SOIL PROPERTIES, CONDITION AND SOIL LOSSES FOR SOUTH AND EAST BRAZILIAN FOREST AREAS
CHAPTER 3
SPATIAL DISTRIBUTED MODEL FOR ASSESSING SOIL EROSION
RISK IN A SMALL WATERSHED
1 ABSTRACT
This study was performed in an experimental forest watershed aiming to
predict the potential average annual soil loss using the Universal Soil Loss
Equation (USLE) and a Geographic Information System (GIS). The studied
watershed is located at the Coastal Plain region of Espírito Santo state,
southeastern region of Brazil. All the USLE factors were generated in a
distributed approach using a GIS tool. The layers were multiplied in the GIS
framework in order to predict soil erosion rates. The C factor values were 0.297
for eucalyptus and 0.017 for Atlantic Forest, the first ones obtained directly from
field experiments in Brazil or even in South America. Results showed that the
average soil loss was 6.2 t ha-1 yr-1. Relative to soil loss tolerance, 86% of the
area presented erosion rate smaller than the tolerable value. According to soil
loss classes, 55% of the watershed had erosion less than 3 t ha-1 yr-1. However,
about 12% of the watershed had erosion rates greater than 12 t ha-1 yr-1, thus,
requiring special attention in order to include sustainable management practices
for such areas. Eucalyptus cultivation showed soil loss greater than Atlantic
Forest (natural ecosystem). Thus, an effort should be made to bring the erosion
rates closer to the native forest. Conservation management practices should
begin for the FX soil and forest roads which had the greatest soil loss. The
implementation of the USLE model in GIS framework was found to be a simple
and useful tool for predicting the spatial variation of soil erosion and identifying
critical areas for conservation efforts.
47
2 RESUMO
O estudo foi conduzido em uma microbacia florestada com o objetivo de
estimar o potencial médio de perdas de solo anuais através da Equação Universal
de Perdas de Solo (EUPS) inserida no Sistema de Informação Geográfica (SIG).
A microbacia estudada localiza-se na região dos Tabuleiros Costeiros no
Espírito Santo, sudeste brasileiro. Todos os fatores da EUPS foram gerados de
forma distribuída utilizando a plataforma SIG. Os mapas foram multiplicados no
ambiente SIG para estimar as taxas de erosão do solo. O fator C foi 0,297 para o
eucalipto e 0,017 para a Floresta Atlântica, estes são os primeiros valores
obtidos diretamente de experimentos de campo no Brasil ou mesmo na América
do Sul. Os resultados mostraram que a perda de solo média foi de 6,2 t ha-1 ano-
1. Em relação à tolerância de perdas de solo, 86% apresentaram taxas de erosão
menores que o limite permitido. Com relação às classes de perdas de solo, 55%
da microbacia tiveram perdas de solo menores que 3 t ha-1 ano-1. Entretanto,
cerca de 12% da área da microbacia apontaram taxas de erosão maiores que 12 t
ha-1 ano-1, exigindo uma atenção especial para conduzir um manejo sustentável
nestas áreas. O cultivo de eucalipto mostrou perdas de solo maiores que a Mata
Atlântica (ecossistema natural). Deste modo, um esforço deve ser realizado a
fim de aproximar as taxas de erosão para próximo dos valores na mata nativa.
As práticas conservacionistas devem se iniciar no solo FX e estradas florestais,
os quais tiveram as maiores perdas de solo. A implementação do modelo EUPS
no ambiente SIG mostrou ser uma ferramenta simples e útil para predição da
variação espacial da erosão do solo e na identificação das áreas críticas para
melhor conservação.
48
3 INTRODUCTION
The process of water erosion occurs in watersheds throughout the world.
It is strongly affected by anthropogenic influences. Modification of natural
ecosystems can cause intense environmental degradation, mainly in soils onsite
and offsite. Advanced erosion not only decreases land productivity but also can
transport nutrients, organic matter and agrochemical contaminants.
Knowledge about soil erosion processes as well as how fast soil is
eroded is necessary in planning of conservationist efforts. Modeling is a way to
provide a quantitative and consistent approach to predict soil loss and the
sediment delivery ratio under a wide range of conditions (Bhattarai & Dutta,
2007). In addition, it can be used to evaluate hypotheses and determine the
appropriate soil management and land use for each site (Beven, 1989; Grayson
et al.,1992; Tucci, 1998). Simple empirical methods such as the Universal Soil
Loss Equation (USLE) (Wischmeier & Smith, 1965; 1978), the Modified
Universal Soil Loss Equation (MUSLE) (Williams, 1975), and the Revised
Universal Soil Loss Equation (RUSLE) (Renard et al., 1991, 1997) have been
used for the assessment of soil erosion at the watershed scale (Jain et al., 2001;
Bhattarai & Dutta, 2007; Pandey et al., 2007; Avanzi et al., 2008; Dabral et al.,
2008; Bahadur, 2009; Beskow et al., 2009; Kouli et al., 2009).
The USLE is the simplest and most widely used model for erosion
prediction, which estimates the long-term annual average rate of erosion with
generally acceptable accuracy. Basically, the USLE estimates the soil loss per
unit area based on the following factors (Wischmeier & Smith, 1978): rainfallrunoff
erosivity (R), soil erodibility (K), topography (LS), cover-management
(C), and support practices (P). The drawback of this model is that it is not
capable of simulating deposition, sediment yield, channel erosion, or gully
49
erosion. In addition, such an equation makes no differentiation between rill and
interrill erosion, thus predicting the two processes together. Despite the abovementioned
limiting aspects, the USLE has presented consistent results when
coupled with a Geographic Information System (GIS) for simulations in
watersheds (Jain et al., 2001; Fistikoglu & Harmancioglu, 2002; Onyando et al.,
2005; Bhattarai & Dutta, 2007; Pandey et al., 2007; Dabral et al., 2008; Ozcan et
al., 2008; Bahadur, 2009; Beskow et al., 2009). The combination of USLE and a
GIS has been found to be an effective and suitable approach for estimating the
magnitude and spatial distribution of erosion. USLE model applications in a GIS
framework allows analyzing soil erosion with much more detail since this
process can account for spatial variability (Pandey et al., 2007). Soil erosion
estimates using GIS techniques enable planners to identify sites which are
susceptible to water erosion and also provides a quantitative measure of soil loss
at different scales (Martin & Saha, 2007). The main reason for using a GIS is
that the erosion process varies spatially, so that cell sizes should be used
allowing spatial variation to be taken into account. In addition, the amount of
data necessary for a great amount of cells is required for an accurate
representation of the watershed. Since it is not practicable to input data
manually, GIS can be used to gather and access databases (De Roo & Jetten,
1999).
The objective of this study is to apply the USLE model coupled with
GIS framework for assessing soil loss in a small watershed located in the
Coastal Plain region (100 million hectares) of Brazil. It is expected that this
methodology will provide a useful tool to identify high risk areas for soil
erosion, thus allowing the targeting of conservation and better management
practices in the studied and similar watersheds of the Brazilian Coastal Plain
region.
50
4 MATERIAL AND METHODS
4.1 The Study Area and Soil Description
The experimental watershed (EAW) is located in the Aracruz Celulose
S.A. area, in the Espírito Santo state, southeastern region of Brazil, between
parallels 19°51’S and 19°53’S and, meridians 40°11’W and 40°14’W (Figure 1).
According to Köppen classification, the climate of this region is Aw (tropical
with rainy summer and dry winter), with annual rainfall equal to 1,400 mm
(Embrapa, 2000). The watershed has a drainage area of about 286 ha, containing
eucalyptus plantations and native forest (Atlantic Forest), with advanced
regeneration.
FIGURE 1 Location of the studied area and soil map for the experimental
watershed.
51
The soil classification in the EAW was performed in accordance with
soil classification by Embrapa (2006), and by Soil Survey Staff (1999), the latter
appearing between parentheses. The EAW includes the following most
representative soils of the Brazilian Coastal Plain (100 Million hectares): a)
dystrocohesive Yellow Argisol – PA1 (Hapludult), b) moderately rocky Yellow
Argisol – PA2 (Hapludult), and c) dystrophic Haplic Plinthosol – FX
(Phinthaquox).
4.2 The Universal Soil Loss Equation (USLE)
The USLE allows an estimate of the long-term annual average soil loss
for specific conditions. This model was applied in a GIS environment to
evaluate potential soil loss and its distribution in a forest watershed at the
Brazilian Coastal Plain. The USLE computes soil loss as the product of six
factors (Wischmeier & Smith, 1978):
A R K L S C P Eq. 1
where A is the average annual soil loss per unit of area (t ha-1 yr-1), R represents
the average annual rainfall-runoff erosivity factor (MJ mm ha-1 h-1 yr-1), K is the
soil erodibility factor (t h MJ-1 mm-1), L corresponds to the slope length factor
(dimensionless), S is the slope- steepness factor (dimensionless), C represents the
cover management factor (dimensionless), and P is the support practice factor
(dimensionless).
A Geographic Information System (GIS) was used in order to obtain
spatially distributed results from USLE predictions. The details of all factors in
Equation 1, which were represented in a 10-meter-resolution map, are described
below.
52
a) Rainfall-runoff erosivity factor (R)
Erosivity is defined as the potential of a given rainfall event to cause soil
erosion due to the raindrop impact and runoff. This factor depends primarily on
the intensity and the amount of rainfall (Lal, 1994).
In order to estimate the rainfall-runoff erosivity factor, the rainfall data
were recorded every 5 minutes from January 1998 to July 2004. Suggestions
described by De Maria (1994) were followed, including only rainfall events with
the following characteristics: (a) amount greater than 10 mm; (b) maximum
intensity greater than 24 mm h
53
-1 within 15 minutes; (c) kinetic energy greater
than 3.6 MJ. The kinetic energy was computed for all the rainfall events by
using the equation suggested by Wischmeier & Smith (1958):
E 0.119 0.0873 Log I Eq. 2
where E is the kinetic energy (MJ ha-1 mm-1), and I is the rainfall intensity (mm
h-1).
The EI index for a specific event was calculated as the product of the
total kinetic energy (E) by the maximum 30 minutes intensity (I30), according to
Wischmeier & Smith (1958). Both annual and monthly values were computed as
the sum of all the EI30 (MJ mm ha-1 h-1) values during the respective period of
time with the given conditions described as above.
The sum of EI values for a given period is a numerical measure of the
erosive potential of the rainfall within that period (Renard et al., 1997). It is
worthwhile to point out that the annual EI value in a particular locality
corresponds to the rainfall-runoff erosivity index (R) for that location. The R
factor for the studied watershed was determined by previous study and presented
by Martins (2005).
b) Soil erodibility factor (K)
The soil erodibility is considered as the soil susceptibility to be detached
by splash during rainfall and/or shallow surface flow (Renard et al., 1997). It is
generally considered as an intrinsic soil property with a constant value. The K
factor of the USLE is represented by the mean ratio of soil loss from a standard
plot divided by the rainfall-runoff erosivity index:
K A R Eq. 3
where A is the soil loss (t ha-1 yr-1) and R is the rainfall-runoff erosivity factor
(MJ mm ha-1 h-1 yr-1). Therefore, K factor is expressed in t h MJ-1 mm-1 units.
The standard plot has the characteristics as follows (Renard et al., 1997): (a) it is
22.1 m long and 1.83 m wide; (b) uniform slope of 9%; and (c) continuous
clean-tilled fallow with tillage up and downslope. The K factor for each soil type
in the studied watershed was determined by previous study and presented by
Martins (2005). The K values used were 0.007, 0.017, and 0.0004 t h MJ-1 mm-1
for the PA1, FX, and PA2, respectively. The K factor map was created with GIS
tools and derived from the soil map of the area.
c) Topographic factor (LS)
Both the slope length (L) and the slope steepness (S) have substantial
influence on water erosion ratio (Wischmeier & Smith, 1978). The effects of
these factors have been evaluated separately in studies which use uniformgradient
plots. However, in erosion prediction, the factors L and S typically have
been evaluated together (Renard et al., 1997).
In this study, a large number of elevation points were surveyed
throughout the watershed, which allowed the generation of a 10-m-resolution
Digital Elevation Model (DEM) (Figure 2). The database inputs for the model
taking into account homogeneous cells as small as possible are necessary, thus
allowing soil loss to be characterized with a good resolution. Bhattarai & Dutta
(2007) verified that DEM resolution influences the LS factors, where 30 m DEM
resolution showed better when compared to 90 m resolution using USLE
method. The researchers emphasized the fact that better results can be expected
54
for the resolution which is closer to the slope length used in the derivation of
USLE relationship.
The DEM was imported to the GIS in order to develop a slope map;
thereafter, the latter map was classified into four classes according to Yuksel et
al. (2008): (i) very gentle to flat (<5%); (ii) gentle (5-15%); (iii) steep (15-30%);
and (iv) very steep (>30%) (Figure 3). The slope length factor (L) and the slope
steepness factor (S) were also generated on a grid cell basis.
FIGURE 2 The Digital Elevation Model (DEM) of the experimental watershed.
55
FIGURE 3 Map of slope categories according to Yuksel et al. (2008) for the
experimental watershed.
L
λ
The slope length factor (L) is expressed by (Renard et al., 1997):
22.13 m Eq. 4
where λ is the field slope length (m), and m is the slope-length exponent
(Wischmeier & Smith, 1978). A grid size of 10 m was used as field slope length
(λ). Similar procedure was adopted by several researchers (Liu et al., 2000; Jain
et al., 2001; Fistikoglu & Harmancioglu, 2002; Bhattarai & Dutta, 2007; Pandey
et al., 2007; Dabral et al., 2008; and Beskow et al., 2009).
56