Multivariate modeling of pan evaporation in monthly temporal resolution using a hybrid evolutionary data-driven method (case study: Urmia Lake and Gavkhouni basins)

Evaporation is an important meteorological variable that has a great impact on water resources. In the current research, climatology data, and seasonal coefficient have been used to estimate monthly pan evaporation (E_pan) for 2005-2018 study years at four selected stations of the Urmia Lake basin with Dsa and six selected stations of Gavkhouni basin with Bsk climate categories, in Iran. Estimation of monthly E_pan was performed using data-driven methods such as artificial neural networks (ANNs), adaptive neuro-fuzzy inference system (ANFIS), and gene expression programming (GEP) as well as wavelet-hybrids (WANN, WANFIS, and WGEP). Based on the evaluation criteria, the WGEP model performance was better than the other models in estimating the monthly E_pan. The results indicated that WGEP and ANN are the best and poorest models for all stations without affecting the climate condition of basins. The values of RMSE for WGEP model for stations of Urmia Lake and Gavkhouni basins were varied from 15.839 to 26.727 and 20.651 to 70.318, respectively. Also, the values of RMSE for ANN model for stations of Urmia Lake and Gavkhouni basins were varied from 29.397 to 38.452 and 30.635 to 85.237, respectively. The model's performance was improved as a result of considering the data noise elimination and applying seasonal coefficient to estimate E_pan of various climatic conditions_. This study with presenting mathematical equations for estimating monthly E_pan has a significant impact on the management and planning of water resources policymakers in the future.