Predicting Tomato Irrigation Intervals and Biomass With Random Forest Modeling: Climate Resilience Strategies in Ethiopia

The accuracy of random forest (RF) machine learning models has not been previously evaluated for predicting tomato irrigation intervals and yield under the combined effects of gypsum and mulching, particularly in the context of climate change in Fura Kebele, Ethiopia. To address this gap, the present study aimed to predict tomato irrigation intervals and biomass using the RF model under projected future climate scenarios. The experimental treatments included control (T1), mulching (T2), gypsum application (T3), and 50% mulching with 50% gypsum (T4). The application rates of straw mulch and gypsum were 15 and 14.5 t/ha, respectively Historical climate data (2010–2023) were collected from the Mirab Abaya meteorological station, while future climate projections for the periods 2030–2060 and 2061–2090 were obtained from the Coupled Model Intercomparison Project Phase 6 (CMIP6) under the Shared Socioeconomic Pathway 5 (SSP5‐8.5) scenario. Model performance was evaluated using root mean square error (RMSE), the coefficient of determination ( R 2 ), and the Nash–Sutcliffe efficiency (NSE). The predicted mean irrigation intervals for T1 and T3 were 7, 6, and 5 days for the respective periods 2010–2023, 2030–2060, and 2061–2090; for T2, the intervals were 9, 8, and 6 days; and for T4, 8, 7, and 5 days. Projected yield reductions by 2030–2060 were 23.1% (T1), 19% (T2), 20% (T3), and 15% (T4), and by 2061–2090 were 41.2%, 37%, 40%, and 30%, respectively. The RF model demonstrated strong predictive performance, with R 2 = 0.73, RMSE = 1.15 days, and NSE = 0.72 for irrigation interval prediction, and R 2 = 0.90, RMSE = 1.06 ton/ha, and NSE = 0.91 for biomass prediction. Among the treatments, T4 exhibited the highest resilience to climate stress. Furthermore, the RF model was more accurate in predicting biomass than irrigation intervals under changing climate conditions.