Modeling and climate prediction of frost damage in apple tree yield in areas of Western Iran (Kermanshah County)

Using modeling and statistical methods, this study describes the relationship between spring frost and apple crop yield in actual data and predicted data from the LARS WG model in mountainous areas of western Iran, focusing on the Kermanshah region. And the model is under RCP4.5 and RCP8.5 scenarios with a base period of 2005 to 2025. To evaluate the predicted data in the LARS-WG model, the error rate of the observational and predicted data was examined using the R2 criteria (RMSE = 1.8), (MSE = 3.24), (MAD = 2.26) and the relationship between apple yield and spring frost forecast data for a ten-year period was performed. The magnitude of the trend indicates a projected 15–20% increase in spring frost risk by 2035. The quantitative assessment suggests that there is one critical risk year within the next 5 years, where the predicted minimum spring temperature (Tmin = -2.2°C) will drive yield to its lowest predicted point (16,523.63 t/ha). Modeling results showed a strong correlation between apple yield and spring frost (R = 0.769, R² = 0.592, p < 0.1), indicating that approximately 59% of yield variability could be explained by temperature-related frost events.