Developing an Erosion risk map for Langkawi Island
Assuming the worst case scenario, i.e. having bare land with no conservation measures applied. The C and P value in the USLE were taken as 1.0. Using the IDRISI function OVERLAY, each thematic layer of the USLE was overlaid over the respective map layers. The output map shows the distribution of soil erosion assessment for the entire Island (Fig. 3)


Fig. 3 The spatial distribution of soil erosion category on the Langkawi Island
The map was then classified into 3 erosion risk classes. These are: less than 5 kg/m² /yr as low, 5 to 15 kg/m² /yr as moderate and above 15 kg/m² /yr as high. The outcome represents a potential erosion risk map with the resolution of 30 m per pixel of Langkawi Island (Fig. 4).


Fig. 4. A potential erosion risk map for the Langkawi Island based on a worst case scenario
Evaluating the Erosion risk map
The produced erosion risk map (Fig.4), was assessed for accuracy by direct comparison of classes, based on geographical location with a friction map of the Langkawi Island, with resolution of 1.7 km² per pixel, produced from fieldwork in 1995 (Sulaiman et. al. 1996). A total of 95 samples were located on both maps. The overall accuracy achieved was 74%. Individual class accuracy ranged from 62% for high, 72% for moderate and 75% for low erosion risk (Table 1). These figures have to be approached cautiously due to the variation in the spatial resolution between the maps. Nevertheless, they can act as general indicators for the correspondence between a map generated based on field work and Fig 4.
Table 1

	remotely sensed data	class1	class 2	class 3	total pixels	% accuracy
ground surveyed data
class 1 (low erosion rate)		45	5	8	58	78
class 2 (moderate erosion rate)		1	12	3	16	75
class 3 (high erosion rate)		0	8	13	21	62
			total		95

sum diagonals is 45 + 12 + 13 = 70, total number of pixels = 95
overall accuracy is (70/95)*100 =74%

It is evident from the topography map (Fig.2) and the erosion risk map (Fig. 4) that high and moderate erosion risk areas are concentrated mainly in the highlands, that is in the central, eastern and northwest region of the Island. Whereas the low erosion risk are in the lowland areas of western and northeast region. This shows that the slope length and steepness of the highlands have influenced greatly the slope length and steepness factor (LS) in the USLE for the erosion study (Morgan, 1986). It is quite difficult to label the class for moderate and high erosion risk in certain part of the Island (Fig. 4) due to the mixed class pixels due to the variation in the pixel size between the maps. Based on visual interpretation, the class with dominant area covered was assigned. This leads to the error in the overall accuracy.

Conclusion
An erosion risk map for the Langkawi lsland was produced using the USLE, remote sensing and GIS. This map was quantitatively compared to a friction map of the Island produced based fieldwork data in 1995. The overall correspondence was 74%, whilst individual class similarities were 62% , 72% and 75% for high, moderate and low erosion risk classes respectively. Major factors influencing soil erosion in Langkawi Island were the slope length (L) and slope steepness (S). Thus, most of the Langkawi Island highland areas are potentially subject to very high predicted soil losses. This study demonstrates the effectiveness remote sensing and GIS in generating essential quantitative information on soil erosion. The outcome for this type of studies represents a valuable resource for decision makers to guard against land acquisition in high erosion risk areas or to issue conditional permits with conservation measures to future development projects in moderate erosion risk areas on the Langkawi Island.

References

• Baban S. M. J., and Kamaruzaman W.Y. 1999. Mapping land use/cover distribution in a Mountainous Tropical Island using remote sensing and GIS. International Journal of Remote Sensing (in submission).
• Baharuddin, M. K. and McGuire, G., 1996. Erodibility of Soil and Erosion Risk Map. In Seminar Kajian Hidrologi Pulau Kecil - Pulau Langkawi. Organised by the Department of Irrigation and Drainage, Malaysia. Section I, paper 3.
• Jusoff, K and Chew, S., 1998. Soil Erosion Risk Assessment Using Satellite Data In Genting Highland . In International Symposium Of Arid Region Soil. Izmir, Turkey.
• Kertesz, A., 1993. Application of GIS Methods in Soil Erosion Modellling. Computer Environment and Urban Systems, 17, 233-238.
• Lal, H., Jones, J. W., and Beinroth, F. H., 1990. Regional Agricultural Planning Information and Decision Support System. In Proc. on Application of Geographic Information Systems, Simulation Models, and Knowledge-based Systems for Landuse Management, Virgina Polytechnic Institute and State Uneversity, Blacksburg, 51-60.
• Moore I. D., and Wilson, J. P., 1992. Length-Slope Factors for the Revised Universal Soil Loss Equation: Simplified Method or Estimation. Journal of Soil and Water Conservation, 45(5), 423-428. Morgan, R. P. C., 1986. Soil Erosion And Conservation. Longman Scientific And Technical, England. 298p
• Perez, L. R., Beinroth, F. H., and Jones, J. W., 1993. AEGIS: Agricultural and Environment Geographical Information System. In Proc. 32d Caribbean Food Crop Society, Santo Domingo.
• Samad R., and Abdul Patah, N., 1997. Soil Erosion and Hydrological Study of the Bakun Dam Catchment Area, Sarawak Using Remote Sensing and Geographical Information System (GIS). In Proc. 18th Asian Conference on Remote Sensing, Kuala Lumpur, Malaysia, B-5-1 to B-5-6.
• Shaaban A. J., and Sahat, R., 1996. Hydrological Characteristics and Surface Water Availability of Pulau Langkawi. In Seminar Kajian Hidrologi Pulau Kecil - Pulau Langkawi. Organised by the Department of Irrigation and Drainage, Malaysia. Section 1, paper 1.
• Sulaiman, W. N. A., Yusoff, M. K., Ghazali, A. W., 1996. An Analysis of Soil Erodibility Distribution of Langkawi Island. In Seminar Kajian Hidrologi Pulau Kecil - Pulau Langkawi. Organised by the Department of Irrigation and Drainage, Malaysia. Section 2, paper 6
• Taher, B., 1996. Developing an Environmental GIS Database for Pulau Langkawi. In Seminar Kajian Hidrologi Pulau Kecil- Pulau Langkawi, Section 4, paper 8.
• Wischmeier, W. H. and Smith, D. D., 1978. Predicting Rainfall Erosion Losses - A Guide To Conservation Planning. USDA-SEA Agriculture Handbook 537. U.S Department Of Agriculture, 1-22