Abstract:Accurate prediction of crop evapotranspiration (ETc) can provide a scientific basis for improving water use efficiency, rational allocation of water resources, and sustainable management of ecosystems. However, conventional statistical methods have the limited application in some regions, because of the complex relationships (nonlinear, linear, exponential, etc) between ETc and its each driving factor. This paper illustrated the utility of the back-propagation neural network (BP) model for ETc prediction and compared the performance of this model with that of the multiple linear regression technique. Combined with the ETc measured by Eddy Covariance and the meteorological data, three-layer BP models, trained by the Levenberg-Marquardt algorithm, were developed with a 4-10-1 architecture (corresponding to four, ten, and one nodes in the input, hidden, and output layers, respectively). The BP models were trained and validated in the MATLAB environment, with three different combinations of maximum air temperature, minimum air temperature, sunshine hours, crop coefficient (Kc), and precipitation. A high correlation was observed between the values measured by Eddy Covariance and those predicted by BP, with a higher coefficient of determination (0.87) and accuracy (91.44%) than that achieved by the multiple linear regression model (0.79 and 82.96%, respectively). Furthermore, the performance of the BP models could be improved substantially by including the dynamic change of Kc and the effect of precipitation, which suggested that these two factors were crucial variables in modeling ETc by BP approaches. In testing sets, the BP model that considered the dynamic change of Kc was found to be superior to the BP model that considered the crop coefficient recommended by FAO-56, with prediction accuracy increased by 19%. Additionally, the accuracy of the BP prediction models that considered the dynamic change of Kc and the effect of precipitation were higher than 95%. Moreover, the BP prediction model in typical weather with precipitation was superior to the BP model in typical weather with no precipitation. The results obtained would be helpful for obtaining the ETc more consistent with the actual growth status of the crop, thereby providing a scientific basis for improving water management during crop production.

Regularized Back-Propagation Neural Network for Rainfall-Runoff Modeling

Identification of the controlling factors for hydrological responses by artificial neural networks

Back-propagation Artificial Neural Networks for Water Supply Pipeline Model

Analysis of Runoff Coefficient Prediction Based on LM-BP Neural Network

Application of BP Neural Network Algorithm in Traditional Hydrological Model for Flood Forecasting

Short-term Runoff Prediction Optimization Method Based on BGRU-BP and BLSTM-BP Neural Networks

Short-Term Rainfall Forecasting by Combining BP-NN Algorithm and GNSS Technique for Landslide-Prone Areas

An improved Back Propagation Neural Network framework and its application in the automatic calibration of Storm Water Management Model for an urban river watershed

Application of Back-Propagation Neural Network to Estimate Precipitation with Doppler Radar in Yishuhe Watershed of China

Runoff Forecasting Using Machine-Learning Methods: Case Study in the Middle Reaches of Xijiang River

The Prediction of Rainfall Based on Superimposed Markov Chains and BP Neural Network Model

Daily runoff forecasting by deep recursive neural network

A new hybrid data-driven model for event-based rainfall–runoff simulation

Application of Back-Propagation Neural Network in Precipitation Estimation with Doppler Radar

Mid- to long-term runoff prediction by combining the deep belief network and partial least-squares regression

Application of Long Short-Term Memory (LSTM) on the Prediction of Rainfall-Runoff in Karst Area

TreeLSTM: A spatiotemporal machine learning model for rainfall-runoff estimation

Comparison of Long Short Term Memory Networks and the Hydrological Model in Runoff Simulation

High Temporal Resolution Rainfall Runoff Modelling Using Long-Short-Term-Memory (LSTM) Networks

Integration of a Parsimonious Hydrological Model with Recurrent Neural Networks for Improved Streamflow Forecasting

Crop evapotranspiration prediction by considering dynamic change of crop coefficient and the precipitation effect in back-propagation neural network model