Integrated probabilistic framework for long-term solar resource reliability assessment for photovoltaic planning at provincial capitals in Sumatra

Reliable solar resource assessment requires consideration of both expected performance and associated uncertainty, particularly under heterogeneous climatic conditions. This study develops an integrated probabilistic framework for evaluating long-term solar resource reliability at ten provincial-capital reference locations across Sumatra, Indonesia. Satellite-derived hourly global horizontal irradiance data from 2015 to 2024 were temporally integrated to obtain daily global horizontal irradiation and analysed through climatological characterisation, probabilistic photovoltaic yield estimation, composite reliability assessment, multidimensional pattern recognition, and Bayesian hierarchical inference. Because the hourly observations were integrated into daily totals, the analysis characterises variability among daily irradiation values over the 2015–2024 period and does not resolve intraday fluctuations. Mean daily irradiation varied only moderately among the analysed locations, whereas temporal variability and photovoltaic yield uncertainty differed substantially. Median annual photovoltaic yield ranged from approximately 1330 to 1474 kWh/kWp/year, showing that locations with similar mean irradiation can exhibit markedly different long-term reliability characteristics. A composite reliability index combining normalised scores derived from temporal variability, annual photovoltaic yield spread, and resource magnitude ranged from approximately 0.15 to 0.80. The Bengkulu and Lampung reference locations obtained the two highest index values, whereas the Riau Islands reference location obtained the lowest. Bayesian hierarchical modelling quantified uncertainty in the location-level reliability estimates and generated partially pooled posterior estimates, thereby supporting more cautious interpretation of extreme deterministic scores. The framework provides a transparent and reproducible basis for preliminary comparative screening and for prioritising locations for detailed ground measurements and techno-economic assessment. The empirical findings are specific to the analysed locations in Sumatra; application to other tropical regions requires independent validation using locally representative and spatially distributed data.