VNU Journal of Science, Earth Sciences 27 (2011) 63-76

Defining required forest area for protection soil from erosion

in Vietnam: a GIS-based application

Tran Quang Bao1,*, Melinda J. Laituri2

1
Vietnam Forestry University, Xuan Mai, Chuong My, Ha Noi, Vietnam
2
Warner College of Natural Resources, Colorado State University, Fort Collins, CO 80523, USA

Received 15 March 2011; received in revised form 19 April 2011

Abstract. Forests play an important role in reducing erosion. In Vietnam, destroying natural
forests in mountainous areas has caused serious environmental problems for sustainable
development. Required forest areas for protection of soils from erosion in Vietnam are defined in
this study. An algorithm of defining required forest area for soil erosion prevention is based on a
comparison of soil loss prediction and its threshold of 10 ton ha-1yr-1 (soil loss tolerance) within
the GIS environment. Soil loss is predicted from rainfall erosivity index, slope, porosity and
vegetation structures in which rainfall index is calculated from 30 year monthly rainfall data of
158 weather stations. A map of erosion risk for Vietnam illustrating potential to erode soil was
generated from slope, rainfall index and soil porosity by using spatial interpolation and map
algebra techniques in ArcGIS. Vegetation index, a function of canopy closure, height, ground
cover and litter cover, is classified into four groups. Required forest areas for protection of soil
from erosion are defined from an erosion risk map in comparison with categories of vegetation
index. An area (a raster cell) requires forest (natural forest or the others) when its erosion risk is
higher than the vegetation index.
Keywords: Soil Erosion, GIS, Required Forest Area, Erosion Risk Map, Soil Loss.

1. Introduction∗ rates are low on land with natural vegetation

cover, about 2 tons ha-1yr-1 in relatively flat land
Soil erosion by water is one of the most and about 5 ha-1yr-1 in mountainous areas
serious environmental problems in the world. It (Pimentel et al., 1998) [3].
causes adverse effects on soils, agricultural In tropical regions where mean annual
production, water quality (Lal, 2001) [1]. sediment yield estimated is greater than 250
Worldwide, soil erosion rate are highest in tons km-2 (Walling at al., 1983) [4], upland
Asia, Africa, and South America, averaging 30 areas are usually protected from erosion by a
to 40 tons ha-1yr-1, and lowest in Europe and the dense vegetation cover. Consequently, cutting
United States, averaging about 17 tons ha-1yr-1 vegetation has caused an increase in runoff and
(Pimentel et al., 1995) [2]. However, erosion erosion (Morgan, 2005) [5]. Sidle et al. (2006)
_______ [6] has summarized some key note papers about
∗
Corresponding author. Tel.: 84-4-33608418. soil erosion in Southeast Asia and concludes
E-mail: baofuv@yahoo.com
63
64 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76

that forest conversion to agriculture and exotic facilitated the estimation of soil erosion and its
plantation (e.g., shifting cultivation) have spatial distribution over large areas. For
significant effects on both surface and landslide example, Yukel et al. (2008) [15] applied the
erosion. The rates of surface erosion depend on CORINE model integrated with remote sensing
the extent dynamic management practices and GIS to generate an accurate and
disturb and compact soil, alter ground cover, inexpensive erosion risk map in Turkey. Wang
and modify soil properties. Therefore, accurate et al. (2003) [16] estimated soil loss by
estimation of soil loss or evaluation of erosion integrating a sample ground data set, TM
risk has become an urgent task. Erosion images, and a slope map and showed that the
prediction can help to address long range land geostatistical method performed significantly
management planning under natural and better than traditional stratification in terms of
agricultural conditions (Angima et al., 2003) overall and spatially explicit estimate. Several
[7]. studies found applied GIS to interpolate
independent factor maps in RUSLE model (or
Efforts to mathematically predict soil CORINE), then to overlay these maps to
erosion by water have occurred only since the generate a regional erosion risk map
1930s. Several models have been developed for (Bissonnais et al., 2001; Lufafa et al., 2003;
estimating soil loss (e.g., Wischmeier and Kheir et al., 2006; Qing et al., 2008) [17-20].
Smith, 1965; Morgan et al., 1984, 2001;
In Vietnam, forests have long been
Woolhiser, 1990; Quynh, 1996) [8-12]. The
recognized to provide an important role in
initial parameters in these models include
environmental protection (Lung et al., 1995;
susceptibility of soil to erosion, potential Quynh, 1996; Dien, 2006) [12,21,22].
erosivity of rainfall and runoff, and soil However, under pressure of economic
protection afforded by plant cover (Renard et development, the demand land for agricultural
al., 1997) [13]. In practice, the Revised and other sectors has increased creating
Universal Soil Loss Equation (RUSLE) model conflicts between land managers. Natural
initially developed by Wishchmeier and Smith forests mostly distributed in mountainous areas
(1965) has been most widely used. It was have experienced high deforestation rates since
originally developed for use on cropland. The 1980s (FPD, 2008) [23]. Consequently, soil
RUSLE has been applied in different land uses erosion in upland has caused serious
(Renard et al., 1997) [13]. However, due to the environmental problems (Lung et al., 1995)
complexity of defining factors of RUSLE for a [21]. There is an essential need to balance
given region, the application of the RUSLE in between agriculture and forests, and minimize
Vietnam has been challenging in term of as much forest land as possible while still
prediction accuracy and its validation (Quynh, ensuring positive environmental effects of
1996) [12]. forest. Responding to those problems, this study
Traditionally, soil loss was predicted at the applies an empirical model for predicting soil
local scale based on the factors usually loss to produce an erosion risk map, and define
calculated from field measurement. Soil erosion required forest areas for protection of soil from
prediction at large scale is often difficult due to erosion for Vietnam. Spatial analyses and
spatial and temporal variability of model’s interpolation techniques in GIS are used for this
factors (Lu et al., 2004) [14]. In recent decades, study. The input data layers for mapping
the development of GIS techniques has include DEM, rainfall and vegetative cover.
 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76 65

2. Methodology
A=
(2.31 *10 −6
* K *α 2 ) (1)
2
 CC 
2.1. Study Sites and Data Sources  + GC + LC  * P
 H 
Required forest areas for protection of soil
from erosion are identified for all territory of Where:
Vietnam, an S-shaped country located in the A = estimate average soil loss (mm year-1)
tropical monsoon area in the southeast of Asia
α = slope (degree)
with a great variety of deltas, mountains, forest
mosaics, and climates. It has a rather high CC = canopy closure (maximum is 1.0)
temperature and humidity, average annual H = forest height (m)
temperature and humidity are above 200C and GC = ground cover (maximum is 1.0)
80%, respectively. Average total annual rainfall
LC = dried litter cover (maximum is 1.0)
is approximately 1940 mm. Total land area is
about 330.000 km2, three fourths of the P = soil porosity (maximum is 1.0)
Vietnam areas is covered by mountains causing K = rainfall erosivity factor, calculated
differences in climate regime between regions based on monthly rainfall (equation 2)
(VNEA, 2006) [24]. Forest cover is about 38.2 12
 R  16+331*lg[(−5.8263+ 2.481*ln(Ri ))/ 25.4] (2)
% of which natural forests is account for 80 % K = ∑ i * 
i=1  25.4  100 
and plantation forests is account for 20% (FPD,
2007) [23]. Data sources used for spatial Where: Ri is rainfall of month ith.
analysis include: National Elevation Dataset The acceptance limits of erosion is 10 ton
(90m x 90m); 30 years monthly rainfall data ha-1 year-1, this is the maximum rate of soil
gauged in 158 weather stations of Vietnam; erosion that can occur and still permit crop
Archives data of 63 research plots of vegetation productivity to be sustained economically
structures and soil loss measurement. These (Hudson, 1977; Renard et al., 1997) [13,25],
plots are representative for different vegetation and approximately equivalent to 0.8mm yr-1. To
types in Vietnam (Quynh et al., 1996) [12]. prevent soil degradation, annual soil loss (A) is
required to less than the sustainably
2.2. Criteria for Defining Required Forest Area replacement rate (0.8 mm yr-1).

The amount of soil erosion by water is an Then, A =

(2.31 *10 −6
* K *α 2 ) ≤ 0.8
2
integration of the effects of vegetation cover,  CC 
topographic features, climatic variables, and  + GC + LC  * P
 H 
soil characteristics (Renard et al., 1997) [13]. In
mm yr-1 (3)
this study, to define required forest areas for
soil erosion protection, average soil loss per  CC 
Let C1 =  + GC + LC  (4)
unit areas was spatially predicted for Vietnam H 
by applying a soil loss equation prediction
is index of vegetation for soil protection. An
developed for Vietnam (Quynh et al., 1996)
area has potential soil erosion less than
[12]. The relationship between soil loss
replacement rate when its C1 meets the
prediction and rainfall, slope, vegetation cover
inequality equation (5) derived from inequality
structures and soil porosity factors can be found
(3).
expression in the following equation.
66 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76

2.3. Spatial Analysis for Defining Required

C1 ≥ (2.31 *10 −6 * K * α 2 ) /(0.8 * P) (5)
Area of Protection Forest
Let C2 = (2.31*10−6 * K *α 2 ) /(0.8 * P) (6)
The digital maps of elevation and rainfall of
is index of erosion risk. C2 does not depend on Vietnam are developed in GIS, using Spatial
vegetation cover structure or other changeable Analysis in ArcGIS 9.2 software (ESRI, 2008)
factors. It is only affected by stable factors (i.e., [26]. We used these maps to produce a map that
slope, rainfall factor, and soil porosity). Based spatially identified erosion risk (C2) of
on value of C2 for a specific area, we can Vietnam. This was compared with the threshold
identify the corresponding vegetation cover of vegetation index (C1) to generate a map of
structure (C1) to protect soil from erosion. required forest area for erosion protection. Figure
1 indicates the methodology used in the model.

Figure 1. Analytical methodology for defining required forest area.

The explanations of each procedure in the by overlaying 12 monthly rainfall maps based
model will be followed: on a raster calculation in equation (2).
(1) Slope data layer was derived from (3) An erosion risk map (C2) for Vietnam
National Elevation Dataset (DEM) was produced from three input layers (i.e.,
(2) Calculated average monthly rainfall for porosity, slope, and rainfall erosivity maps), in
158 meteorological stations in Vietnam, then which P was assumed to equal 0.4, this is
spatially interpolated 12 monthly rainfall maps equivalent to the average porosity of fallow
from these point data. A map of rainfall land following one year of traditional swidden
erosivity factor (K) for Vietnam was generated cultivation (Quynh at al., 1996) [12]. The raster
 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76 67

calculation for the erosion risk map was based Algebra functions (i.e., If Then Else) in Spatial
on equation (6). Analyst Tool of ArcGIS 9.2 (Theobald, 2003)
(4) From the data of vegetation cover [27]. Total areas of forested cells are required
structures (i.e., canopy closure, ground cover, forest areas for protection soil from erosion in
litter cover, and height) of previous study Vietnam.
(Quynh et al., 1996) [12], calculate C1 for
2.4. Rainfall Interpolation
different main cover types in Vietnam (equation
4). Index of vegetation covers (C1) are Monthly rainfall maps are interpolated from
classified into five classes based on their 30-year averaging rainfall data of 158 weather
relationship with soil loss (Table 1). stations relative evenly distributed in Vietnam
Table 1. Classes of vegetation cover structure index
(Fig. 2). The interpolation method used is
in Vietnam Inverse Distance Weighted (IDW), in which an
unknown point is interpolated from usually
Cover types C1 scattered set of known points (Bartier et al.,
1996) [28].
Natural Forests >1.7 n

Plantation forest, agro-forests 1.3 - 1.7 ∧ ∑ Z ( s )λ i i

Z ( s0 ) = i =1
n (7)
Industrial plants, fruits 0.9 - 1.3
∑λ
i=1
i
Agriculture 0.6 - 0.9
Where:
(5) Defining required protective forest area
Z(si) is rainfall of station ith
Algorithm of this step is based on a ∧
comparison between actual value of erosion risk Z ( s 0 ) is interpolated rainfall for location so
(C2) and threshold of vegetation index (C1) in n is number of the nearest stations used for
Table 1. An area (a raster cell) is required interpolation, n is chosen equal 3.
natural forest when its C2 is greater than 1.7
λi is weighted value for station ith,
(i.e., C1 of natural forest). It is required natural
forest, or plantation forest, or agro-forest, when λi = 1 , where di is distance from location
d i2
its C2 is less than 1.7, but greater than 1.3 (i.e.,
si to location so.
C1 of plantation forest, agro-forest). These
conditional statements were executed by Map
68 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76

­
Legend
Weather Station
Vietnam

0 3060 120 180 240

Kilometers

Figure 2. Map of Vietnam showing the locations of 158 weather stations in Vietnam.
 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76 69

3. Results and analysis

3.1. Rainfall Interpolation and Rainfall Erosivity Factor

The temporal and spatial distributions of monthly rainfall in Vietnam are illustrated in Figure 3
from January to December.

Jan. Feb. March April

May June July August.

70 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76

Sept. Oct. Nov. Dec.

Figure 3. Interpolated average monthly rainfall for Vietnam.

As shown in Figure 3, average annual Thiet there is either no rain for 2-3 months or
rainfall varies dramatically ranging very little rainfall. The highest monthly rainfall
approximately from 1000mm in Nha Ho to occurring in August and September is 900–
4000mm in Bac Quang. The rainfall is 1000mm (e.g., Bac Quang, Nam Dong). The
unevenly spatio-temporally distributed. The rain season starts from April to October,
variation of rainfall is the main cause of particularly from July to December in the
droughts in the dry season and floods in the central coastal area. The rainfall in rainy season
rainy season. In some areas like Ham Tan, Phan accounts for 80% of the total annual rainfall.
T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76 71

a) b)

Figure 4. Map of slope (a) and rainfall erosivity factor (b).

72 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76

3.2. Erosion Risk and Required Forest Areas consequently created a great variability within
erosion risk map of Vietnam (Fig. 5a). The
As indicated above, about three fourths of northwest and central west areas of Vietnam
the total natural land area of Vietnam is covered (red color) have the highest potential to erode
by hills and mountains, with a general soil. The two large areas having the lowest
downward slope from west to east (Fig. 4a). A erosion risk (blue color) are located in Red
high gradient of slope, together with unevenly River Delta (northern) and Mekong River Delta
distribution of rainfall erosivity (Fig. 4b), (southern).

(a) (b)

Figure 5. Maps of Vietnam showing (a) erosion risk and (b) required protective forest areas.
 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76 73

The map of required forest areas for value of each factor in the equation is fixed,
Vietnam (Fig. 5b) was generated from erosion which only can be established after it has been
risk map in comparison with vegetation index measured (Hudson, 1977) [25]. In Vietnam,
(inequality 5). Total required forest areas for there are limited applications of the RUSLE to
protection of soil from erosion for Vietnam are predict erosion from land surface due to a lack
7,191,436 ha, of which 2,469,497 ha is natural of references to qualitatively assess the factors
forest. Fifteen out of 64 provinces do not for given circumstances. Lung et al. (1995) [21]
require forests for erosion prevention, most are has defined factors in the equation (8) for the
distributed in the Red and Mekong river deltas. Central Highlands, and also identified C factor
Provinces requiring high percentages of forest for different forest covers in this area (Table 2).
cover are mainly located in the northwest and However, there are some disadvantages when
south central of Vietnam. applying this equation to predict soil erosion;
these include: (1) there is no verification for
method used to define factors; (2) vegetation
4. Disscussion classifications are not detailed enough; (3) and
experimental plots were designed in a small
The revised universal soil loss equation
range of the factors.
(RUSLE) is an erosion model predicting
longtime average annual soil loss, it is a Table 2. An example of USLE factors calculated for
powerful tool that is widely used in the United the central highland of Vietnam
States and many foreign countries (Renard et
al., 1997) [13]. The RUSLE was developed Locations R Ka LS C P
initially by Wischmeier and Smith (1965, 1978)
Konhanung 872.5 0.021 2.37 (100) 0.0083b 1.0
[29] for original use on cropland. It has been
being applied to different land uses (e.g., Pleiku 943.3 0.024 4.38 (150) 0.0076c 1.0
rangeland, forestland). The RUSLE is
Sources: Lung et al. (1995) [21]
expressed as:
a
K factor for Bazan soil; b C factor for
A = R * K * L*S * C * P (8)
bamboo forest; c C factor for grass
Where: A = estimated spatial average soil
These disadvantages are resolved by
loss per unit area
applying the erosion prediction equation (1)
R = rainfall-runoff erosivity factor used in this study. This equation was
K = soil erodibility factor established based on observations of 63 field
L = slope length factor plots of different cover types, including natural
forests, plantation forests, orchards, abandoned
S = slope steepness factor
land, grazing land and paddy field (Quynh et
C = cover-management factor al., 1996) [12]. Soil erosion in each plot is
P = support practice factor measured and estimated by using the triangle of
three steel poles. In the middle of each pilot
The essence of universal soil loss equation
plot, place three steel poles in a triangle form.
is to isolate each variable and reduce its effect
The length of each side of the triangle (the
to a number. Soil loss is predicted by
distance from each pole) is 3 m. Each pole is
multiplying the numbers. For a given situation
placed deeply into the soil and left about 20 cm
(e.g., soil type, cover, slope and length) the
74 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76

higher than the surface of the land. Use a long slope, and rainfall. The authors have found a
plastic durable string to connect the three poles close relationship among these variables (Fig.
at the height of 10 cm from the surface, then 8a). They used monthly rainfall as a
measure the distance at 9 points (3 points in replacement of rainfall intensity (Fig. 8b) for
each side of the triangle) from the string to the calculation of rainfall erosivity factor. The root
surface before and after each rain event to mean squared error (RMSE) of soil loss
estimate the thickness of soil layer eroded by prediction by using the equation (1) is about
each rain (mm). Soil loss depth was analyzed in 16%. Recently, the equation has been widely
relation to vegetation structures (e.g., height, applied in Vietnam (Quynh et al., 2006) [12].
canopy closure, ground cover, and litter cover),

16 12
14
10
12 Rainfall Intensity (mm hr-1)
Soil Loss (mm yr-1)

10 8

8 6
6
4
4
2
2

0 0
0 0.5 1 1.5 2 0 100 200 300 400 500 600 700
(CC/ H+GC+LC) Monthly Rainfall (mm)

Figure 6. Bivariate plots of (a) vegetation cover structure (i.e., canopy closure, height,
ground cover, and litter cover) and soil loss (mm yr-1), R2 =0.73; and (b) monthly rainfall (mm)
and rainfall intensity (mm hr-1), R2=0.78, (Quynh at al., 1999) [30].

5. Conclusions threshold for soil loss in tropical regions (10 ton

ha-1 yr-1), we have established two criteria to
Soil erosion by water continues to be define required forest area, one is index of
serious environmental problems in Vietnam. erosion risk (C2), the other one is index of
The primary objectives of this study were vegetation (C1). The map of erosion risk was
applying GIS techniques to define required interpolated from mean 30-year monthly
forest areas for protection soil from erosion in rainfall data, slope, and porosity. The index of
Vietnam. vegetation was calculated for main cover types
Due to difficulties in identifying factors for in Vietnam from available data (i.e., height,
Revised Universal Soil Loss Equation (RUSLE) canopy closure, ground cover, and litter cover).
in Vietnam, the spatially potential soil loss was Applying raster analysis techniques in ArcGIS,
predicted by an equation developed by Vietnam the map of required forest areas for soil erosion
itself, in which soil erosion prediction is a prevention was generated from erosion risk map
function of vegetation cover structures, slope, in comparison with vegetation index. This map
erosivity rainfall index, and soil porosity. Based is a spatial distribution of required natural
on the selected soil loss equation and the forests, other forests, or non forests.
 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76 75

References [13] K.G. Renard, G.R. Foster, G.A. Weesies, D.K.

McCool, D.C. Yoder, Prediction soil erosion by
[1] R. Lal, Soil degradation by erosion. Land water: A guide to conservation planning with the
Degradation & Development, 12 (4) (2001) 519. revised universal soil loss equation (RUSLE).
USDA, Agriculture HandBook, No. 703. (1997)
[2] D. Pimentel, C. Harvey, P. Resosudarmo,
K.Sinclair, D.Kurz, M.McNair, S.Crist, [14] D. Lu, G. Li, Valladares, G.S., and Batistella,
L.Shpritz, L.Fitton, R.Saffouri, R.Blair, M., Mapping soil erosion risk in Rondônia,
Environmental and Economic Costs of Soil Brazilian Amazonia: using RUSLE, remote
Erosion and Conservation Benefits. Science, 267 sensing and GIS. Land Degradation and
(1995) 1117. Development 15 (2004.) 499.
[3] D. Pimentel, N. Kounang, Ecology of Soil [15] A. Yuksel, R. Gundogan, A.E. Akay, Using the
Erosion in Ecosystems. Ecosystems, 1 (5), Remote Sensing and GIS Technology for
(1998) 416. Erosion Risk Mapping of Kartalkaya Dam
Watershed in Kahramanmaras, Turkey. Sensors,
[4] D.E. Walling, B.W.Webb, Patterns of Sediment 8 (2008) 4851.
Yield. In Morgan, R.P.C. (3rd ed.), Soil Erosion
and Conservation, Blackwell Publishing, 1983, [16] G. Wang, G. Goerge, S. Fang, A. Alan, Mapping
pp. 1-10. multiple variables for predicting soil loss by
geostatistical methods with TM images and a
[5] R.P.C. Morgan, Soil Erosion and Conservation. slope map. Photogrammetric engineering and
Third Edition, Blackwell Publishing, 2005. remote sensing, 69 (8) (2003) 889.
[6] R.C Sidle, A.D. Ziegler, J.N. Negishi, A.R. Nik, [17] Y.L. Bissonnais, C. Montier, M. Jamagne, J.
R.Siew, F. Turkelboom, Erosion processes in Daroussin, D. King, Mapping Erosion Risk for
steep terrain-Truths, myths, and uncertainties Cultivated Soil in France. Catena 46 (2001) 207.
related to forest management in Southeast Asia.
Forest Ecology and Management 224 (2006) [18] A. Lufafa, M.M. Tenywa, M. Isabirye, M.J.G.
199. Majaliwa, P.L. Woomer, Prediction of Soil
Erosion in a Lae Victoria Basin Catchment using
[7] S.D. Angima, D.E. Stott, O’Neill, M.K., Ong, a GIS-based Universal Soil Loss Model.
C.K., Weesies, G.A.., Soil erosion prediction Agricultural Systems 76 (2002) 883.
using RUSLE for central Kenyan highland
conditions. Agriculture Ecosystems and [19] R.B. Kheir, O. Cerdan, C. Abdallah, Regional
Environment 97(2003) 295. Soil Erosion Risk Mapping in Lebanon.
Geomophology, 82, 347 (2006)359.
[8] W.H. Wischmeier, D.D. Smith, Predicting
rainfall erosion soil losses from cropland east of [20] X.Y. Qing, S.X. Mei, K.X. Bin, P. Jian, C.Y.
the Rocky Mountain: A guide for selection of Long, Adapting the RUSLE and GIS to Model
practices for soil and water conservation. USDA, Soil Erosion Risk in a Mountains Karst
Agriculture Handbook, 1965, No. 282. Watershed, Gouzhou Province, China.
Environmental Monitoring and Assessement,
[9] R.P.C. Morgan, D.D.V. Morgan, H.J. Finney, A 141 (2008) 275.
predictive model for the assessment of soil
erosion risk. Journal of Agricultural [21] N.N. Lung, V.D. Hai, Research on scientific
Engineering Research 30 (1984) 245. foundation of technical economic solutions for
planning and designing watershed, upstream
[10] R.P.C. Morgan, A Simple Approach to Soil Loss protective forests, and coastal mitigating storm
Prediction: A Revised Morgan-Morgan-Finney forests. Report for National Project No. K03-09,
Model. Catena 44 (2001) 305. Ha Noi, 1995.
[11] D.A. Woolhiser, R.E. Smith, D.C. Goodrich, , [22] P.V. Dien, Water retention capability of some
KINEROS: a kinematic runoff and erosion vegetation types in protection areas of Hoa Binh
model: documentation and user manual. USDA hydropower plant. Ph.D Dissertation, Forestry
Agricultural Research Service ARS-77, 1990. University, Vietnam, 2006.
[12] V.V. Quynh, N.N. Lan, Capability to protect soil [23] FPD, 2008. Vietnam Annual Forest Cover, 2007.
of different vegetation covers in Ham Yen, Tuyen Forest Protection Department (FPD), Ministry
Quang. Report for Vietnam - Swedish of Agriculture and Rural Development,
Cooperation Program, SIDA, 1996. http://www.kiemlam.org.vn
76 T.Q. Bao, M.J. Laituri / VNU Journal of Science, Earth Sciences 27 (2011) 63-76

[24] VEPA, 2006. Environmental Report of Vietnam, [28] P.M. Bartier, C.P. Keller, Multivariate
2006. Vietnam Environment Protection Agency. interpolation to incorporate thematic surface data
[25] N.W. Hudson, Soil Conservation. Cornell using inverse distance weighting (IDW).
University Press, Ithaca, New York, 1977. Computers and Geosciences 22 (7) (1996) 795.
[26] ESRI, 2008. ArcGIS Spatial Analysis, [29] W.H. Wischmeier, D.D. Smith, Predicting
www.esri.com/spatialanalyst. rainfall erosion soil losses: A guide to
conservation planning. USDA, Agriculture
[27] D.M. Theobald, GIS Concepts and ArcGIS Handbook, 1978, No. 537.
Methods. 1st Edition, Conservation and Planning
Technologies Publisher, USA. 2003, pp. 238- [30] V.V. Quynh, N.H. Tuat, H.Q. Khai, Forest
266. Criteria for Soil Conservation. Report for
Project Funded by Ministry of Agriculture and
Rural Development, 1999.