Land Surface Temperature-Aware Optimization of Grid-Connected Photovoltaic Systems for Enhanced Energy Yield: A Comparative Assessment of Metaheuristic Algorithms in Adamawa State, Nigeria

Accurate photovoltaic (PV) energy-yield estimation in thermally variable environments requires consideration of spatial differences in environmental conditions beyond conventional solar-resource assessment. This study developed and comparatively evaluated a land surface temperature (LST)-aware optimization framework for grid-connected PV systems across seven locations in Adamawa State, Nigeria: Yola North, Yola South, Mubi North, Mubi South, Gombi, Numan, and Michika. Global horizontal irradiance, ambient temperature, wind speed, and satellite-derived LST were integrated within a temperature-coupled PV performance model in which LST was treated as a thermal indicator rather than being directly equated with module temperature. PV tilt angle, azimuth orientation, and installed capacity were optimized using Parametric Sweep, Genetic Algorithm (GA), and Particle Swarm Optimization (PSO). Performance was evaluated using specific energy yield, improvement relative to baseline configurations, convergence behaviour, and solution stability. The optimized configurations exhibited tilt angles of 12–16°, azimuth angles of 165–180°, and installed capacities of 100–108 kWp, demonstrating spatial variation in the preferred PV configuration. At the algorithm level, PSO achieved the highest average specific yield of 2,550 kWh/kWp and a 7.6% improvement over its corresponding baseline, followed by GA at 2,530 kWh/kWp and 7.3%, and Parametric Sweep at 2,475 kWh/kWp and 6.5%. PSO also reached the reported convergence point at 30 iterations, compared with 45 iterations for GA, and exhibited lower reported variability (standard deviation = 1.8 versus 2.5). At the site level, LST exhibited a strong inverse association with optimization gain (Spearman’s ρ = −0.964, p = 0.00045).