Determination of soil erosion and sediment yield, and development of a 3D map for Sawah, Agroforest fields and forests in Sumani Watershed

Determination of soil erosion and sediment yield, and development of a 3D map for Sawah, Agroforest fields and forests in Sumani Watershed, West Sumatra, Indonesia

Aflizar1

Supervisor
Tsugiyuki Masunaga
1) Faculty of Life and Environmental Science, Shimane University, Japan;


Abstract

Determination of erosion and sediment yield is essential in planning proper land uses and soil conservation actions for sustainable technologies. The study which was conducted in 3 land use areas, i.e. Sawah, Agroforest field and Forests aimed at determining erosion rates and sediment yield in the surrounding watershed. Two models i.e. USLE model run in Surfer software (USLEs models) and Sediment Delivery Ratio models using observed sediment sample (SDR models) were deployed to compare erosion rates in Sawah, Agroforest field, Forest contrasted in the two different rainfall periods i.e. 1992-1993 period and 1996-2007 period. The study objectives also included the development of a three dimension (3D) map of soil erosion. In the watershed indicated on the map, it was evident that 94%. Soil loss was from erosion and 6% pf the area was deposited with sediment. Soil loss occurred in upland areas classified as very low to low in sawah and forest areas. It was moderate to extreme severe in Agroforestry areas. 6% soil deposition in Sumani watershed spread 67% in both upland areas occupied with sawah rice terrace and down slope forest areas. It was 33% in lowland areas of sawah rice terrace. The result simulated from USLEs models was 46.64 ton ha-1y-1 while SDR models simulated 49.25 ton ha-1y-1. These results are not significantly different from each other. However in Subwatersheds, the USLEs models measured more than half of the erosion rate predicted by SDR models. The erosion rates in sawah simulated from 1992-1993 year period was 0.91 ton ha-1y-1 where as from 1996-2007 period rainfall data was 0.94 ton ha-1y-1. rates from Agroforest field were 98.94 and 102.62 ton ha-1y-1 respectively. From forests, rates were 0.55 and 0.57 ton ha-1y-1 respectively. Sawah and forest areas are land uses with minimum soil erosion cases in Sumani Watershed. This , further suggests that usage of lands as Sawah and Forest would contribute greatly to minimizing soil erosion. In contrast 51% of Agroforest fields require soil conservation measures. It was also noted in the study that besides being located in the low lying terrain, the presence of terraces in Sawah enhanced the accumulation of sediments from water erosion.

Key words: Soil erosion, Sediment yield, soil deposited, USLEs models, SDR models, Sumani watershed

Introduction

Estimation of soil loss and sediment yield is essential for assessing soil erosion hazard. It is also vital for determining suitable land uses and soil conservation actions for a watershed (Moehansyah et al. 2004). Soil erosion in Indonesia is one of the nation’s most serious environmental degradation problems. Although many agent cause soil erosion, In Sumani watershed, it is caused mainly by high rainfall. Detection and measurement of erosion problems have an essential role in influencing better land uses and conservation practices .
There are six parameter inputs in USLE model to simulate soil erosion caused by water. In West Sumatra precipitation rates are characterized by high intensities within short periods of time.
The term sawah refers to a leveled and bounded rice field with an inlet and outlet for irrigation and drainage (Wakatsuki et al. 1998). Agroforestry refer to a system of farming where trees are grown together with crops ( Morgon 1985) and it is similar to mixed farming in Sumani Watershed. Sumani Watershed has erosion hazard categorized as: severe to extreme severe (26.23%), moderate (24.59%) and very low to low (49.18%).
In the experiment, it was found that annual average soil loss for Sumani Watershed is 48.38 ton ha-1yr-1. A relatively lower erosion hazard was found in Sawah and Forest areas.
Young (1991) stated that Agroforestry has a potential to control erosion. Kusumandari et al. (1997) reported that Agroforestry is a best choice for land use to minimize soil erosion rate. Lal (1990) stated that forested areas have quite lower erosion risk than arable lands or plantation areas. These studies have emphasized the importance of water flow and soil loss by water in different land uses within the watershed. The importance of Agroforestry and Forest practices in controlling erosion have been well studied. Unfortunately, there is little quantitative data in Indonesia on the effectiveness of Sawah to control erosion .
The USLE model can be applied to forecast erosion in watersheds in Indonesia (Moehansyah et al. 2004; Kusumandari et al. 1997). Some models with strong theoretical base such as: SEMMED (Jong et al. 1999), WEPP (Amore et al. 2004), EUROSEM (Morgon et al. 1998), GUEST (Ciesiolka et al. 1995), ANSWERS ( Ahmadi et al. 2006), FUERO (Matternicht et al. 2005), AGNPS (Walling et al. 2003), LISEM (Takken et al. 1999), MMF (Morgon 1999) and Erosion 3D (Schmidt et al. 1999), may not be very suitable in the context of Indonesia since it is a developing country mostly with unreliable input data like rainfall, topographic information required to run the models, and most of the time they are not available or difficult to collect due to resource constraints (Mohansyah et al. 2004).
At present the most commonly used methods of predicting the average water erosion rate from agricultural lands are the Universal Soil Loss Equation (USLE) (Wischmeier and Smith 1978) and the Revised Universal soil Loss Equation (RUSLE) (Kenneth et al. 1994).
USLE model run in Surfer software and serves as a tool for analysis of the environmental risk .As such it also was used to map a three dimension (3D) erosion rate and sediment yield for this study and was overlaid with land use map. Soil erosion models, such as the USLE estimates gross soil erosion rate at plot-scale. Since, the USLE estimates gross soil erosion rate at plot-scale, erosion rates estimated by USLE are higher than those measured at watershed outlet (Lu 2006). Sediment delivery ratio (SDR) was used to correct this reduction effect. SDR is defined as the average annual sediment yield per unit area divided by the average annual erosion over that same area (Walling et al. 2003; Simon et al. 2007).
In surrounding watershed, the process of soil erosion in the upland area may result into accumulation by deposition at lowland area and other losses in the river. Therefore, there is need to estimate the land use to determine accumulation of erosion product as well as erosion distribution and determine conservation measures.
This study had two main aims, the first was to measure and understand how the erosion rates and sediment yield are distributed in Sumani Watershed by using USLE model run in Surfer software . This would help to document the erosion rates of different land uses (Sawah, Forested and Agroforestry), and compare soil erosion rates of the watershed measured by the USLE as compared to result of the observed SDR model. The second was to develop a 3D map of soil loss and estimate the magnitude of erosion and accumulation by deposition in land uses type in the surrounding areas of Sumani Watershed in West Sumatra, Indonesia.

2. Material and methods

2.1. Study area in Sumani watershed

The Sumani Watershed, covers 58330 ha (Saidi 1995) and is located in Solok Regency (latitude 00 42’17” to 10 2’2” S, longitude 1000 32’ 41”- 1000 40” E), West Sumatera, Indonesia. (Figure 1). Elevation ranges from 300 to 2500 m asl and it is a typical hilly and mountainous area. Tropical climate (mean annual temperature 25oC, mean annual precipitation 2450 mm, occurring through out the month in all year round dominates the areas).
Sumani watershed is important areas in West Sumatra which is used for hydroelectric power generation, irrigation, domestic water supply, transportation and recreation. Since, the change in land use, these change may have affected the extent of soil erosion and the soil productivity. Sumani watershed also central to its significance in West Sumatra is the production of high quality rice, and contain various land uses, including Sawah, forestry (both plantation and natural), Agroforestry (mixed farming), upland crop land (vegetables field), and settlement, thereby allowing comparisons. This watershed was most appropriate for one of the objectives of the research: to compare the soil erosion rates among Sawah, Agroforestry and forestry areas. The study also aimed to creating three dimension erosion map for Surrounding Sumani Watershed.
Base on latitude, Sumani watershed can be separated into two topographic unit that is lowland areas (<500>500 m asl).According geology map, parent material in the Sumani Watershed area are of tertiary age and permacarbon age (12 million to 240 million years B.P.) (BPTP DAS Sumbar 2000; Haugh 1958; Saidi 1995; Geologi Singkarak 199..). The Upland areas of the watershed are characterized by alluvium fan, tuff volkan and Andesit Mt. Talang ( 12 million-60 million years B.P.). The Lowland areas developed on the alluvium during the Oligocene, eosin (40 – 60 million years B.P.) and tuff volkan during the permacarbon (240 million years B.P.). Dominant land uses in Sumani watershed are Forest (21%), Agroforest field (43%), Sawah (30%) and settlement (3%) and other (3%). Sawah areas occurred dominant in Lembang and Sumani watershed and lowland areas of Aripan, Gawan and Imang subwatershed. Forest areas occurred dominant in altitude more than 1000 m asl in steep slope at Lembang, Sumani and Gawan subwatershed. Agroforestry areas occurred around Sumani Watershed in altitude from 450 to 1200 m asl.

2.2.Landscape and USLE factor mapping and soil sampling

From July to September 2007, the Sumani watershed was mapped on a 125 x 125 m grid (0.25 x 0.25 cm cell on a map scale 1:50000 were get 65536 node) using digitize method in Surver ver. 8 Software. Surver is a contouring and 3D surface mapping program that include gridding option with kriging method which allow to interpolate data onto a user specified grid to produce accurate maps and images from XYZ data (Scientific software group 2008). Pronounced change in vegetation or topography between the grid points as well as creeks and rivers were mapped.
At each of the grid nodes, ground vegetation from Landsat ETM 2002 (Farida et al. 2005) was determined a 0.25 x 0.25 cm cell in map scale 1:50000 that used to determine C and P factor of USLE.
Determine R factor of USLE from three climatology station were marked by step : first step a polygon map was made, second step identified R factor of 81 soils were sampled at the site after that were interpolated based on the parameter of their semi-variograms using the kriging technique, third steep was to remap R factor by grid method to create digital data base using Surver ver.8 software.
Digital Elevation model (DEM) was used to make topographic map and calculation LS factor of USLE. LS factor calculated based on a step by step procedure: First step was to make distribution percentage slope class, second step was transform a slope map into digital map by grid method and make them ready to calculated LS factor .
K factor for the USLE was determined from a soil survey where 81 sites were sampled. Erodobility K factor was obtained from 81 samples augered from a variety of geomorphic positions at 0-20 cm and 20-40 cm depths. Bulk density and soil permeability, was assessed by a 100 cm3 soil core same locations where sampling took place.Thus, in some sites soil structure were identified by soil survey manual. GPS was used to record sampling positions. Distribution map of K factor was made using similar method used in creating map for R factor.

2.3. Sample preparation and laboratory analyses

The composite samples were air dried immediately after sampling, bulk density was determined from corer samples. Soil permeability was determined from corer sample by De Boodt (1967) method. Soil particle-size fraction were determined using the pipette method following H2O2 treatment to destroy organic matter and dispersion of soil suspensions by Na-hexametaphosphate. Soil organic carbon was determined to each soil sample by standard of method from Walkley and Black (1934) (ASA-SSSA 1982; Soil Survey staff 1993a, 1993b).
To predict soil loss in the spatial domain. The grid-cell were set 125 m by 125 m. which was resolution possible with the available data and computer facilities. Althought the USLE is recommended for small plot and fields (Wischmeier and Smith 1978), the cell sizes used in this study were considered to give adequate detail because this study was of reconnaissance scale (Mati et al. 2000). Each grid was viewed as a single slope plane for which the USLE could be applied individually. Grid was labeled according coordinate zone 47, WGS 84 southern hemispire for each single grid data in subwatershed to make possible estimated soil loss within a given subwatershed in the Sumani Watershed.

2.4. USLE model and SDR method predict erosion in Sumani Watershed

The overall methodology involved use of a soil erosion model, the universal soil loss equation (USLE) (Wischmeier and Smith 1978) in a Surver ver. 8 software. Climatological data obtain from weather stations during 1996-2007 (BMG Sicincin 2007), field survey and soil analyses data of 81 sites, land use cover was interpreted from Landsad TM satellite image taken in July 2002 (Farida et al. 2005) with reconnaissance survey.
In the USLE model, mean annual soil loss is expressed as a function of six erosion factors:
A = R K L S C P (1)
Where : A is the estimated soil loss in t ha-1y-1 ; R is the erosivity of rainfall in MJmm ha-1y-1 ; K is inherent soil erodibility in t.ha.h-1MJ mm-1; L is length of the slope factor, dimensionless; S is slope factor, dimensionless; C is crop cover factor, dimensionless; and P is a factor that accounts for the effects of soil conservation practices, dimensionless.
In general, rainfall erosivity (R) and soil erodibility (K) are the most important factors that need evaluation based on local conditions for successful application of the model (Chris et al. 2002). The Erosivity index was calculated for one year as the sum of erosivity index values of individual months for ten year rainfall data collected from 3 stations climatological . Values for factors C and P were estimated using the procedure outlined in Wischmeier and Smith (1978) and the range of values suggested by Morgon et al. (1978) and Abdurachman et al. (1984). Vegetative cover in Sumani Watershed was
Investigated in Landsat TM imaginary in 2002 and by field observation conducted in 2007.
For computing the monthly value of the R factor, the following equation proposed for Indonesia by Bols (1978) was used :
R = 6.19(Rf)1.21(Rn)-0.47 (Rm)0.53 (2)
where: R is monthly erosivity; Rf is total monthly rainfall; Rn is number of rainy days per month; and Rm is the maximum rainfall during 24 hour in the observed month.
The value for factor K is computed using the following equation (Wischmeier and Smith, 1978):
100K= 2.713M1.14(10-4)(12-a)+3.25(b-2)+2.5(c-3) (3)
where: M is given by [(St – Svf)/100] – Cf ; a is the percentage of soil organic matter content; b is the structural code; c is the permeability class code of the soil; St is silt fraction of soil in percentage; Svf is very fine sand fraction in soil in percentage; and Cf is clay fraction in soil in percentage.
To predict R factor and K factors of the grid that did not have own data, Interpolation by the Nearest Neighbor kriging method (Golden software 1995) that assigns the value of data of the nearest point to each grid node. This method is useful when data are already evently spaced, but need to be converted to a surfer grid file. Alternatively, in cases where the data are nearly on a grid with only a few missing values, this method is affective for filling in the holes in the data. Good performance of this method was reported by Goovaerts (2000), Granadas et al. (2002), Takata et al. (2007b), Bohling (2005). Yamamoto (2005) used kriging method for assessing uncertainties at unsampled location . Thus, estimating of O3 exposures across united stated (Reagen.1984, Lefohn et al.1987, Knudsen and Lefohn. 1988).Also, kriging used the technique to more accurately predict the extent of gold deposit in Unsampled areas (Leadon 1991).

For LS factor calculation at the original USLE formula for estimating the slope length and slope steepness can be used (Wischmeier and Smith 1978). Liu et al. (2000) reported that with an increase in the slope steepness from 20% to 40% and 60%, the slope length exponent does not change. Therefore, in the present study two equation for slope gradient 20% as given in the USLE (eq.4) and for slope gradient >20% as incorporated in the USLE (eq. 5) were separately used (Renard et al. 1994; Wu et al. 2007)
LS = (l/22)m .(65.41. sin 2 X – 4.56. sin X + 0.065) (4)
LS = (l/22)0.7 .(6.432. sin (X0.79 ).cos (X)) (5)
Where : l is the slope length in m; s is percentage of slope; x is the slope in degrees; m is exponent that varies with slope as in (0.2 for <> 5%)
Substitutes are used for C and P factors. Land use types in Sumani Watershed were investigated by interpreting image photo of Landsat TM 2002 confirming with field survey data in August 2007. C factor values were taken as 0.001 for primary forest, 0.29 for grasslands (Brachiaria sp), 0.4 for agriculture lands on upper slope mainly cultivated by crop like chili, onion, soybean, maize , 0.2 for a mixed garden dominated by perennial crops as coconut, clove, coffee, teak, mahagony, sawo, avocado, melinjo, rubber. cinnamons, 0.3 for coconut, 0.01 for sawah , 0.01 for bush and shrub, 0.002 for pine plantation and 0.95 for settlement. Sawah is a dominant conservation practice in sawah area traditional terrace with P factor value 0.4. Agricultural fields, and mixed gardens and coconut have P factor of 0.5 because they have plantation crop with cover ground. Very minimal conservation practices are done on the other land use pattern. In these areas very small area has conservation practices, P factor values were assumed as 1.

Sediments from sixth outlet of major river in Sumani Watershed were collected every month for one year observation in 1992 - 1993. Sediment delivery ratio (SDR) was used to calculate erosion to correct for USLE model. It is a dimensionless scalar and conventionally expressed as :

Y= E (SDR) A (6)

Where Y is the average annual sediment yield per unit area and E is the average annual erosion over that same area (Walling, 1993; Richards, 1993), SDR = α Aß , where A is watershed area (in km2) , α and ß are empirical parameters ( Lane et al.2007; Roehl 1962). This study used instant, actual values of SDR derived from experimental data from Roehl (1962).
Amore et al. (2004) reported that In order to obtain the total sediment yield, soil loss from each area should be multiplied by its, respective, SDR. Yearly average sediment yield Ys was thus assessed by the following expression

Ys = RKLSCP SDR A (7)

2.5. Surver processing and statistical analysis
Data on soil properties , erosion and sediment yield from sawah, Agroforestry and Forest areas were tested using a t-test with significant level of 1 and 5%. Soil properties to measure K factor from location not sampled were interpolated based on the parameters of their semivariograms using the kriging technique. Surface three dimensional maps of these were produced using the Surfer software

3. Result and discussion

3.1. Topography, land uses and soil types

The Sumani Watershed had five sub watershed. The eastern part have Lembang and Aripan subwatershed that comprise 45% % of the watershed area. The western part have Sumani, Gawan and Imang subwatershed that comprise 55%of the watershed area. Base on latitude, Sumani watershed can be separated into two topographic unit that is lowland areas (<500>500 m asl). In the central to northern part, lowland areas occur with little slope (< 10%) and comprise 24% of the Sumani Watershed and around lowland areas occur upland areas and comprise 76% of the Watershed areas. Five major river, ie. Lembang, Sumani, Bagawan, Ujung karang and Barus rivers flow in the watershed and finally the river water drain into Lake Singkarak. (figure 1).
Sumani watershed consist of various land uses, including primary forest, mix farming, upland crop land, sawah and settlement. Relative flat areas with less than 10% of slope covering 24% of total watershed area are mostly in the lower elevation (<500>500 masl) with slope 10 – 30% (40% of the area) commonly planted as Agroforestry like mixed garden (Carica papaya,sp; Arachis hipogea.sp; Avogado sp; Musa indica.sp; Cinnamomun sp; Durio sp; Coconut, and vegetables field combine with cinnamons and passion fruit) and Sawah rice terrace. Distribution vegetation on subwatershed shows on table 1 and table 3. Widianto et al. (2003) stated that Agroforestry had along time practiced by farmer in Indonesia. Agroforestry divided as 1. Classical Agroforestry that found in landscape of villages agro ecosystem where was local people planted tree with arable land and poultry and 2. modern Agroforestry than popular as tumpangsari in Indonesia. Gouyon et al. (1993) stated that mixed Rubber in Jambi Sumatra is Agroforestry rubber. Agroforestry at wet tropical areas is used for name and identity for land use manajement which was perennial tee (shrub, bamboo, tree etc) and arable land or poultry tree planted together in the same areas and time (Foresta et al 1991; Michan et al 1992). Michan et al (1986) found that mixed farming, popular calsed as Parak in Maninjau, West Sumatra is Agroforest field.
In foothills of Mt. Talang, agriculture lands like mixed garden, vegetables are found in this slope class below 1000 m asl. In the higher elevation in Bukit Barisan (>1000 m asl), forest dominate this slope class. Combination of steep slopes more than 30% up to 100% appears as dissected plateau in the west side of the basin. These vary steep areas are covered by natural vegetation like forest, shrubs and grass, also patches of less intensive agricultures like mixed garden (Farida et al. 2005).
These very steep areas are covered by natural vegetation like forest (Dipterocarpaceae), shrubs and grass (Imperata cylindrical; Tithonia diversifolia sp; Mimosa invisa ), also patches of less intensive agriculture, like Agroforestry. Soil types indentified include six family ie. Aeric Tropaquept, Typic Kandiudult, Typic Distropept, Oxic Hapludand , Typic Eutropept and Andic Humitropept (Saidi 1995). The Andisol occurred in inland areas of Lembang subwatershed in around Lake Dibawah. The Ultisol occurred in inland areas of Aripan subwatershed. In contrast, most Inceptisol found in Sumani, Gawan, Imang , Aripan and Lembang subwatershed in Uplan and lowland areas (Table 1).

3.2. Estimation Erosion and Sediment yield

Both the USLE model in Surfer software and the observed Sediment Delivery methods (SDR model) were able to measure erosion and sediment yields in Sumani Watershed.
Table 1 shows soil physico chemical analysis measuring erodibility of soil used as the input data for USLE model. This result indicated that land uses different have various soil characteristic and erosion hazards.
Erosion rates in 1992 – 1993 calculated by USLE in Surfer (46.64 ton ha-1y-1) were similar at the rate calculated by Saidi (1995) using observed Sediment Delivery (SDR) method (49.25 ton ha-1y-1) while for subwatershed (SW) the USLE model in Surfer over estimates erosion in Sumani-SW(34%), Lembang-SW(67%), Gawan-SW(81%), Aripan-SW(71%) and Imang-SW(74%) as compared with SDR model (Table 2). Ideally, land uses maps are similar based on year of data “R” value. Unfortunately, only data Landsat ETM 2002 were available. Moehansyah (2004) reported that USLE higher measure erosion rate compare than ANSWER and AUSLE in 0.4 ha plot in Riam Kanan watershed, Indonesia. Kusumandari (1997) stated that USLE higher 50% measure erosion rate compare than AGNPS model in Citarik subwatershed, Indonesia. In suggestion USLE model is reliable to measure erosion rate in Indonesia.Wischmeier and Smith (1978) stated that Among six factor of USLE, C factor is probably the greatest source of error for the models. Wu et al.(1993) reported that the error in the prediction of soil loss due to raindrops is related to antecedent moisture and crop cover. The higher values of predicted soil loss USLE models than those SDR model in subwatershed in the study might be related to large area of Sub watershed and appropriate value of C factor . Nearing (1998) reported that evaluation of various soil erosion models with large data sets have consistently shown that these models trend to over-predict soil erosion for small measured values, and under-predict soil erosion for larger measured values. The USLE was designed only to predict long-term, average annual soil loss. Tonitt et al. (2007) stated that use of maximum and minimum values can be implemented to obtain different threshold value.

In other fact, this different in the soil erosion rates at subwatershed was measured by USLE in Surfer and SDR models are mostly due to the distribution in large of Sawah area in subwatershed (Table 3). Table 4 shown that large size Sawah was affected with significant negative correlation erosion with USLE and balance erosion, while mixed farming (Agroforestry) and Settlement were significant in positive correlation with erosion USLE and balance erosion. This result bears testimony to the fact that erosion product in upland area which were dominated with Agroforestry area flow to lowland and it is not all the erosion product that flow to the outlet of the river as sediment yield but some part of erosion from upland is deposited in lowland at subwatershed in located sawah area because sawah are traditionally terraced. When it was measured sediment delivery in outlet of was observed to be in low quantity. This result indicated that USLE run in Surfer software was suitable in Sumani watershed to investigate erosion in different land uses and also might be predicted deposition areas..
Ni et al. (2007) reported that terrace stop the downslope transport of soil, so the soil accumulates upslope of boundary, and erodes downslope of the boundary. Terracing, an effective method of soil conservation on steep slopes, has been used extensively to control water erosion in hilly area. Farmers have dissected entire hillslope into a number of slope segment, i.e. terracing, for the sake of minimizing soil loss and for the convenience of field management operation (Zhang et al.2004b).

3.3. Spatial distribution erosion and sediment yield in Sawah, Agroforest field and Forest.

This evidence was found in Figure 2 that from simulated Soil redistributed was found as soil loss 94 % and soil deposited 6% of watershed areas. Soil loss occurred in upland areas classified as very low to low at sawah and forest areas and moderate to extreme severe found at Agroforestry areas. From 6% soil deposited in Sumani watershed that spread 67% in upland areas at sawah rice terrace and down slope forest areas and 33% in lowland areas at sawah rice terrace. Deposites areas was identified by erosion minus 50 t/ha/yr which is distributed in subwatershed (Lembang-SW, Sumani-SW, Aripan-SW, Gawan-SW and Imang-SW). Other researches reported that observations show that sediment yield from watershed are often about an order of magnitude lower than the soil erosion rates measured from hillslope plots (Edwards 1993) and is deposited (Hua lu et al. 2006). Lim et al. (2005) reported that A sediment reduction ratio of 50%, indicating that half of the sediment retention basin and the rest of the sediment leaves the sediment retention basin to downstream areas. Estimated erosion in 1996 to 2007 was higher than August 1992 to July 1993 because of rainfall change (Table 1).
The erosion rate from 1996 to 2007 of Sumani watershed was 48.38 ton ha-1y-1 with average bulk density (0.9426 g cm-3) is equal to 5.133 mm yr-1 of soil washed away annually and for ten years it was erosion 51.3 mm. This data connected after compared to actual changes in the field, as shown by tree root, sheet erosion when field survey in September 2007. as a result, we concluded that the USLE model in Surfer provides realistic result so that we can identified specific land use with distribution erosion map.
Table 2 and Table 3 shows that the sediment yield in 1992 – 1993 of Sumani Watershed was 264439.08 ton y-1 compared with USLE in Surfer higher 90% (2720511.2 ton y-1). This result indicated that half of sediment yield was to be deposited and accumulated and we need to know the possible land uses where deposition took places.
Table 5 shows that erosion rates between 1996 – 2007 both in Sawah and Forest quite similar (0.94 ranging 0.003 – 11.8 ton ha-1y-1) and with a (0.57 ranging between 0 and 4.05 ton ha-1y-1) respectively, while erosion rates in Agroforestry was 102.6 (with a range 0.047 – 893.1) ton ha-1y-1 which was higher compared to forest area and it was much higher than the permissible erosion rate (10 to 14 ton ha-1y-1) as set by the Indonesian government (Kusumandari et al. 2000). This result shows that Sawah area is best suited to minimize erosion compared to Agroforestry. Thus, Sawah areas might be important land use to be conservation measures for Sumani watershed for a long periods.

Table 6 shows that average R value in 1996 – 2007 determined by USLE method estimated sediment yield in the Agroforestry to be (315610 ton y-1). Settlement was 75775.5 ton y-1 and Grass was 13371.1 ton y-1. These results are higher compared to sediment yield at Sawah (2133.2 ton y-1), Forest (1317.2 ton y-1) and shrub (1317.2 ton y-1). This shows that Sawah can function in similar way a forest in conservation of Watershed while Agroforestry , Settlement and grass areas should be put under vegetative and mechanical conservation technique to minimize erosion.

Figure 2a and 2b indicate that the erosion rate in Agroforestry areas is closely associated with the rainfall erosivity while Sawah and Forest areas were not associated with rainfall erosivity. It mean that when rainfall erosivity increases, erosion rate in the Sawah and the Forest areas is stable while in Agroforestry area it increases the erosion.
Kusumandari et al. (1995) stated that the erosion rate in the forested area is negligible and stable, although different rainfall erosivity is applied. In this context, a forest is a natural forest with a multilayered canopy, dense ground cover, and in an undistributed condition. Including decrease in the volume of runoff brought about by the interception of rainfall at the canopy, higher rates of infiltration associated with better aggregated soil, and the opening up of macrospores in the soil by root growth.All helps to reduce erosion to neglible rates. However, the dense ground cover in the forest area is the main factor which decreases erosion rate. While, Sawah has traditional bench terraces which makes erosion stable although different rainfall erosivity exist. Terraces, as well function in similar way as a with multilayered canopy and dense ground covers in Forest in reducing the volume of runoff and it has the ability to concentrate deposited sediment. Terraces can be a control measure and it minimizes erosion in hilly area as testified by (Hua lu et al. 2006; Zhang et al. 2004b; Ni et al. 2007).

Table 5 and 6 indicates that Agroforestry areas have high variation in erosion (128%) and large interval between minimum and maximum erosion, because not all areas of Agroforestry are identical with high erosion. Specific areas in Agroforestry with high erosion rates can be determined by USLE model in Surfer and can show the distribution of erosion in 3D map as show in (Figure 4 and 5).

To validation of distribution erosion map was make the variogram (Figure 3). The variogram of R, K,LS, C and P factor of USLE fitted to linear model while Erosion (A) fitted to the polynomial model. The variogram of all the RKLSCP factor that fitted to linear model exhibited transitive feature, i.e. their variogram increase up to a point variogram as well as increased out at the maximum lag distance , while Erosion model, their variogram increased up to a point and then fluctuated and flatted out the lag distance. The variogram represent the a priori variance (Oyedele et al. 1998). The transitive nature of the variogram of these USLE factor exerted a similar influence to Soil loss. The max lag Distance is the maximum separation distance to be considerated during variogram model. The Lag Distance indicated of the XY coordinates (Golden Software 1988). The Lag Distance indicates the distance to which the USLE factor (RKLSCP) can be assumed to be related, implying that sample of soil collected to calculated K factor ( Table 1) within distance are more similar or more related to each other than samples collected outside distance in watershed (Webster 1985).
Sonoko et al. (2008) stated, Although plenty of models intended for various scales exist, a perfect model for including all flow does not exist. Predicting the future by using models is becoming more and more important, and developing models that match to the characteristic of a regional scale is highly required.
Figure 5 shows that distribution of erosion in a 3D map of Sumani Watershed when compared with land use pattern in figure 1 was found to have 51% of Agroforestry areas in high erosion hazard areas in classes moderate to extreme severe which it was located generally in upland area of surrounding watershed.
Figure 4 also shows that deposit areas in Sumani watershed (- 50 ton ha-1y-1, minus value indicated deposited of erosion).Generally, 6% of Sumani watershed was deposited areas. 33% was found in Sawah area at lowland and 67% at upland in Sawah rice terrace and down slope of forest areas. In lowland sawah areas, deposits of erosion was concentrated generally at Sawah in Imang-SW>Gawan-SW>Lembang-SW>Aripan-SW>Sumani-SW. In Upland sawah areas deposited was founded as Sumani-SW>Lembang –SW and Upland down slope Forest deposited founded as Aripan-SW>Gawan-SW>Imang-SW. To know specific areas with coordinate position where deposited concentrated can be digitize the map which was created.

Conclusion

In surrounding watershed, soil loss a class of moderate to extreme severe had high occurred at upland areas and soil accumulation was present at sawah areas in upland and lowland. The USLE in Surfer measured erosion rate almost similar with SDR model (46.64 and 49.25 ton ha-1y-1). However, In Subwatershed, the USLE in Surfer measured more than half of the erosion rate predicted by SDR models. The Sediment yield in Sumani watershed was 245587.5 ton y-1 in 1992-1993 rainfall and , 259808.84 ton y-1 in 1996-2007 rainfall using USLE in surfer. SDR models measure sediment yield in 1992-1993 was 264439.07 ton y-1. It is clear that combination USLE model in surfer and SDR models, its accuracy and suitable to predict both soil loss and sediment yield in Sumani watershed. Also, USLE model in Surfer can be calculated erosion and sediment yield in Sawah, Agroforestry and Forest.
The erosion rates of Sawah area (146.68 km2) were 0.91 and 0.94 ton ha-1y-1, Agroforestry areas (232.74 km2) were 98.94 and 102.62 ton ha-1y-1 and Forest (114.39 km2) were 0.55 and 0.57 ton ha-1y-1 in rainfall 1992-1993 and 1996-2007.
This fact suggest that Sawah area and Forest surrounding Sumani watershed is contributes greatly in minimizing soil erosion and were not associated with rainfall erosivity affect while 51% of Agroforestry has erosion hazard in class from moderate to extreme moderate and it need soil conservation action because Agroforestry erosion rate was from 7 to 63 times grater than the permissible erosion rate (10 to 14 ton ha-1y-1) determined by the Government of Indonesia. It is suggested to planning appropriate conservation of Sumani watershed can be estimated from single numerical values as cover and management factor (CP). CP = Tolerable erosion (T)/ RxKxLxS. Base on overlay land uses and Distribution erosion in 3D map, it is clear that Sawah terrace in lowland and upland areas to be a place whereas water erosion result has been accumulated and need to investigate affect accumulation nutrient affected by erosion result in Sawah area.



Acknowledgements

My deepest acknowledgements should go to Ministry of Education, Science, Sport and Culture of Japan for the financial assistance in this study. More appreciation goes to Polytechnic of Agriculture Payakumbuh and Soil department of Andalas University in Indonesia for providing the necessary support during soils sampling.

References

Abdurachman A, Abuyamin S, and Kurnia U 1990: Pengelolaan Tanah and Tanaman untuk Usaha Konservasi (Soil and Crop Management to Conservation), PPT Bogor (in Indonesian).
Affandi, O 2000. Mempromosikan Kembali Kearifan Lingkungan Tradisional dalam Konservasi Sumberdaya Alam. Buletin Leuser Edisi Vol. 3 No. 6 Maret 2000.
Ahmadi SH, Seifollah Amin, Ali Reza Keshavarzi and Naser Mirzamostafa 2006: Simulating Watershed outlet sediment Concentration using the ANSWERS Model by applying Two Sediment Transport Capacity Equations. Biosystems Engineering, 94(4), 615-626.
Badan Meteorologi and Geofisika Sicincin 2007: Data curah hujan di Stasiun Klimatologi Kelas II Sicincin (Rainfall data in Climatology Class II Sicincin), Padang (in Indonesian).
Bols P C 2000: The Iso-Erodent Map of Java and Madura. Bogor, Indonesia, Belgian Technical Assistant Project ATA 105, Soil Research Institute, Bogor, Indoesia.
Chris SR and Hon Harbor 2002: Soil erosion assessment tools from point to regional scales-the role of geomorphologists in land management research and implication. Geomorphology, 47, 189-209.
Ciesiolka C A, Coughlan KJ, Rose CW, Escalante MC, Mohd. Hashim G, Paningbatan Jr EP and Sombatpanit S 1995: Methodology for a multi-country study of soil erosion management. Soil Technology, 8, 179-192.
Boodt DM 1967 : water saturated permeability determination on undisturbed soil samples, eoropean methods for Soil Structure determination, V106-V107.
Jong DSM, Paracchini ML, Bertolo F, Folving S, Megier J and De Roo APJ 1999:Regional assessment of soil erosion using the distributed model SEMMED and remotely sensed data. Catena, 37, 291-308.
Departemen Pertanian Badan Penelitian dan Pengembangan Pertanian 1979: Penuntun Analisa Fisika Tanah (Soil Physic analysis Manual), Lembaga Penelitian Tanah, Bogor (in Indonesian).
Edwards K 1993: Soil erosion and conservation in Australia In:Pimentel D (eds), World Soil Erosion and Conservation, Cambridge, pp 147-169.
Amore E, Carlo Modica, Mark A. Nearing and Vincenza CS 2004: Scale effect in USLE and WEPP application for soil erosion computation from three Sicilian basins. Journal of Hydrology, 293, 100-114.
Farida, Kevin Jeanes, Dian Kurniasari, Atiek Widayati, Andree Ekadinata, Danan prasetyo Hadi, Laxman Joshi, Deshi Suyamto and Meine van Noordwijk 2005 : Rapid hydrological appraisal (RHA) of Singkarak Lake in the context of Rewarding Upland Poor for Environmental services (RUPES).ICRAF South Asia. Bogor, Indonesia, Working Paper 1-76.
Foresta D, Michon HG 1991 : Agroforesteries Indonésiennes: systèmes et approches. Makalah (bahasa Prancis) disampaikan pada seminar internasional “Quelles agroforesteries pour l'Orstom?”, Paris, Oktober 1991.(Translation in Indonesia)
Getis A, J Getis and J Fellmann 1991. Introduction to geography. Third Edition, Wm. C. Brown Publishers, USA.
Golden Software 1988. SUERFER User’s Manual. Golden Software Incorporation.
Goovaerts P 2000: Geostatistical in soil science: state-of-the-art and perspectives. Geoderma, 89, 1-45.
Granados LF, Jurado EM, Atenciano S, GarciaFA, Orden MS and Tores GC 2002: Spatial variability of agricultural soil parameters in soutern Spain.Plant Soil, 246, 97-105.
Lu H, Moran CJ and Ian PP 2006: Modelling sediment delivery ratio over the Murray Darling Basin. Environmental Modelling & Software, 21, 1297-1308.
Knudsen HP, Lefohn AS 1988. The use of geostatistics to characterize regional ozone exposures In Heck WW, Taylor OC, Tingey DT eds. Assessment of crop loss from air pollutants. London, UK, Elsevier Science publishers, LTD, pp, 91-105.
Kusumandari A and Bruce Mitchell 1997: Soil erosion and sediment yield in forest and agroforestry areas in West Java, Indonesia. J. Soil and Water Cons, 52(4), 376-380
Kusumandari A. 1995. Factor Influencing Soil Erosion Rate in Citarik Subwatershed, West Java. Bulletin Fakultas Keutanan UGM. Faculty of Forestry, Gadjah Mada University, Yogyakarta, Indonesia. 2(6):33-45.
Lal R 1990. Soil erosion in the tropics” Principles and Management. Mc. Graw Hill, USA.
Lane SN, Reid SC, Tayefi V, Yu D and Hardy RJ 2004: Reconceptualising coarse sediment delivery problems in rivers as catchments-scale and diffuse . Geomorphology, xx, xxx-xxx.
Lefohn AS, Knudsen HP, Logan JA, Simpson J and Bhumralkar C 1987: An evaluation of the kriging method to predict 7-h seasonal mean ozone concentration for estimating crop losses. J. Air Pollut. Control Assoc,37, 595-602.
Liu BY, Nearing MA, SHI PJ and Jia ZW 2000: Slope Length Effects on Soil Loss for Steep Slopes. Soil Sci. Soc. Am. J, 64, 1759-1763.
Metternicht G, Gonzalez S 2005: FUERO: foundations of a fuzzy exploratory model for soil erosion hazard prediction. Environmental Modeling & Software, 20, 715-728.
Michon, G., Mary, F. and Bompard J.M. 1986 : Multistoried agroforestry garden system in West Sumatra, Indonesia. Multistoried agroforestry garden system in West Sumatra, Indonesia. Agroforestry Systems, 4:,315-338.
Michon, G and Foresta HD 1992 :Complex agroforestry systems and conservation of biological diversity 1/Agroforestry in Indonesia: a link betweentwo worlds. in YS Kheong and LS Win (eds.) In Harmony with Nature. An International Conference on the Conservation of Tropical Biodiversity,Kuala Lumpur, Malaysia, The Malayan Nature Journal. Golden Jubilee issue. 457-473
Moehansyah H, Maheswari BL and Armstrong J 2004: Field Evaluation of Select Erosion for Catchment Management in Indonesia. Biosystems Engineering, 88(4), 491-506.
Morgon RPC 1999:A simple approach to soil loss prediction : a revised Morgon-Morgon-Finney model. Catena, 44, 305-322.
Morgon RPC 1985: Soil Erosion and Conservation. Longman, London, pp, 127.
Morgon RPC, Morgon DDV and HJ Finney 1982. Stability of agricultural ecosystem: documentation of a simple model for soil erosion assessment. Int. Inst. Appl. Syst. Anal. Collaborative Paper CP-82-50.
Morgon RPC, Quinton JN, Smith RE, Govers G, Poesen JWA, Auerswald K, Chisci G, Torri D and Styczen ME 1998: The European Soil Erosion model (EUROSEM)” A Dynamic Approach for Predicting Sediment Transport from fields and Small Catchments. Earth Surf. Process. Landform, 23, 527-544.
Nearing MA 1998 : Why soil erosion models over-predict small soil losses and under-predict large soil losses. Catena, 32, 15-22.
Ni S J and J H Zhang 2007. Variation of chemical properties as affected y soil erosion on hillslopes and terrace. European Journal of Soil Science,58, 1285-1292.
Wu Q and Wang M2007: A framework for risk assessment on soil erosion by water using an integrated and systematic approach. Journal of Hydrology, 337, 11-21.
Reagan J 1984: Air quality interpretation In National Crop Loss Assesment Network (NCLAN) 1982 annual report. Corvallis, OR: U.S. Environmental Protection Agency; EPA-600/3-84-049, pp. 198-219.
Renard G. Kenneth and Jeremy R. Freimund 1994: Using monthly precipitation data to estimate the R-factor in the revised USLE. Journal of Hydrology, 157, 287-306.
Roehl JE 1962: Sediment sources areas, and delivery ratios influencing morphological factors. International Association of Hydrological Science, 59, 202-213.
Saidi A 1995: The Affecting Factors of Runoff, Sedimentation and their Impacts on Land Degradation in Sumani Watershed Solok West Sumatera.Disertasi Doctoral in Padjadjaran University, Bandung (in Indonesian).
Schmidt J, Werner Mv and Michael A 1999: Application of the EROSION 3D model to the CATSOP watershed, The Netherlands. Catena, 37, 449-456.
Scientific Software Group 2008. Environmental & Water Resources Software & Equipment. www.scientificsoftwaregroup.com
Simon CR, Stuart NL, David RM and Christopher JB 2007: Does hydrological connectivity improve modeling of coarse sediment delivery in upland environment. Geomorphology, 90, 263-282.
Soil Survey Staff 1993a: National Soil Survey Handbook.Title 430 VI. (Washington, DC:USDA)
Soil Survey Staff 1993b: Soil Survey Manual. Handbook no 18. (Washington, DC:USDA)
Takata Y, Funakawa S, Akshalov K, ishida N, Kosaki T 2007b. Spatial prediction of soil organic matter in northern Kazakhstan based on topographic and vegetation information. Soil Sci. Plant Nutr., 53, 289-299.
Takken I, Beuselinck L, Nachtergaele J, Govers G, Poesen J and Degraer G 1999:Spatial evaluation of physically-based distributed erosion model (LISEM). Catena, 37, 431-447.
Tonitto C, David MB, Drinkwater LE, Li CS 2007: Application of the DNDC model to tile-drained Illinois agroecosystems: model calibration, validation, and uncertainty analysis. Nutr. Cycl. Agroecosys., 78, 51-63.
Walling DE, He Q and Whelan PA 2003: Using 137 Cs measurement to validate the application of the AGNPS and ANSWERS erosion and sediment yield models in two small Devon catchments. Soil & Tillage Research, 69, 27-43.
Wakatsuki T, Shinmura Y, Otoo E, Olaniyan GO 1998:African-based sawah system for the integrated watershedmanagement of the small inland valleys of West Africa.FAO Water Report No. 17, pp. 56–79.
Wischmeier WH and Smith DD 1978: Predicting rainfall erosion losses: a guide to conservation farming, USDA Handbook: No. 537 US DEpartement of Agriculture, Washingon, DC pp 1-58.
Yamamoto JK 2005: Comparing ordinary kriging interpolation variance and indicator kriging conditional variance for assessing uncertainties at unsampled location In Application of computers and operations research in the mineral industry-Dessureault, Ganguli Kecojevic & Dwyer editorial, Balkema.
Young RA, C A Onstad, D D Bosch and W P Anderson 1989. AGNPS. Agriculture non-point Source Pollution Model. A watershed Analysis Tool. USDA, Conservation Research Report. No. 35.
Zhang JH, Lobb DA, Li Y and Liu GC 2004b. Assesment of tillage translocation and tillage erosion by hoeing on the steep land in hilly areas of Sichuan, China. Soil Tillage and Research, 75, 99-107.