981)
and Lane (1982) respectively. Although the model is more accurate, it can be applied
to small watersheds ranging from a few hectares to approximately 20,000 hectares
only.
Some previous works listed above are not applicable for prediction of longterm variations of sediment discharge. The objective of this study is to establish a
simplified method to predict the long-term sediment discharge from large river basins
to coastal environment by using Revised Universal Soil Loss Equation (RUSLE) and
Gross Erosion-Sediment Delivery method (GESD) in combination with GIS.
2
Target Areas and Input Data Sets
Four large river basins in Asia including Mekong River, Red River, Yangtze River
and Irrawaddy River are selected in this study to demonstrate the applicability of the
model. The Abe River in Japan is also selected to test the model. The following
sections describe the data used. Figures 1 and 2 show the sample data maps of global
average precipitation in September and the sediment yields of major rivers in the
world, respectively.
2.1
ETOPO-5
The ETOPO-5 data set is original from the U. S. National Geophysical Data Center. It
has elevation readings sampled from every five-minute latitude/longitude crossing on
the global grid and a one-meter contour interval. It includes bathymetry from
approximately 10,000 meters below sea level, and extends up to heights of approx.
8,000 meters above sea level. The original NGDC data file was reformatted by
UNEP’s Global Resource Information Database (GRID) to place the origin at 180
degrees West instead of 0 degrees Greenwich Meridian.
2.2
Digital soil map of the world
The Food and Agriculture Organization (FAO) Soil Map of the World includes
estimated 1,650 different mapping units, consisting of soil units or associations, which
occur within the limits of a mappable physiographic unit. When a given map unit is
non-homogeneous, it is composed of dominant and associated soils and inclusions
(which cover respectively at least 20%, and less than 20% of the unit). The number of
soil type classes, which compose the FAO Soil Map legend, is 106; and these are
often grouped into 26 major categories. A total of 12 soil phases, three texture classes,
three slope classes and so-called "miscellaneous (e.g., non-soil) land units" are also
recognized in this digital version of the FAO-UNESCO Soil Map.
2.3
Landuse map
The Matthews Vegetation, Cultivation Intensity and Albedo data set (land use map) is
used in this study. The data set comes from a global map of vegetation types, which
was compiled from up to 100 existing map sources at the Goddard Institute of Space
Studies (GISS), Columbia University, in New York.
It shows the predominant
vegetation type (one out of 32 classes) within each one degree-square
latitude/longitude grid cell. Matthews Cultivation Intensity data set is based on
existing maps of vegetation and satellite imagery, and it shows the percentage of each
one-degree square latitude/longitude grid cell that is under cultivation, versus the
percentage of natural vegetation, including five classes. The data have a spatial
resolution of one degree latitude/longitude, and one byte/eight bits per pixel.
2.4
Precipitation data set
The IIASA Climate Database was created at the International Institute for Applied
System Analyses (IIASA; Laxenburg, Austria) by Rik Leemans and Wolfgang P.
Cramer to represent current global climate. The GRID versions of this data set include
12 monthly average precipitation data files in two-byte (16-bit) image format, to
accommodate values above 255, and 12 one-byte (eight-bit) data files with
precipitation values compressed into a 0-255 range. Both sets of data files have image
arrays of 360 rows (lines, records) by 720 columns (elements, pixels, or samples)
covering the entire globe. Non-land areas (the oceans) have a data value of 9999.
Future precipitation scenarios of two different time periods (2010-2039 and
2040-2069) generated by HadCM2 model – a coupled atmosphere-ocean general
circulation model - is used as the input to the current model. HadCM2 simulation
results include data files of different climate variables such as temperature, wind
speed, cloud cover, precipitation, vapor pressure, etc. Concerning the sediment
discharge computation, only precipitation data is needed. Each precipitation data files
have 7,008 data (96x73 grids) for each month. Table 1 lists the calculated average
precipitation of the selected river basins in different time periods.
(Table 1)
2.5
Observed sediment yield of major rivers in the world
This data represent the long-term sediment observation of major river basins in the
world and was summarized by Milliman and Meade (1983). Figure 2 shows the
observed average sediment of the basins, the map also presents total observed
sediment delivery to Oceans.
2.6
River basin boundary
The river basin boundary map with a resolution of 1 degree by 1 degree
longitude/latitude grids was produced by Oki (1998) as a part of the project called
"Total Runoff Integrating Pathways (TRIP)”. The data are organized into two files,
one is the boundary coordinate file and the other is the index file. This index file
indicates the river basin numbers and corresponding river names, and also the
longitude and the latitude of the river mouth.
(Figures 1 + Figure 2)
3
Methodology
3.1
Soil erosion model
The Revised Universal Soil Loss Equation (RUSLE) is used as a soil erosion model to
estimate the physical amount of soil loss in river basins. RUSLE is a simple
multiplicative model that was derived from over 10,000 plot-years of data. In RUSLE,
the amount of soil loss is a product of six coefficients representing the nature of
rainfall and basin characteristics as follows,
A=RKLSCP
(1)
where A = computed spatial soil loss per unit area [ton km-2 yr-1]; R = rainfall-runoff
erosivity factor [kJ mm km-1 hr-1]– the rainfall erosion index plus a factor for any
significant runoff from snowmelt; K = soil erodibility factor [(ton km2 hr)(km2 kJ
mm)-1] – the soil-loss rate per erosion index unit for a specified soil as measured on a
standard plot, which is defined as a 22.1 meters length of uniform 9% slope in
continuous clean-tilled fallow; L = slope length factor [unitless] – the ratio of soil loss
from the field slope length to soil loss from a 22.1 meters length under identical
conditions; S = slope steepness factor [unitless] - the ratio of soil loss from the field
slope gradient to soil loss from a 9% slope under otherwise identical conditions; C =
cover-management factor [unitless] – the ratio of soil loss from an area with specified
cover and management to soil loss from an identical area in tilled continuous fallow;
P = supporting practices factor [unitless] – the ratio of soil loss with a support practice
like contouring, strip-cropping, or terracing to soil loss with straight-row farming up
and down the slope.
3.1.1 Rainfall-runoff erosivity factor (R)
The energy of a particular storm depends upon the intensities at which the rain occur
and the amount of precipitation that is associated with each intensity value. Within the
RUSLE rainfall erosivity is estimated using EI30, which is a product of total rainfall
energy (E) and maximum rainfall intensity (I30) (Renard et al., 1997). The rainfallrunoff factor is the average annual total of all computed EI30 values of the storms for
one-year period. The storm energy indicates the volume of rainfall and runoff, but a
long, slow rain may have the same E value as a shorter rain at much higher intensity.
Raindrop erosion increases with intensity. The I30 component accounts the prolonged
peak rates of detachment and runoff. The product term EI is a statistical integration
term that reflects how total energy and peak intensity are combined in a given storm.
Technically, the term indicates how particle detachment is combined with transport
capacity (Renard et al., 1997). In the calculation, only storms with the amount of
rainfall more than 12.5 mm are considered, and a storm period with total rainfall less
than 1.25 mm is used to divide a longer storm period into two storms (Renard et al.,
1997). The rainfall energy of a given storm is calculated as
E = 0.29 [1-0.72 exp(-0.05 im)] (Brown & Foster,1987)
(2)
where E is rainfall energy [MJ.ha-1.mm-1] and im is rainfall intensity [mm.hr-1].
The rainfall-runoff factor is then calculated by using the formula
j
R
 (EI30 )i
i 1
N
(3)
where (EI30)i = EI30 for storm i, and j = number of storms in an N year period.
If detail rainfall data are not available, the method proposed by Arnoldus
(1980) is employed. In this method, the value of R factor is estimated with only
monthly and yearly rainfall data by the following equation
R  (4.17  MFI )  152
(4)
where MFI stands for Modified Fourier Index and is calculated by
pi2
MFI  
i 1 P
12
(5)
where pi = monthly rainfall [mm] and P = yearly rainfall [mm].
3.1.2 Soil erodibility factor (K)
The soil erodibility factor K represents the influence of soil properties on soil loss
during storm events on upland areas and is defined as the rate of soil loss per rainfall
erosion index unit as measured on a unit plot. The unit plot is 22.1 m long, has a 9%
slope, and is continuously in a clean-tilled fallow condition with tillage performed
upslope and downslope (Wischmeier and Smith, 1978; Renard et al., 1997).
It
denotes the average long-term soil and soil profile response to the erosive power
associated with rainfall and runoff. That means this factor is a lumped parameter that
represents an integrated average annual value of the total soil and soil profile reaction
to a large number of erosion and hydrological processes. These processes consist of
soil detachment and transport by raindrop impact and surface flow, localized
deposition due to topography, and rainwater infiltration into the soil profile (Renard et
al., 1997).
RUSLE model utilizes the technique proposed by Wischmeier at at. (1971) to
calculate the K factor value of a soil. This method involves grouping many
measurable soil properties into five parameters that are most closely correlated with
soil erodibility. The five parameters include: percentage of modified silt, percentage
of modified sand, percentage of organic matter (OM), class for structures (s), and
permeability (p) (Renard et al., 1997). Determination of K factor includes the
assignment of values that corresponded to the soil texture (Corbittt, 1990) and age of
the soil (Kappas, 1996) of each location. The FAO soil map of the world provides a
classification scheme which indicate the presence of a primary soil type, secondary
soil type, and in some cases, tertiary category within a soil unit. Since each soil type
must have a single K value associated with it, a method of allocating a weight to each
of the soil categories within a unit is utilized to calculate the final K value of that
location. The weighting value of a soil category is proportional to the percentage of
area of that category.
3.1.3 Slope length (L) and slope steepness (S) factors
The effect of topography on erosion in RUSLE is accounted for by the LS factors.
Erosion increase as slope length increases, and is considered by the slope length factor
(L). Slope length is defined as the horizontal distance from the origin of overland flow
to the point where either the slope gradient decreases enough that deposition begins,
or runoff becomes concentrated in a defined channel (Wischmeier and Smith, 1978;
Renard et al., 1997). The slope steepness factor S reflects the influence of slope
gradient on erosion. Slope is often estimated in the field by use of an inclinometer or
other devices, or can be estimated from elevation map. Both slope length and
steepness substantially affect sheet and rill erosion estimated by RUSLE. The effects
of these two factors have been evaluated separately in research using uniform-gradient
plots. However, in erosion prediction, the factors L and S are usually evaluated
together by following formula
  
2
LS  
 65.41 sin   4.56 sin   0.065
22
.
13


m
(6)
where  is slope length in meters and  is the slope angle.
3.1.4 Cover management factor (C) and supporting practices factor (P)
The C factor is used within RUSLE to reflect the effect of cropping and management
practice on erosion rate. As with most other factors within RUSLE, the C factor is
based on the concept of deviation from a standard, in this case an area under cleantilled continuous-fallow conditions. The soil loss ratio is then an estimate of an ratio
of soil loss under actual conditions to losses experienced under the reference
conditions. Past works indicated that the general impact of cropping and management
on the soil loss can be divided into a series of subfactors. In this approach the
important parameters are the impacts of previous cropping and management, the
prediction offered the soil surface by the vegetative canopy, the reduction in erosion
due to surface cover and surface roughness, and in some cases, the impact of low soil
moisture on reduction of runoff from low-intensity rainfall (Renard et al., 1997). The
C factor depends on the type of land cover and its value ranges from 0.001 for ever
green forest to 1.0 for drainage/water and buildup area (McKendry et al., 1992). The
large value for buildup area reflects large percentage of smooth paved surface. This
means that once eroded soil would be available in buildup area because of, for
example, soil inflow from rural area with a flood event, the soil could be drained
quickly from the area (Kurata et al., 1999).
By definition, the support practice factor P in RUSLE is the ratio of soil loss
with a specific support practice to the corresponding loss with upslope and downslope
tillage. These practices principally affect erosion by modifying the flow pattern,
grade, or direction of surface runoff and by reducing the amount and rate of runoff.
For cultivated land, the support practices considered include contouring, stripcropping, terracing, and subsurface drainage. On dryland or rangeland areas, soildisturbing practices oriented on or near the contour that result in storage of moisture
and reduction of runoff are also used as support practices (Renard et al., 1997). P
stands for erosion inhibition effect and reflects partly human’s effort not to allow soil
erosion. For example, the relatively small value of 0.5 is estimated in paddy field, and
this is due to maintenance of rice paddy field, e.g. the construction of footpath
between rice paddies, so that the soil in the field will not be able to be transported
away (Kurata et al., 1999).
C and P are depended on the kind of land cover and therefore are estimated
from the landuse map. Table 2 shows the values of C and P for different land covers.
(Table 2)
3.2
River sediment yield estimation model
Sediment yield is the amount of sediment passing a particular channel location and is
influenced by a number of geomorphic processes. Sediment yield is usually less than
the amount of soils actually eroded in the river basin. It is normally expressed as the
total sediment volume delivered to a specified location in the basin, typically the river
mouth, divided by the effective drainage area above that location for a specified
period of time. Yield typically has the units of cubic meters per square kilometer per
year or metric tons per year. In some cases, it is also necessary to estimate yield from
a river basin from individual storm events of specified frequency. Individual event
yields are measured as metric tons or cubic meters per event. In some basins, single
event sediment yields often exceed average annual values by several orders of
magnitude.
Spatial and temporal variations in physical and biological characteristics of the
river basins make estimation of sediment yield a difficult and imprecise task.
Important variables include soils and geology, relief, climate, vegetation, soil
moisture, precipitation, drainage density channel morphology, and human influences.
Dominant processes within a river basin may be entirely different between
physiographic or ecological regions, and may change with time. The problem
becomes even more complex when grain size distributions and sediment yield for
particular events must be estimated for input to sedimentation transport simulation
models. There is no widely accepted procedure for computing basin sediment yield
and grain size distribution directly from basin characteristics without measured
information.
Sediment transport is influenced primarily by the action of wind and water,
and deposition occurs in a number of locations where energy for transport becomes
insufficient to carry eroded sediments. Colluvial deposits, floodplain, and valley
deposits, channel aggradation, lateral channel accretion, and lake and reservoir
deposits are examples of typical geomorphic deposition processes. The stability and
longevity of sediment deposits vary. Lake and reservoir deposits tend to be long-term,
whereas some channel and floodplain deposits may be remobilized by the next largescale flood event, only to be deposited downstream. The spatial and temporal
variability of sediment production, transport and deposition greatly complicates the
task of estimating sediment yield from a watershed.
The RUSLE can be used to compute average erosion in the various parts of a
watershed, but deposition and channel-type erosion must be estimated by other means.
The soil transport from grid cells to river mouth is often computed by the soil
transport models. However, these models are very complicated and require the detail
information about land surface condition, river data (cross section, longitudinal
profile), hydrological data, etc., therefore the applicability of these models is very
limited.
In this study, river sediment yield is estimated by analyzing the relationship
between the measured sediment data and the total soil loss of the river basin obtained
from RUSLE. The Gross Erosion-Sediment Delivery Method (GESD) (Neibling and
Foster, 1977) is used. In GESD the sediment yield relates with the gross soil erosion
by the following formula:
Y = E(SDR)/W
(7)
where Y is sediment yield [ton/km2/yr], E is the gross erosion [ton/yr] computed by
summation of annual soil loss of all cells estimated by RUSLE; SDR is sediment
delivery ratio, depends on the basin area and basin characteristics and was estimated
by analyzing the measured sediment yield and the RUSLE result; W is basin area
[km2].
3.3
Procedures
3.3.1 Data processing procedures
Digital data obtained from different sources have different types of data formats. In
order to use these data together, each data layer must be transformed into a common
data format. The standard format used in this study is one-degree lat/long grid raster
format. In basic data processing, those raster format data with grid sizes other than
one-degree must be resampled to obtain the desired one-degree lat/long standard
format data. Vector format data are transferred directly into the standard format with
an appropriate scale.
The original global data sets are first converted into ASCII raster file format.
In order to get the data of a particular river basin from global data sets map overlay is
used. In map overlay, if a cell is located outside the basin it will be assign no-data
value (-9999). The obtained basin data are then converted into ArcView Grid and
stored in ArcView as a theme. Theme is a map layer in ArcView containing both
spatial and attribute data (the latter are in database tables). A theme file contains
graphic information required to draw a set of geographic features together with
information about those features. Figure 3 describes the basic data processing
procedures.
(Figure 3)
3.3.2 Calculation procedures
The RUSLE and GESD models are implemented in the ArcView GIS environment.
ArcView GIS and its extensions are used as an modeling tool which facilitated the
data processing, model parameters computation, input/output presentation, and result
analysis. Modeling is done using the ArcView built-in Avenue - a macro
programming language, and Fortran subroutines built into Dynamic Link Library
(DLL) files. In the calculation, different input data sets are used to generate data
layers of RUSLE factors in ArcView. The data layer of soil loss rate is generated by
the multiplication of these factor data layers. The gross erosion is taken as the total
soil loss of all the cells. SDR is calculated from the estimated gross erosion and
observed sediment discharge. The computed SDR value is considered constant and is
used to predict future sediment discharge. The general modeling framework is shown
in Figure 4.
(Figure 4)
4
Results and Discussion
4.1
Model examination
This section describes the application of the model to calculate total sediment
discharge of Abe River basin in Japan. The reason for selection of this river is that it
has long records of rainfall and sediment data as well as high-resolution digital maps
of elevation, landuse, and soil, etc. The objective of the calculation is to test the model
and explore its applicability by comparing the calculated sediment discharge and
observed values. For this river basin, a different data set, which is more detail than
global data sets is used.
RUSLE’s factors and soil loss distribution maps of Abe River in 1995 are
given in Figure 5. The soil loss rate varies widely. The estimated soil loss rate may be
as small as zero in the rocky area or it may reach up to thousands of ton per square
kilometer per year. Although soil loss depended on the soil, rainfall, land cover and
the slope of the land surface, it is clear from the result that rainfall has the biggest
impact on soil erosion.
(Figure 5 + Figure 6)
Figure 6 shows the calculated and measured sediment discharges. The result
shows that, in general, the agreement of calculated sediment discharge and the
observed one is obtained. However, there are two exceptions in 1974 and 1982 where
the observed sediment discharges are much greater than the calculated values. In
1974, the observed sediment discharge is 520x103 m3/year and is nearly 4 times as
much as the calculated value (132.84x103 m3/year). Similarly, the observed sediment
discharge in 1982 is nearly 3 times as much as the calculated one. This happened
because during these two years, there were landslides in the area and this caused a
large amount of rock and sand to be transported to the river mouth. From the result, it
may be stated that the model is capable of estimating the total sediment discharge
from a river basin to coastal area in normal years where the effect of landslides is
relatively small. If big landslides occur, a separate module that calculates the amount
of sediment contributed by landslide is needed.
4.2
Model application
The model is applied to forecast the change of sediment discharge of four large river
basins in Asia including Mekong River, Red River, Yangtze River and Irrawaddy
River due to global climate change. Future precipitation scenarios of two different
time periods (2010-2039 and 2040-2069) generated by HadCM2 model is used as the
input to the current model. The result of future sediment discharge is then compared
with the current values.
Figures 7 and 8 show the soil loss rate distribution maps of the river basins in
different time periods. The results show that there is big spatial variation of soil loss
rate within a river basin. For example, the soil loss rate in the Irrawaddy River varies
from as low as 800 to as high as 2,600 tons per square kilometer per year. This
variation is mainly due to the difference in soil property, rainfall, land cover and
slope. If a cell is located in the area where the soil is easily eroded and the rainfall is
high, the soil loss rate is extremely high. In other hand, if a cell is located in the area
where soil is difficult to be eroded or the soil is well protected by human activities, the
soil loss rate is very low. Rocky areas often have negligible or zero soil loss rate.
(Figure 7 + Figure 8)
Average soil loss rates of Red River and Mekong River are bigger than the
rates of Yangtze River and Irrawaddy River. The result suggests that there is no direct
correlation between average soil loss rate and SDR. However, it is observed from the
result that in large river basins such as Mekong River and Yangtze River sediment
delivery ratio is lower than in smaller river basin. In large river basins, sediment
particles must travel a very long distance before they reach the sea. In the process of
sediment transport, a certain amount of sediment particles is deposited along travel
path. As the travel distance becomes longer, the amount of deposited sediment
becomes larger leading to less sediment discharge to coastal area.
Figure 9 shows the calculated total sediment discharge of the river basins in
different time periods and Table 3 compares the future sediment discharges with the
current values.
(Figure 9 + Table 3)
The result shows that precipitation has a direct impact on river sediment
discharge to the ocean. As the precipitation increases, the sediment discharge
increases accordingly and vice versa. The variations of total sediment discharges from
Red River and Mekong River are significant. Sediment discharge values decrease by
5.06~7.97% in 2010-2039 period and 16.9~17.53% in 2040-2069 period respectively.
The total sediment discharge from Yangtze River increases by 5.51% and 9.4% in
2010-2039 and 2040-2069 time periods respectively. In other hand, the calculated
result of Irrawaddy River shows a decrease in total sediment discharge by
0.28~4.68%.
5
Conclusions
This paper proposes a method to calculate sediment discharge from rivers to coastal
areas using RUSLE in combination with GESD in the ArcView GIS modeling
environment. The model was tested using a Japanese river basin. Results showed that
the estimated sediment discharge values is agreed with the observed ones in normal
condition. During the periods when an episodic event such as big landslide occurs, the
model may underestimates the sediment discharge.
The application of the model in four river basins in Asia suggests that the
model is capable of estimating the sediment discharge from basin’s digital data such
as elevation, soil, land use, rainfall, etc. The model can be applied to predict future
sediment discharge variation due to the impacts of global climate or land use changes.
There is, however, a need for further improvement of the model to take into account
the amount of sediment created by other causes such as landslide. It is also noted that
the accuracy of the result is largely dependent on the resolution of digital data
especially when small-scale or medium-scale river basins are to be considered.
References
Beasley, D.B. et al. (1982) Modeling sediment yields for agricultural watersheds.
Jour. of Soil and Water Conservation 37(2): 113-117.
Bennet, J.P. (1974) Concepts of Mathematical Modeling of Sediment Yield. Water
Resources Research 10: 485-496.
Busacca, A.J. et al. (1993) Comparing landscape scale estimation of soil erosion in the
Palouse using Cs-137 and RUSLE. Jour. of Soil and Water Conservation 48:
361-367.
Campbell, F. B. and Bauder, H. A.(1940) A rating-curve method for determining siltdischarge of streams, EOS (Trans. Am. Geophys. Union) 21: 603-607.
Corbitt, R.A. (1990) Standard Handbook of Environmental Engineering. McGrawHill.
Cox, M.D. (1984) A primitive equation, three-dimensional model of the ocean. GFDL
Ocean Group Technical Report No. 1, GFDL, Princeton, 143pp.
Desmet, P.J.J. and Govers, G. (1995) GIS-based simulation of erosion and deposition
patterns in an agricultural landscape: a comparison of model results with soil map
information. Catena 25: 389-401.
Endo, T. and Shibasaki, R. (1996) An analysis on soil erosion factors at the global
scale. Jour. of Global Environment Eng 2: 9-21.
Foster, G.R. et al. (1981) Estimating erosion and sediment yield on field-sized areas.
Trans. of the American Society of Agricultural Engineers 24(5): 1253-1262.
Foster, G.R., and Wischmeier, W.H. (1974) Evaluating irregular slopes for soil loss
prediction. Trans. of the American Society of Agricultural Engineers 17(2): 305309.
Goodchild, M.F., et al eds. (1996) GIS and Environmental Modeling: Progress and
Research Issues. Fort Collins: GIS World, Inc.
Grunwald, S. and Frede, H.G. (1999) Using AGNPS in German watersheds. Catena
37(3-4): 319-328.
Grunwald, S. and Norton, L.D. (1999) An AGNPS-based runoff and sediment yield
model for two small watersheds in Germany. Trans. of the ASAE 42(6): 17231731.
Hairsine, P.B., and Rose, C.W. (1991) Rainfall detachment and deposition: Sediment
transport in the absence of flow-driven processes. Soil Science Society of America
Jour. 55(2): 320-324.
Holmgren, P. (1994) Multiple flow direction algorithms for runoff modelling in gridbased elevation models: An empirical evaluation. Hydrologic Processes 8: 327334.
Johns,T.C. et al. (1997) The second Hadley Centre coupled ocean-atmosphere GCM:
model description, spinup and validation. Climate Dynamics 13: 103-134.
Kurata, G. et al. (1999) Estimation of soil erosion and sediment transport in the
Mekong River basin using global mapping data. Jour. of Global Environment
Eng. 5: 111-125.
Lane, L.J. (1982) Development of a procedure to estimate runoff and sediment
transport in ephemeral streams. pp. 275-282 in Walling, D.E., (ed)., Recent
Developments in the Explanation and Prediction of Erosion and Sediment Yield.
Wallingford: International Association of Hydrological Sciences Publication No.
137.
Milliman J.D. et al. (1983) World-wide delivery of river sediment to Oceans. The
Jour. of Geology 91: 1-21.
Mitasova, H. et al. (1996) Modeling topographic potential for erosion and deposition
using GIS. Int. Jour. of Geographical Information Science 10(5): 629-641.
Moore, I.D., and Wilson, J.P. (1992) Length-slope factors for the Revised Universal
Soil Loss Equation: Simplified method of estimation. Jour. of Soil and Water
Conservation 47(5): 423-428.
Murphy, J.M. (1995) Transient response of the Hadley Centre coupled oceanatmosphere model to increasing carbon dioxide. Part I, control climate and flux
adjustment. Jour. of Climate 8: 36-56.
Neibling, W.H. and Foster, G.R (1977) Estimating deposition and sediment yield
from overland flow processes. 1977 International Symposium on Urban
Hydrology, Hydraulics, and Sediment Control Procs, Univ of Kentucky,
Lexington.
Renard, K.G. et al. (1991) RUSLE: Revised Universal Soil Loss Equation. Jour. of
Soil and Water Conservation 46(1): 30-33.
Renard, K.G. et al. (1997) Predicting soil erosion by water: A guide to conservation
planning with the Revised Universal Soil Loss Equation (RUSLE). US
Department of Agriculture, Agriculture Handbook No.703, 404 pp.
Srinivasan, R., and Arnold, J.G. (1994) Integration of a basin scale water quality
model with GIS. Water Resources Bulletin 30(3): 453-462.
Stevens, H.H. (1985) Computer program for the computation of total sediment
discharge by the modified Einstein procedure. U.S. Geological Survey WaterResources Investigations Report 85-4047, 77 pp.
Wilson, J.P. (1996) Spatial models of land use systems and soil erosion: The role of
GIS. In Wegener, M., and Fotheringham, A.S. (eds) GIS and Spatial Models:
New Potential for New Models?, London: Taylor and Francis.
Wischmeier, W.H. (1976) Use and misuse of the universal soil loss equation. Jour. of
the Soil and Water Conservation 31(1): 5-9.
Young, R.A. et al. (1989) AGNPS: A nonpoint source pollution model for evaluating
agricultural watersheds. Jour. of the Soil and Water Conservation 44(2): 168-173.
Zhang, X. et al. (1997) Global scale overland flow and soil erosion modelling using
remote sensing and GIS techniques: model implementation and scaling.
Proceedings of RSS ‘97 Remote Sensing in Action, pp. 379–384.
Table 1: Basin’s average precipitation (mm/year)
Period
Mekong
Yangtze
Red
Irrawaddy
1961-1990
2010-2039
1358.75
1312.48
1027.63
1041.42
1264.00
1226.61
1818.32
1807.67
2040-2069
1102.78
1126.71
1179.54
1781.94
Table 2: C and P factor for different land covers (McKendry et al., 1992)
Land cover
C factor
P factor
Drainage/Water
1.00
1.0
Buildup area
1.00
1.0
Barren area
0.28
1.0
Forest
0.10
1.0
Agricultural area (crop field)
0.65
0.5
Paddy field
0.10
0.5
Grassland/Shrub
0.15
0.5
Wetland
0.56
1.0
Mixture
0.40
0.5
Table 3: Prediction result of sediment discharge change
of the selected river basins in Asia
River
Area
Total annual sediment discharge
Change
Basin
(103 km2)
(106 ton/year)
(%)
1961-1990
2010-2039
2040-2069
2010-2039
2040-2069
Mekong
790
160
147.24
131.95
-7.97
-17.53
Red
120
130
123.42
108.03
-5.06
-16.90
1940
502
529.68
549.20
+5.51
+9.40
430
265
264.30
252.59
-0.26
-4.68
Yangtze
Irrawaddy
PRECIPITATION - SEP
Figure 1: Global average precipitation in September (Leenmans and Wolfgang)
100
16
13
12 3 14 6
52
Yield
(T/km2/yr)
20
8
4
10
210
6
7
100
220
210
900
6
92
40
43
17
81
108
4780
130
96
265 160
20
33
30
Figure 2: Sediment yields (ton/km2/year) and measured total sediment
discharge (106 ton/year) of major rivers in the world (Milliman et al., 1983)
Global Data Sets Acquisition
FORTRAN
Programming
ASCII Raster Format
Data Files
ArcView Data
Import
ArcView’s GRIG Format
Global Data Files
Figure 3: Basic data processing procedures
DIGITAL
MAPS
BASIN
FACTORS
RUSLE
GESD
Elevation
Rainfall & Runoff
(R)
Landuse
Observed Sediment
Soil Erodibility (K)
Soil
GIS
Rainfall
Distribution
Basin
Boundary
Land Cover
Management (C)
Annual
Soil Loss
Support Practice
(P)
Topographic (LS)
Others
Figure 4: General modeling framework
Sediment
Delivery Ratio
Total Sediment
Discharge
(a)
(d)
(c)
(b)
(e)
(f)
Figure 5: RUSLE’s factors and soil loss distribution maps of Abe river: (a) K factor;
(b) LS factor; (c) C factor; (d) P factor; (e) R factor (1995); (f) Soil loss rate (1995)
550
450
Calculated values
400
Observed values
350
300
250
200
150
100
50
00
98
96
94
92
90
88
86
84
82
80
78
76
74
72
70
68
0
66
Sediment discharge (10^3 m3/y)
500
Year
Figure 6: Comparison of calculated and observed total sediment discharge - Abe
River
2236
0 (ton/km2/yr)
(a)
1961-1990
2010-2039
2040-2069
Soil Loss Rate
(ton/km2/yr)
(b)
1961-1990
2010-2039
2040-2069
Soil Loss Rate
(ton/km2/yr)
(c)
1961-1990
2010-2039
2040-2069
Figure 7: Soil loss rate distribution maps of the river basins in different time periods:
(a) Mekong Riber; (b) Irrawaddy River; (c) Red River
(not to scale)
1961-1990
2010-2039
2040-2069
Figure 8: Soil loss rate distribution maps of the river basins in different time periods –
Yangtze River
600
Total sediment discharge
(mill.tons/year)
500
1961-1990
2010-2039
400
2040-2069
300
200
100
0
Mekong
Red
Yangtze
Irrawaddy
Figure 9: Basin total sediment discharge in different time periods