Integrating land cover dynamics and projections to assess surface drought susceptibility using frequency ratio and cellular automata in the Takkalasi Watershed

Abstract

Land cover degradation from increasing anthropogenic pressure is a key driver of rising surface drought susceptibility in watersheds, yet the spatial relationship between land cover dynamics and drought susceptibility distribution remains understudied in an integrated framework.This study examines historical land cover change in the Takkalasi Watershed, Barru Regency, South Sulawesi, from 2016 to 2025; projects land cover patterns through 2045; and maps surface drought susceptibility using an integrated approach.Land cover change was analyzed through supervised classification of Landsat 8 imagery, while the 2045 projection was generated using a Cellular Automata model in LanduseSim.Surface drought susceptibility was modeled using the frequency ratio (FR) method based on the drought susceptibility index (DSI), incorporating LST, NDVI, NDBI, precipitation, slope, soil texture, distance to river, hotspots, and land cover.Model validation used a 70:30 data split, with AUC-ROC evaluation conducted in SPSS.Results show that rice paddy and settlement classes experienced the greatest historical area increases, while dryland agriculture and shrubland declined most significantly.Projections for 2045 indicate a dramatic expansion of residential areas in the middle and lower watershed reaches.FR analysis identified building density, residential land cover, and rice paddies as the factors most strongly associated with elevated drought susceptibility.Approximately 28.63% of the watershed falls into the high-to-very-high susceptibility category, concentrated in the middle and lower reaches.The FR model demonstrated strong performance, with an AUC Success Rate of 0.807 and a Prediction Rate of 0.727.Rather than positioning FR-CA integration as a methodological innovation, this study frames its contribution as a reproducible, integrated assessment workflow linking historical land-cover dynamics, cellular automata-based projection, and drought susceptibility mapping for watershed-scale risk planning.The findings indicate that anthropogenic land-cover change is a key ecological pathway increasing surface drought susceptibility in the Takkalasi Watershed, providing a basis for ecological engineering interventions riparian and forest restoration, agroforestry, and infiltration enhancement and for local-level climate adaptation planning.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint