ANALISIS MULTIKRITERIA DAN REGRESI LOGISTIK TERHADAP KERAWANAN BANJIR DI KECAMATAN KOTO XI TARUSAN, KABUPATEN PESISIR SELATAN

Abstract

The objectives of this research are: (1) Mapping flood vulnerability in Koto XI Tarusan Sub-district, Pesisir Selatan Regency. (2) To determine the most influential factors in determining the level of flood vulnerability in Koto XI Tarusan Sub-district, Pesisir Selatan Regency. This type of research is quantitative research with descriptive, and inferential approaches. Using a combined method consisting of three stages of analysis. Firstly, Analytic Hierarchy Process (AHP) was used to determine the priority importance of each parameter based on expert judgement. Second, Binary Logistic Regression was applied to build the most stable and significant statistical prediction model. Thirdly, bivariate analyses (Chi-Square Test and Independent Sample T-Test) were applied to test individual relationships between variables. The six main parameters were rainfall, elevation, slope, soil type, distance from river and land cover. Validation of the regression model was conducted with a train-test data split of 80%-20%. The results showed that (1) The results of the AHP method flood vulnerability mapping are classified into three levels of vulnerability, namely low with an area of 22,584.94 hectares (51%), medium with an area of 18,704.34 hectares (42%), and high with an area of 2,987.69 hectares (7%). Mapping binary logistic regression method classified into three levels of vulnerability, namely low with an area of 31,054.538 hectares (70%), moderate with an area of 2,644.558 hectares (6%), and high with an area of 10,649, 199 hectares (24%), concentrated along the alluvial plain of Tarusan watershed (2) Rainfall (priority, 0.38) and Elevation (priority, 0.23) are the most influential factors according to AHP. Binary logistic regression model of significant variables showed Distance from River (p = 0.002) and Slope (p = 0.015) as the best model with 100% validation accuracy. Bivariate analysis proved that the factors of soil type, land cover and elevation had very strong individual relationships (p < 0.05) with flood occurrence. In conclusion, triggering factors (hydrometeorology) and physical vulnerability (geomorphology and land cover) jointly determine the level of flood vulnerability.