Search

Japan International Cooperation Agency (JICA) and National Building Research Organisation (NBRO) have been conducting projects to prevent and minimize landslide damages. Landslide early warning is one of the targets of the JICA-NBRO projects based on the analysis of rainfall indices. The analysis of these indices, such as working rainfall and Soil Water Index (SWI), was carried out. The Soil Water Index is an output of conceptual hydrological model representing soil water content. Time series of observed rainfall at 25 gauging stations from 2014 to 2020 were utilized to calculate rainfall indices. Moreover, past landslide records were utilized to analyze the correlation between landslide occurrences and the calculated rainfall indices. Results showed that most of the past landslides were caused by severe rainfall events in which Soil Water Index exceeded 112. Results also showed that small-scale slope failures were caused by relatively minor rainfall events. The critical values of rainfall indices causing landslides depend on the regional characteristics. The critical value in the south-western region tends to be higher than in the northern and southern regions. It seems that the higher rainfall in the south-western region increases the critical rainfall value of landslide occurrence. There is a possibility to improve the accuracy of landslide early warning by using the rainfall indices considering regional characteristics.

Fourteen districts were identified as landslide prone in Sri Lanka. These districts fall in the wet and intermediate climatological zones where annual rainfall is over 2000 mm. Short term heavy rainfall and long term cumulative rainfalls are active factors for landslide occurrences. In this study, to enhance the landslide prediction capacity and accuracy, two indices were analyzed; Short Term Rainfall Index—1.5 h. half period Working Rainfall and Long-Term Rainfall Index—Soil Water Index. Also, regional critical Soil Water Index values were evaluated for the analysis of regional geological and other parametric impact on landslides. Hydrological Tank Model simulations were used to calculate Soil Water Index and to predict landslides with snake curves and probability curves. In regional aspect, tank model parametric analysis was carried in determination of suitable parameter set for each soil conditions. Rainfall intensity acts as the main contributing factor for Soil Water Index. With probability curves, better predictions were obtained on regional based soil water index of greater than 100 values to identify the landslide occurrences even though the Short-Term rainfall did not exceed 10 mm. Region-wise Critical Soil Water Indexes were obtained by critical lines with soil parameters derived for Sri Lankan soil conditions. Better practices on landslide early warnings could be performed through these derived regional long-term Soil Water indexes. As a future development, it is expected to develop warning levels based on these critical values.