Determination of soil erosion parameters for Malaysian conditions using remote sensing and geographic information system approach

Abstract

The use of Universal Soil Loss Equation (USLE) for predicting an average annual soil loss in any area is influenced by the characteristics of the environmental setting of that specific area. In the previous studies that have been carried out in Malaysia, the parameters used were based on various studies in many other areas, and this affected the accuracy of the result. This study is carried out in Pasoh Forest Reserve which covers an area of 8,100 km2 located in Pahang, Negeri Sembilan and Selangor. The elevation of the study area range from 40 meters to 1760 meters. Permanent crop constitutes the largest land cover of the study area (46 %), followed by forest (45%), cropland (5 %), and settlement and non-agriculture lands (4 %). The accuracy of the results was tested using regression analysis and Root Mean Square Error (RMSE). Notably, the results of this study shows strong correlation between factors that were generate and the RMSE values are smaller. It is noted that crop management (C,) slope length and steepness (LS) and soil erodibility (K) factors that have been applied for USLE model in this study gave better results compared to other previous studies using inapplicable environmental characteristics. Estimation of soil loss in the year 1995 showed an average of 7.97 t/h/yr and the average soil loss for the year 2003 was 6.83 t/h/yr, a decline of 15 percent. Secondary forest, oil palm, sundry non tree cultivation and rubber have contributed to the highest amount of soil loss for both years. Error estimation for the year 1995 shows that annual rate of soil loss is 18.93± 2.37 t/yr compared to 16.59 ± 1.13 t/yr for the year 2003, apart from experiencing loss of soil surface ranging from 4 mm to 8 mm in a period of 8 years. Thus, the approach used in this study is not only suitable for Malaysia, but can also be applied to any other country with similar environmental characteristics.