where Pi are the predicted values, Oi are observed values,and are the average of Pi and Oi, and n is the total number of data.
2.6 Statistical analysis
To achieve the best comparison between the models and measurements, we need to select the dominant meteorological factors affecting the measured ET. Given that it may not be appropriate to explore results based solely on the coefficient of independent variables in multiple regression analysis, owing to the strong collinearities and nonlinearities among meteorological factors, we adopted a boosted regression trees (BRT) model to quantitatively evaluate the relative influences of meteorological variables on measured ET. In the past, the BRT method has been widely used to improve the performance of a single model through by fitting a large number of models, ultimately yielding an overall prediction (Martínez-Rincón, Ortega-García, & Vaca-Rodríguez, 2012). Most importantly, the BRT can evaluate the relative influence of an independent variable on a dependent variable, without transformations, and can cope well with non-linear relationships. Furthermore, the BRT displays good performance in dealing with stronger collinearities and nonlinearities. Thus, the BRT was adopted to evaluate the individual influences of controlling factors on measured ET. All statistical analyses were conducted in R software version 3.03(R Development Core Team, 2006), and all figures were plotted by Origin 9.0.