Impact of Different Vegetation Scenarios on Solar Power Plant Performance: An ENVI-Met and Machine Learning–Based Approach

Abstract

Solar energy systems are highly sensitive to microclimatic conditions, particularly in hot-summer continental regions where elevated temperatures significantly reduce photovoltaic (PV) efficiency. This study investigates how different vegetation configurations influence the microclimate and performance of a 250 kW operational PV plant located on a 5000 m2 area at Dicle University, Turkey. A dual-method approach was applied by combining (i) ENVI-met three-dimensional microclimate simulations under four vegetation scenarios grass, broad-leaved deciduous plants, coniferous plants, and mixed vegetation and (ii) machine learning–based multiple linear regression (MLR) using real operational PV data, including air temperature, relative humidity, wind speed, solar irradiance, and physiological equivalent temperature (PET). The machine learning model showed high predictive capability (R2 = 0.9737, RMSE = 378,939.55, and mean absolute error (MAE) = 484.44), revealing strong relationships between climatic factors and PV efficiency. ENVI-met simulations demonstrated that mixed vegetation reduced summer air temperature by up to 1.7°C, while coniferous vegetation provided the greatest improvements in thermal comfort (PET). These findings confirm that vegetation-based microclimate design can mitigate thermal stress and support improved PV performance.