STUDY ON ASSESSMENT OF FLOODS & GROUNDWATER SUSCEPTIBLE ZONES IN IDUKKI DISTRICT, KERALA USING GIS BASED APPROACH

Abstract

Floods constitute a recurrent and intensifying hydro-meteorological hazard in the Idukki district of Kerala, driven by the coupled influence of steep topography, high monsoonal rainfall variability, land use transitions, and complex catchment scale hydrological responses. This thesis undertakes a comprehensive assessment of flood susceptibility and its consequent impacts using an integrated geospatial and hydrological modelling framework, providing a scientific basis for understanding their broader environmental significance. The study begins with the development of a flood susceptibility map for the Idukki district using twelve hydrological and geomorphological parameters such as geology, distance from river, land use, Topographic Wetness Index (TWI), elevation, slope, Topographic Roughness Index (TRI), soil, aspect, rainfall, Stream Power Index (SPI), and Sediment Transport Index (STI) within a GIS-based Analytical Hierarchy Process framework. The developed flood susceptibility map was categorized into five susceptibility categories namely very low, low, moderate, high, and very high and these classes occupied the areas of 609.0417 km2, 1222.83 km2, 1180.45 km2, 950.48 km2 and 395.9487 km2 respectively. The analysis also identifies over 30% of the district as highly susceptible, with prominent hotspots in Thodupuzha and central Idukki where low terrain gradients, dense drainage networks, and proximity to major rivers enhance flood generation potential. To characterise subsurface conditions, groundwater potential zones were modelled using a GIS-enabled machine learning approach incorporating AdaBoost, Gradient Boosting, and Random Forest algorithms, capturing the influence of litho-structural features, slope, land use, and other recharge- controlling factors. These groundwater potential outputs were subsequently integrated with the flood susceptibility map to derive a groundwater susceptibility assessment, enabling evaluation of how flood affected areas respond in terms of recharge capacity. The coupled analysis reveals that regions repeatedly subjected to inundation experience elevated runoff coefficients, increased sediment detachment, and reduced infiltration, collectively constraining groundwater replenishment even in zones with otherwise favourable structural characteristics. To evaluate how successive flood events alter the district’s surface conditions, Land Use and Land Cover (LULC) changes associated with the major floods of August 2018 vi and October 2021 were analysed using multi-temporal satellite imagery classified with a Random Forest algorithm on the Google Earth Engine platform. The 2018 floods affected approximately 20.86 km², influencing built-up (0.48 km2), forest (10.60 km2), agricultural (5.11 km2), and barren (4.67 km2) areas, whereas the 2021 event impacted about 19.24 km² across built-up (0.24 km2), agricultural (6.23 km2), barren (2.82 km2), and forest (9.95 km²) classes. These spatial patterns illustrate the substantial modifications to vegetation cover, agricultural land, and terrain stability driven by repeated high-intensity rainfall and flooding in the region. The impact of flooding on surface runoff was further quantified using the Soil and Water Assessment Tool (SWAT) applied to the Periyar River Basin, the largest basin within the Idukki district. Model simulations indicate substantial amplification of runoff during extreme rainfall events, with several outlets exhibiting nearly a 128% rise and others showing increases exceeding nearly 126% compared to non flood conditions. These elevated discharge responses spatially coincide with the high and very high flood susceptibility zones, reinforcing the reliability of the susceptibility modelling and highlighting the presence of hydrologically sensitive regions within Idukki. Collectively, the findings demonstrate strong interactions between surface flooding, groundwater recharge dynamics, landscape transformations, and basin scale runoff behaviour. The integration of multi-criteria analysis, machine-learning modelling, remote sensing, and hydrological simulation provides a comprehensive framework for characterising flood impacts in a complex mountainous environment. The outcomes offer valuable insights for flood mitigation planning, groundwater management, and sustainable land use decision making, while establishing a robust methodological foundation for future hydrological assessments in similar data-scarce regions.