Synergistic application of fuzzy logic and geo-informatics for landslide vulnerability zonation-a case study in Sikkim Himalayas, India

The present study comprises the application of various Fuzzy algebraic functions for assessment of landslide vulnerability in Rumtek-Samdung area of Sikkim, which is a Himalayan state in India. The thematic layers for probable causative factors for landslides were collected from various spatial data sources. The main causative factors identified include pedologic factors viz. soil depth, soil texture, soil drainage behavior, soil stoniness, soil erosion and soil hydraulic conductivity, geologic factors such as lithology and foliation along with land use, slope, existence of road and drainage, etc. Landslide spots for the past few years were identified and detected using Cartosat panchromatic image of 2.5 m resolution and further augmented with Google image and field verification. A relation between the occurrence of landslides and each sub-category of the causative factors was established through their frequency ratio and converted to fuzzy membership values. In addition to the five Fuzzy operators used in previous researches, we have introduced two new operators named as Fuzzy AVERAGE(AR) which is computed as the average of AND, and OR operators, and Fuzzy AVERAGE(SP) which is computed as the average of Fuzzy algebraic SUM and Fuzzy algebraic PRODUCT operators. Landslide Susceptibility Index values were computed employing various Fuzzy operators based on which zonation maps were generated classifying the study area into five zones such as least vulnerable zone, low vulnerable zone, moderate vulnerable zone, highly vulnerable zone and most vulnerable zone. Vulnerability assessment accuracy was then computed based on the occurrences of past landslides in the higher three vulnerability zones. The FuzzyGAMMA Operator exhibited the highest vulnerability assessment accuracy at Lambda=0.5. The two newly formulated Fuzzy operators showed significant improvement in the vulnerability assessment accuracy in comparison to the two conventional Fuzzy operators' namely Fuzzy algebraic PRODUCT and Fuzzy algebraic AND operator. Performance Index for the Fuzzy Operators is computed as the ratio of vulnerability assessment accuracy to the percentage of area in the higher three vulnerability zones. All the three newly formulated Fuzzy Average operators exhibited higher Performance Index compared to Fuzzy OR-and Fuzzy SUM-based operators indicating higher proficiency in landslide vulnerability assessment