Publication

Abstract : With climate change presenting a significant global threat, semi-arid regions are particularly at risk due to rising global temperatures and changing rainfall patterns. This study explored climate risk and livelihood vulnerability in Dongarampur, located in the drought-prone semi-arid region of northeastern Karnataka, India, utilizing machine learning and participatory rural appraisal techniques. The findings reveal significant interannual variability in rainfall over the past 73 years, characterized by a declining trend with a Mann–Kendall τ of –0.148. In contrast, maximum temperature exhibits a strong and statistically significant increasing trend (τ = 0.534, p < 0.0001). Drought risk assessment, employing eight indices, consistently identified multiple drought events throughout the study period, detecting 21, 24, and 14 drought years through the standardized precipitation, streamflow, and soil moisture drought indices, which capture different aspects of meteorological, hydrological, and agricultural drought conditions respectively. A livelihood asset pentagon was formulated using 24 indicators to evaluate households’ livelihood capital, revealing that households had a high physical (0.686) and financial (0.624) capital, moderate social (0.478) capital, and limited natural (0.379) and human (0.351) capital. The livelihood vulnerability index (LVI), based on seven primary and 32 sub-indicators, indicates a high level of household vulnerability to climate change, with adaptive capacity (0.439) and sensitivity scores (0.379) significantly lower than the exposure score (0.608). This emphasizes the urgent need to enhance community capacities. Additionally, land use and land cover analysis revealed an increase in fallow land by 186.59 hectares, alongside reductions in cropland and water bodies by 76.16 hectares and 29.40 hectares, respectively. This trend suggests a shift away from productive land use within the community. In conclusion, the study provides practical recommendations for harnessing this information to develop context-specific climate-smart adaptations to tackle current and projected climate challenges.