Nonlinear Drivers and Volatility of Solar Radiation Variability in Asian Megacities: A Functional Time-Series Approach

Solar radiation is a significant renewable energy source, but its variability in Asian megacities is driven by advanced meteorological drivers that are poorly represented by traditional linear models. This study employs a functional time series method to investigate the nonlinear drivers and volatility of solar variability in five typical cities Dhaka, New Delhi, Jakarta, Manila, and Kuala Lumpur from July 2016 to June 2025. Solar irradiance was represented as continuous daily curves with functional data analysis (FDA), and the first three components explained more than 97% of the variance with functional principal component analysis (FPCA). Seasonal clustering indicated monsoon-dominated regimes for New Delhi, Manila, and Dhaka, and equatorial stability for Kuala Lumpur and Jakarta. Threshold and nonlinear modelling indicated city-specific tipping points: rainfall (~43 mm) for Dhaka, humidity (~48%) for New Delhi, wind (~4 m/s) for Jakarta, and precipitation (~108 mm) for Manila. Volatility analyses via GARCH-family models confirmed clustering and persistence, with EGARCH capturing the asymmetric effects of negative shocks. VAR impulse–response functions demonstrate that precipitation shocks cause immediate but transitory reductions in solar returns. These findings demonstrate that solar variability over Asian megacities is controlled by nonlinear thresholds and clustered volatility and has implications for forecasting, grid integration, and climate-resilient energy planning.