Forecasting Daily Reference Evapotranspiration by the Hybrid Wavelet-Nonlinear Autoregressive Neural Network Approach
Abstract
The accurate forecast of reference evapotranspiration (ETo) has a vital role in real-time decisions on water resources management by quantifying the prospective changes in agricultural and hydrological processes. The real-time decisions on irrigation scheduling are mainly made based on the agricultural water demand predictions, which themselves strongly depend on ETo. Reference evapotranspiration is a complex process driven mainly by weather variables, and thus is characterized by high non-linearity and non-stationarity. In this study, the nonlinear autoregressive (NAR) neural network, and hybrid wavelet-NAR (WNAR) approaches are used to forecast ETo for the 7 days ahead at six sites (namely, Alliance, Champion, Dunning, McCook, Mead, and North Platte) in Nebraska. These sites are chosen to cover various climatic conditions. At each site, 60%, 20%, and 20% of the daily ETo measurements from 1994 to 2015 are used respectively to train, test, and validate the NAR and WNAR networks. Thereafter, the trained NAR and WNAR networks are utilized to forecast ETo in 2016. The number of hidden nodes and layers is determined empirically to optimize performance of the networks for the particular site. Three training functions (Levenberg-Marquardt backpropagation, resilient backpropagation, and scaled conjugate gradient) and three transfer functions (tansig, logsig, and radbas) are employed in the networks. The Levenberg-Marquardt Backpropagation and tansig functions generate the best results. The findings show that the NAR and WNAR approaches can accurately forecast ETo. The six-site average root-mean-square-error (RMSE) of ETo forecasts from NAR is 1.70 (mm/day). The WNAR approach decreases the corresponding RMSE to 1.34 (mm/day), implying WNAR outperforms NAR. The results also show that both the NAR and WNAR approaches perform well in dry sites, but their performance degrades in wet sites.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.H33L2124A
- Keywords:
-
- 1847 Modeling;
- HYDROLOGY;
- 1873 Uncertainty assessment;
- HYDROLOGY;
- 1906 Computational models;
- algorithms;
- INFORMATICS;
- 1942 Machine learning;
- INFORMATICS