Modeling and Forecasting Monthly Humidity in South Asia: A SARIMA Approach

This study employs the Seasonal Autoregressive Integrated Moving Average (SARIMA) model to forecast monthly average humidity trends in South Asian countries, utilizing data spanning January 1981 to December 2023 from NASA Power Data. The analysis focuses on Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan, and Sri Lanka. Following the Box-Jenkins methodology, optimal SARIMA models were developed for each country to project humidity trends for the period January 2024 to December 2026. Model validation was conducted using in-sample data (1981-2023), while the out-of-sample forecast covered 2024-2026. The SARIMA models demonstrated robust predictive accuracy within a 95% confidence interval, revealing significant changes in humidity patterns potentially driven by climate change across the region. The selected SARIMA models include: Bangladesh (ARIMA (1,0,0) (0,1,1) [12] with drift), Afghanistan (ARIMA (4,0,0) (0,1,1)[12]), Bhutan (ARIMA (1,0,2) (0,1,2) [12]), India (ARIMA (1,0,2) (2,1,0)[12]), Maldives (ARIMA (1,1,1) (2,0,0) [12]), Nepal (ARIMA (1,0,1) (0,1,1) [12] with drift), Pakistan (ARIMA (0,1,3) (2,0,0) [12]), and Sri Lanka (ARIMA (0,0,2 )(2,1,1) [12]). These findings provide insightful contributions to understanding the evolving climatic dynamics of South Asia and offer a solid foundation for future research on humidity-related climate variations and their environmental implications.