A hybrid PCA-SEM-ANN model for the prediction of water use efficiency

This study employs a Structural Equation Model (SEM), Principal Component Analysis (PCA) and Artificial Neural Network (ANN) to construct a hybrid PCA-SEM-ANN model, for the prediction of Water Use Efficiency (WUE). The structural relationship and the degree of influence among factors is determined by SEM, and is transformed into ANN's topology, where PCA is employed to reduce spatial dimensionality. The applied results, in Kashgar, Xinjiang, China, show that different influencing factors on WUE present a diversity with different levels. The ANN structure optimized by SEM fits better, and the PCA-SEM-ANN model has high explanatory and precision for environmental control of the ecosystem as well as WUE simulation. The model can be widely applied to the vegetation ecosystem in the entire Xinjiang or elsewhere, providing a theoretical basis and a simulation method for improving the efficient water use capacity as well as predicting the future response of WUE to climate change.

Automated summary

This study utilizes SEM, PCA, and ANN to construct a hybrid model for predicting WUE. The application in Kashgar, Xinjiang, China shows that different factors have diverse influences on WUE, and the ANN structure optimized by SEM fits better. The PCA-SEM-ANN model has high explanatory power and precision, providing a theoretical basis and simulation method for improving water use efficiency and predicting future responses to climate change.