Assessing Solar Irradiance Trends and Temperature Extremes by Applying Machine Learning Based Statistical Modelling for Renewable Energy Optimization

In the pursuit of sustainable development goals, understanding the temporal trends and variability of solar irradiance (SI) is crucial, especially for the promotion of clean energy production and environmental pollution management. The topic has received scant attention in research literature, with no attention given to arid regions. This study investigated the variabilities and trends of SI and the influence of maximum temperature (T_max) events by applying machine learning (ML) based statistical modelling in Kuwait during 1984–2022. Monthly variabilities were identified by calculating descriptive statistics. ML based linear trend models including Mann-Kendall test (MKT), and Innovative Trend Analysis (ITA) were used for daily, monthly, and annual trend analyses. Density plots and Pearson’s correlation coefficients (r) were used to assess linear associations and dominance range. Time-frequency relationships and associations were assessed using wavelet coherence transform (WCT) analysis. Outcomes revealed that Kuwaiti cities experienced peak SI in June and July (≥ 7.0 kW.hr/m²/day), suggesting a high solar energy production period. All three trend models exhibited consistent outcomes, with increasing SI trends in all cities except in Mina Suud. The values of r suggest a strong positive influence of T_max on rising trends of SI in all cities, with a pronounced influence between the 42^oC and 47^oC Tmax ranges. The in-phase arrow directions in WCT maps suggest a positive association between T_max and SI, with a strong power spectrum at a 200 ~ 700 time-frequency scale and a 95% confidence level. The insights from this study will be instrumental in policymaking for sustainable environmental management by achieving SDG 7-Affordable and Clean Energy.