Pan Evaporation Prediction Based on Grey Correlation and Machine Learning

Pan evaporation serves as the most intuitive and fundamental data for calculating watershed evaporation. Its trends and variation patterns are instrumental in analyzing watershed water resource balance and improving water utilization efficiency. However, due to the influence of multiple factors and the complex nonlinearity inherent in the evaporation process, traditional evaporation prediction methods suffer from poor generalizability and low accuracy. To enhance the accuracy of pan evaporation predictions and provide methodological references for related research, this study uses long-term pan evaporation data and meteorological observations from the Yingpan Evaporation Experiment Station. First, a correlation analysis is conducted between pan evaporation and its influencing meteorological factors. On this basis, two machine learning models are constructed for pan evaporation prediction. The results reveal that the relationships between various meteorological factors and evaporation exhibit significant monthly variations. Both prediction models developed in this study achieve an accuracy rate exceeding 80%, confirming their effectiveness as reliable approaches for pan evaporation prediction. Additionally, this study provides a methodological reference for evaporation prediction in other regions.