Variability of climate-induced rice yields in northwest Bangladesh using multiple statistical modeling

Because of evident climatic variations and significant contribution to national food production, Bangladesh is a climate extreme hotspot region of examination for climatic consequences for rice (Oryza Sativa) crop production. This study intends to explore the variability of climatic variables (e.g., variations in mean temperature, rainfall, relative humidity, and sunshine duration) with rice yields (e.g., Aus, Aman, and Boro rice varieties) in northwest Bangladesh. The modified Mann–Kendall test, Theil–Sen slope estimator, and multiple linear regression (MLR) modeling were used to estimate the association among these factors. Heteroskedasticity and autocorrelation constant standard error (HAC) and feasible generalized least square (FGLS) technique were adopted to measure the climate-rice crop nexus using the regional level dataset for 1976–2015. Furthermore, the spatiotemporal variation of rice yield trends with climatic variables was mapped and assessed by the coefficient of variation. The results show that observed temperature and humidity trends were beneficial for Aus and Aman yields but not Boro yields. In contrast, observed rainfall and sunshine trends were negative for all three rice seasons. The outcomes of the MLR model explained 67%, 92%, and 83% of the variability in Aus, Aman, and Boro rice yields in the study region. The model outcomes showed that humidity and rainfall have negatively affected Aus and Aman rice crops, while temperature and rainfall positively influence Boro rice yield. Regarding the climate change issues and safeguarding food safety at the regional level, the concerned authorities should provide substantial attention to improving heat and drought-tolerance high-yielding varieties against climate effects on Aus and Aman rice varieties.