International Journal of Sediment Research 23 (2008) 167-173

Assessment of soil erosion and sediment delivery ratio using remote sensing
and GIS: a case study of upstream Chaobaihe River catchment, north China
Weifeng ZHOU1 and Bingfang WU2

Abstract
Soil erosion in catchment areas reduces soil productivity and causes a loss of reservoir capacity.
Several parametric models have been developed to predict soil erosion at drainage basins, hill slopes
and field levels. The well-known Universal Soil Loss Equation (USLE) represents a standardized
approach. Miyun reservoir, which sits on Chaobaihe River, is the main surface source of drinking
water for Beijing, the capital of China. Water and soil loss are the main reasons for sediment to enter a
reservoir. Sediment yield is assessed using a version of the universal soil loss equation modified by
Chinese researchers. All year 2001 and 2002 data for factors in the equation are obtained from remote
sensing or collected to form an analysis database. These factors are computed and mapped using
Geographic Information System tools. Based on the complex database, the modified model is
developed. Through pixel-based computing the sediment yield per hydrological unit is calculated. The
model does not consider sediment deposition occurring on hillslopes. Gross soil loss is often higher
than the sum of those measured at catchment outlets. The sediment delivery ratio (SDR) per
hydrological unit is also computed. This study analyzes the main contributions of sediment yields on
sub-basins of the Chaobaihe River to the Miyun Reservoir, and discusses the possible reasons for the
difference between SDRs in 2001 and 2002 at different outlets. The result shows that in the upper
basin of the Miyun Reservoir, in 2001 the area of erosion that could be neglected was 8,202.76 km 2,
the area of low erosion 3,269.59 km2, the area of moderate erosion 3,400.97 km2, the area of high
erosion 436.89 km2, the area of strong erosion 52.19 km 2 and the area of severe erosion 3.13 km 2. The
highest soil loss was 70,353 t/km 2.yr in Fengning County in 2001, followed by 64,418 t/km 2.yr by
Chicheng County in 2001. The SDR in 2002 was lower than that in 2001. The main reasons are the
decreasing rainfall erosivity and total runoff.

Key Words: Modified USLE, Gross erosion, Sediment delivery ratio

1 Introduction
Soil erosion in catchment areas and the subsequent deposition in rivers, lakes and reservoirs are of great
concern for two reasons. Firstly, rich fertile soil is eroded from the catchment areas. Secondly, there is a
reduction in reservoir capacity as well as degradation of downstream water quality (European
Environment Agency, 1995). Although sedimentation occurs naturally, it is exacerbated by poor land use
and land management practices adopted in the upland areas of watersheds. Uncontrolled deforestation due
to forest fires, grazing, incorrect methods of tillage and unscientific agricultural practices are some of the
poor land management practices that accelerate soil erosion, resulting in large increases in sediment
inflow into streams (David Pimentel, 1998).
The Miyun Reservoir, which sits on Chaobaihe River, is the main surface source of drinking water for
1
Doctor candidate, Graduate School of the Chinese Academy of Sciences(CAS), Beijing 100039, China,
E-mail: zhou_wf@hotmail.com
2
Prof., Institute of Remote Sensing Applications, Chinese Academy of Sciences(CAS), Beijing 100101, China,
corresponding author, E-mail: wubf@irsa.ac.cn
Note: The original manuscript of this paper was received in March 2005. The revised version was received in
 Dec. 2007. Discussion open until June 2009.
International Journal of Sediment Research, Vol. 23, No. 2, 2008, pp. 167–173 - 167 -
Beijing, the capital of China. Water and soil loss is the main reason for sediment entering the reservoir and
this process potentially reduces water quality. Soil erosion in this area strongly influences the capital’s
ecological health. The study site for this work is located 115 o23’-117o35’E and 40o19’ -41o38’N, and
covers an area of approximately 15,378 km 2 (Fig. 1). In Fig. 1, the area cinctured by the yellow line is the
Upper Basin of the Miyun Reservoir, and is the study area for this work; the area enclosed by the bright
line is the catchment for Beijing. Conditions in the study site affect the water quality of 12 counties.
This paper assesses soil loss by a model-based approach. Use of the well-known Universal Soil Loss
Equation (USLE) is a standardized approach. To develop monitoring of soil losses in the upstream
Chaobaihe River Catchment, a modified version of the USLE is used. Data for the equation from years
2001 and 2002 are obtained from remote sensing or collected to form an analysis database. Factors that
will be used in the analysis are computed and mapped using Geographic Information System (GIS) tools.
The modified model is implemented based on the complex database. Through pixel-based computing, the
sediment yields per county and per hydrological unit, and the sediment delivery ratio (SDR) per
hydrological unit, are calculated.

Fig. 1 Location of the study site

2 Data and methodology

A model-based approach was used to assess soil loss. The USLE was used because it is one of the least
data demanding erosion models that has been developed and it has been applied widely at different scales.
The USLE is a simple empirical model, based on regression analyses of soil loss rates on erosion plots in
the USA. The model is designed to estimate long-term annual erosion rates for agricultural fields.
Although the equation has many shortcomings and limitations, it is widely used because of its relative
simplicity and robustness (Desmet, 1996). It is also a standardized approach (van der Knijff et al., 2000).
Soil erosion is estimated using an empirical equation:
A = R·K·L·S·C (1)
where A represents mean (annual) soil loss, R is the rainfall erosivity factor, K is the soil erodibility factor,
L is the slope factor, S is the slope length factor, and C is the cover management factor. It has been very
difficult to calculate the soil losses for the valley when a traditional sample method was used to collect the
data for the USLE model. Based on field data, researchers in China have modified parameters and
computing methods to suit reality. In this study, to develop the monitoring of soil losses in the upstream
Chaobaihe River Catchment, remote sensing data, digital elevation model (DEM), and land use and land
cover GIS data were used. The key point of the procedure was data fusion, which was based on an image
pixel un-mixing technique. The values of parameters extracted from satellite sensor data and
 - 168 - International Journal of Sediment Research, Vol. 23, No. 2, 2008, pp. 167–173
generated by GIS for the USLE model soil erosion calculation were pixel based. This improves the
accuracy in every parameter calculation.

2.1 Rainfall data derivation of rainfall erosivity factor

The energy of a given storm depends on all intensities at which the rainfall occurred and the amount of
precipitation associated with each intensity. However, among 44 meteorological stations there are only 5
that recorded the course of every storm. A storm’s maximum 30-minute precipitation intensity must be
known to compute the storm’s erosion index. If a station has not recorded 30-minute intensities and only
monthly and annual rainfall, the 30-minute intensity of the nearest station was assumed to be
representative. From long-term monthly and annual rainfall totals, and rainfall intensities from the 5
meteorological stations, the rainfall R-factor for each station is found by Equation (2) (Bu et al., 2003).
The rainfall R-factor station data were interpolated using an inverse distance interpolation through GIS.
The computing formula is
fB B R PI I = 30 − *1575.0**1281.0 30 (2)

where Pf is annual rainfall (mm), R is mean annual erosivity (MJ.mm/ ⋅ ⋅ )yhha and I30B is a storm’s
maximum 30-minute intensity (mm/h). The results for 2001 and 2002 data in the form of an erosivity map
are shown in Figs. 2 and 3.

2.2 Soil type, texture data and derivation of soil erodibility factor
The soil erodibility factor K represents the average long-term soil and soil-profile response to the erosive
power associated with rainfall and runoff. The soil data used in this study were collected and derived
from the Second Soil Investigation in China. This investigation uses data on soil properties and maps of
soil type distribution. Information on soil surface texture was derived. For each soil type, percentages of
clay, silt and sand were used to estimate K based on the class descriptions. K was estimated using
Equation (3) (Walter and Wischmeier, 1978):
{ } ( ) [ (100*12101.2 )] ( ) ( ) 100/3*5.22*25.3 4 −+−+−∗−∗∗= − K Ssa Sc b c (3)
14.1

Fig. 2 R factor for 2001 Fig. 3 R factor for 2002

where K is the soil erodibility factor (t•ha•h/ ⋅ ⋅mm)MJha , Sc is the percentages of clay, St is the
percentage of silt, Sd is the percentage of sand, and a is the percentage of organic matter. a and b
represent the soil structural status and are given in Table 1. c is the soil saturation capability and is given
in Table 2. Figure 4 shows spatially distributed K values for different soil types in the basin.

Table 1 Soil structural status criteria

A ≤0.5 0.51-1.5 1.51-4.0 ≥4.0

B 4 3 2 1

International Journal of Sediment Research, Vol. 23, No. 2, 2008, pp. 167–173 - 169 -
Table 2 Soil saturation capability criteria
Sc <=10 10-15.9 16-21.6 21.7-27.4 27.5-39 >39.1

c 1 2 3 4 5 6

2.3 Satellite image data and vegetation cover

Vegetation cover is possibly the most crucial element in the process of soil erosion, since it is the one
factor that can readily be altered to control the affects of water and soil loss. With an improved dimidiate
pixel model (Li et al., 2004), the vegetation fraction in the study area was estimated using Landsat-7
ETM+ images. The study area is crossed by two satellite paths. For there to be little cloud in the images,
image data was acquired on 22 and 29 May, 2002. Besides geometric correction, an atmospheric correction
is required to keep images consistent. The raw image data were processed by the EADAS ACTOR2
model. The normalized vegetation index NDVI was then calculated by
NDVI = − + RNIRRNIR )/()( (4)
where NIR is reflectance in the near-infrared channel and R is reflectance in the red channel. The
vegetation cover fraction is calculated by
−
= (5)
NDVI NDVIsoil c
f −
NDVIveg NDVIsoil
where fc is the vegetation cover fraction and NDVIsoil and NDVIveg are two input parameters of the
improved model. Based on the land use categories extracted by overlaying the land cover and soil type
data, 5% and 95% frequencies of NDVI values for individual land use types are parameters derived from
Landsat TM data 5%, 5% for NDVIsoil and 95% for NDVIveg .

2.4 Land use data and cover management factors

Information on land use is difficult to obtain from remote sensing images. Therefore land use maps
covering this study area were collected as assisting data to be combined with the vegetation map to assess
cover management factors. Depending on the existing land use information, the land surface was divided
into three systematic areas, namely natural vegetation areas, low-land areas and artificially disturbed
areas, using different functions to simulate the C-values (Fig. 5).
 Fig. 4 K factor Fig. 5 CP factor

2.5

Topographic factors
To produce a DEM and then deduce slope steepness and slope length, DLG (Digital Line Graphic) data
from 1:50,000 topographical maps were collected. There are 60 topographical maps covering this study
area. The points and lines representing surface elevations and other surface water features were separated
from the DLG data.
Erosion is the detachment of earth material from the surface. Once detached, agents like water or wind
transport the material to a new location where it is deposited. Because the most ubiquitous erosion is the
action of water, it is essential to generate a DEM suitable for hydrological analysis. Arc/Info

- 170 - International Journal of Sediment Research, Vol. 23, No. 2, 2008, pp. 167–173
TOPOGRIDTOOL is a suitable tool to generate a hydrologically correct elevation grid from point, line,
and polygon information (ESRI, 2000). TOPOGRIDTOOL requires parameters, such as contour, elevation
point, boundary and stream locations, which can be obtained from the DLG data. A 30-m resolution DEM
was then generated.
Compared with the existing algorithm (Bu et al., 2003), the slope steepness of a pixel in the DEM was
calculated using modified equation (6), in which slope steepness is calculated from the eight surrounding
pixels. The procedure is programmed in AML and executed on the platform of an ARC/INFO
workstation. It is assumed that water flows along the steepest direction around the center pixel. The
assumption is more in accordance with reality. Slope was estimated using
hh ji
−
=
−

θ j = 1,2,…,8 (6)
ji )(tanmax 1 81
D
−=
where D represents the distance between pixel centers, h or h h
represents the elevation of pixel i or j, and i
i j
is the center pixel of the nine pixels of the computing. Slope length of a pixel based on the DEM was
estimated by
⎤⎢ ⎡⎟ ⎞⎜ ⎛ − ⎞⎜ ⎛
⎣ ⎠ ⎝ ⎟ ⎠ ⎝ = −+
+

1 1 m
1 i
m
i

θ θθ (7) () i

cos
∑ ∑ cos 13.22cos
 D
i DL i i ⎥⋅D
m

1 i

⎦
ii
1

The slope- and slope length factors were estimated using

065.0sin56.4sin41.65 2 S = θ + θ + (8)

=
∑ ∑− = (9)
−

i Dl i cosθ i Di cosθ i D cosθ 111

ii

( ) ( ) ii

where S is the slope factor, is the slope length factor, D is the distance between pixel centers and i l θ is
the angle of the slope. Figure 6 shows the product of SL factors.

2.6 Sediment delivery ratio (SDR)

Gross soil loss is often higher than the sum of losses measured at catchment outlets. The sediment
delivery ratio (SDR) is used to correct sediment yield for this reduction effect. It is defined as the fraction
of gross erosion that is transported from a given area in a given time interval and it is a measure of
sediment transport efficiency, which accounts for the amount of sediment that is actually transported from
the eroding sources to a measurement point or catchment outlet compared to the total amount of soil that
is detached over the same area above that point (Lu et al., 2003). Mathematically, it is expressed as
Y
SDR = (10)
E
where Y is the average annual sediment yield per unit area and E is the average annual erosion over the
same area. Sediment yield data for 2001 and 2002 were collected at six hydrographic stations (units) (Fig.
7). The stations define the sub-basin catchments as a spatial statistical unit. Based on the gross erosion per
unit, the SDRs are calculated.
 Fig. 6 SL factors Fig. 7 Hydrological units International Journal of Sediment Research, Vol. 23, No. 2, 2008, pp.
167–173 - 171 -
3 Results and discussion
The maps of estimated annual soil erosion for 2001 and 2002 are shown in Figs. 8 and 9. The erosion
risk is expressed in qualitative terms rather than in actual rates of soil loss. Six grades are used from very
low to very high. In the upper basin of the Miyun Reservoir for 2001, the area of erosion that could be
neglected was 8,202.76 km2, the area of low erosion 3,269.59 km 2, the area of moderate erosion 3,400.97
km2, the area of high erosion 436.89 km 2, the area of strong erosion 52.19 km 2 and the area of severe
erosion 3.13 km2.

Fig.
8 Result for 2002 Fig. 9 Result for 2002

To support decision making in soil and water conservation, the area of each grade and percentage of
each risk grade of erosion per administrative unit is given. The highest soil loss was 70,353 t/km 2.yr in
Fengning County in 2001, followed by 64,418 t/km2.yr in Chicheng County in 2001.

(a) (b)
 (C)
Fig. 10 Relationship of soil loss, sediment yield and sediment delivery ratio

From Table 3, the SDR in 2002 was lower than that in 2001. The main reasons were decreases in rainfall
erosivity and total runoff. However, many factors influence SDR including hydrological inputs (mainly
rainfall), landscape properties (e.g., vegetation, topography, and soil properties) and their complex
interactions on the land surface. The multitude of such interactions makes it difficult to identify the
dominant controls on catchment sediment response and on catchment-to-catchment variability. Hence the
- 172 - International Journal of Sediment Research, Vol. 23, No. 2, 2008, pp. 167–173
ratios for stations (sub-basins) are not equal.
SDR is a scaling factor used to accommodate differences in areal averaged sediment yields between
measurement scales. The value obtained at a downstream station is often lower than that of an upstream
station. Since the area increases, sediment is more prone to deposition. This is the main reason that the
construction of a series of reservoirs and other irrigation works upstream of the Miyun Reservoir changes
the process of the water cycle and holds up sediment.

Table 3 Measured sediment yield and computed gross erosion (t/km2.yr),

and SDR at each hydrological station

Record Station code Station name Sediment yield

Soil loss in
2001 Sediment yield in 2002 Soil Loss in
SDR in
SDR in
in 2001
2002
2001
2002

1 30302200 Dage 70.4 435.2715 4.16 116.1441 0.162 0.036

2 30302300 Daiying 103 484.0011 1.32 107.4024 0.213 0.012

3 30302600 Xiahui 53.6 144.8334 0 31.05981 0.370 0.000

4 30301300 Xiapu 115 380.133 26.2 164.64375 0.303 0.159

5 30301900 3.26 148.7853 2.16 56.34162 0.022 0.038

Sandaoying

6 30301600 1 237.0546 0.12 107.4312 0.004 0.001

Zhangjiafen

4 Conclusions
This work indicated there are a number of advantages in using the modified USLE equation including
the ability to combine it with a raster-based GIS to produce a cell-by-cell basis for mapping spatial
patterns of soil erosion rates. The advantage of using a GIS raster based framework is that it allows one to
quantify the impact of a single factor on the overall result and it can also easily be updated with improved
datasets. The results could embody the temporal change and spatial pattern of soil erosion in the upstream
area of the Chaobaihe River Catchment, which could support soil conservation strategies.
The current work should be regarded as preliminary and is somewhat limited. Nonetheless, the work to
date provides the foundation to prepare the necessary spatial data as well as information protocols for
further use. This will improve the efficiency in delivering spatial information to external users as well as
standardizing the spatial layers related to these factors.

Acknowledgements
The authors thank Mr. Zhaohong Bu, Institute of Soil Science, Chinese Academy of Sciences, for his
advice on soil erosion mapping. We also acknowledge the Tianjin Investigation, Design & Research
Institute of Water Resources and Hydropower (TIDI) of China for providing data for this project.

References
ArcInfo 8 Help files. 2000, Environmental Systems Research Institute, Inc.
Bu Zhaohong, T. W., Yang Linzhang, Xi chengfan, Liu Fuxing, Wu Jiayu, and Tang Henian. 2003, The progress of
quantitative remote sensing method for annual soil losses and its application in Taihu-Lake catchments. Acta
Pedologica Sinica (in Chinese), Vol. 40, No. 1, pp. 1–9.
David Pimentel N. K. 1998, Ecology of soil erosion in ecosystems. Ecosystems (1), pp. 416–426. Desmet P. J. J. a.
G. G. 1996, A GIS procedure for automatically calculating the USLE LS factor on topographically complex landcape
units. Journal of Soil and Water Conservation, 51, pp. 427–433.
European Environment Agency. 1995, CORINE Soil erosion risk and important land resources - in the southern
regions of the European Community, Commission of the European Communities.
Lu Hua, Moran Chris J., Prosser Ian P., Raupach Michael R., Olley Jon, and Petheram Cuan. 2003, Sheet and rill
erosion and sediment delivery to streams: A Basin Wide Estimation at hillslope to Medium Catchment Scale:
Report E to Project D10012 of Murray Darling Basin Commission: Basin-wide Mapping of Sediment and Nutrient
Exports in Dryland Regions of the MDB.CSIRO Land and Water, Canberra Technical Report 15/03.
J. M. van der Knijff, R. J. A. J., and Montanarella L. 2000, Soil Erosion Risk Assessment in Europe, EUR 19044 EN.
Li Miaomiao, W. B., YAN Changzhen, and ZHOU Weifeng. 2004, Estimation of vegetation fraction in the upper
basin of Miyun reservoir by remote sensing. Resources Science, Vol. 26, No. 4, pp. 153–159. Walter H. and
Wischmeier D. D. S. 1978, Predicting rainfall erosion losses: a guide to conservation Planning. Washington DC,
USDA, ARS.

International Journal of Sediment Research, Vol. 23, No. 2, 2008, pp. 167–173 - 173 -