Exploring the effectiveness of Artificial Intelligence-powered insights in drought study in Nagaland, India

Abstract

Drought, as a natural calamity, has serious economic and environmental implications, especially as the impacts of climate change continue to escalate globally. In many regions, monitoring and comprehending changes in drought patterns have become imperative. As climate change increasingly influences hydrological cycles, there is a need to grasp and interpret drought behaviour in diverse geographical areas. This study is particularly focused on a landlocked state in the north-eastern region of India, which is characterised by a predominantly monsoon climate with high humidity and an annual rainfall of 1800–2500 mm. The study focuses on the state of Nagaland, India, and is aimed at evaluating the efficacy of artificial intelligence (AI) models such as Multilayer Perceptron (MLP), Long Short-Term Memory (LSTM), and Genetic-Algorithm Adaptive Neuro-Fuzzy Inference System in predicting drought. For analysing the drought conditions, the Effective Drought Index (EDI) is used. By utilising rainfall data from 1987–2021, the EDI drought index has been computed, recognising the pivotal role of rainfall in comprehending prevailing drought conditions. The drought conditions are categorised from extremely dry to near normal, excluding the wet conditions in the study region. The investigation into the effectiveness of AI in predicting and detecting drought yielded insightful results, highlighting the informative and promising capabilities of AI models. The results of the study facilitate a comparative analysis of the three models, MLP, LSTM, and GA-ANFIS, using the evaluation metrics. The study findings indicate that LSTM exhibits superior prediction accuracy in the study region in terms of its ability to predict drought conditions in the given geographical area. This outcome is crucial for understanding and addressing the impacts of drought. This study contributes to the broader understanding of drought prediction and emphasises how AI models can improve their ability to predict drought conditions, which will ultimately contribute to enhanced water resource management and climate adaptability.