Application of K-nearest neighbors (KNN) technique for predicting monthly pan evaporation

• Main Authors: Abed M., Imteaz M., Ahmed A.N., Huang Y.F.
• Other Authors: 36612762700
• Format: Conference Paper
• Published: American Institute of Physics Inc. 2024

Finding a reliable computational technique for determining pan evaporation (Ep) may be helpful in the assessment and application of techniques for the development of organic agricultural systems and management of water resources. In this study, monthly evaporation (Ep) was estimated utilising a K-Nearest Neighbours (KNN) model, with monthly climatological statistics obtained from a Malaysian weather station used for training and verifying the model, including climatic factors such as maximum and minimum temperature, mean temperatures, relative humidity, solar radiation, and wind speed, for the period 2000-2019. Several models were devised utilising different combinations of input components and other model factors, and the performance of the devised model was evaluated employing standard statistical indices. The outcomes of evaluation in the testing stage for the final suggested KNN-6 model were R2= 0.946, MAE=0.085, RMSE=0.115, RAE=0.211 and RSE=0.053. The results of the investigations in terms of various performance evaluation criteria highlighted that the proposed KNN structure can model the monthly evaporation losses with reasonable accuracy and thus be used to help local interested parties in discussing the ongoing management of water resources. � 2023 Author(s).