Temporal Analysis and Forecasting of Rainfall of Kerala: A Machine Learning Based Approach

In agrarian societies, rainfall variability significantly affects water availability, impacting both drought and flood risks. This study explores the implications of rainfall variations in Kerala, using CHIRPS data and Google Earth Engine. Rainfall patterns from 2000 to 2021 were analyzed to establish a baseline, and data from 2022 revealed deviations from the average. Despite Kerala’s high average rainfall of 3,269 mm in 2022, months like January, February, October, and November showed negative deviations, suggesting potential meteorological drought conditions. The Standardized Precipitation Index (SPI) identified drought-like conditions in districts such as Wayanad, Palakkad, Idukki, and Thiruvananthapuram, while districts like Kottayam, Ernakulam, Alappuzha, and Kannur experienced above-normal rainfall, raising flood concerns. A random forest regressor model, trained on long-term data, was employed to predict rainfall patterns. The model's predictions, validated against CHIRPS data, achieved a 95% accuracy score, demonstrating its effectiveness in forecasting rainfall trends. The methodology developed can be applied to other regions and timeframes.