LAUSR

Abstract. Initial assessment of landslide prone area is important in designing landslide mitigation measures. This study, a part of our study on developing landslide spatial model, presents initial signal of landslide prone area. In here, we use soil depth to hardpan to assess landslide prone area in Western Central Java, a relatively small region where 23 % of Indonesian landslide occurs. To this end, we interpolated soil depth to hardpan in a regular grid from irregularly distributed data. To do this, we employed three different methods: Inverse Distance Weighting (IDW), Ordinary Kriging (OK) and Co-Kriging (CK). For the latter, we experimented with several potential covariates. To determine the best fitting model, several tests on model performance and its corresponding errors were done. Error measures used in this study are Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE), while statistical measures employed are Standard Deviation, Variance, Interquartile Range (IQR), Mean Absolute Deviation and Median Absolute Deviation. The result shows that CK with covariate of slope and soil cohesion is the best fitting model and exhibits clear pattern related to recorded landslide disaster sites. We found that 64 % of landslide disaster events occur in the area having soil depth to hardpan of 5–10 m. Moreover, 84 % of landslide occurrences happen in regions where soil depth to hardpan ranges from 5 to 15 m. Hence, we suggest that landslide prone area is an area possessing soil depth to hardpan of 5–15 m. This finding is advantageous for policy makers in planning and designing efforts for landslide mitigation.