The generalized nonlinear advection aridity (GNAA) model has been widely applied for evaporation estimations and drought assessments at regional and global scales. Here, the GNAA model was combined with a semi-empirical function for the parameter alpha e-B to estimate actual evaporation (E) in nine Level-I water resources re-gions and one typical revegetation area (i.e., the Loess Plateau) in China. The results showed that the semi-empirical function for alpha e-B with aridity index rescaled the original independent variable (xB) in GNAA; thus, the model satisfied the boundary condition of y = 0, at xB-alpha f =0. Moreover, the data points of E/Epa (Epa: apparent potential evaporation) and Epo/Epa (Epo: potential evaporation) were captured well by the GNAA curve with the semi-empirical function of alpha e-B. Further, the GNAA model showed high accuracy in simulating evapo-ration (ECR-B) when compared with the water balance equation, with the mean absolute error, root mean square error, and Nash-Sutcliffe efficiency values being 52 mm, 66 mm, and 0.84 respectively, over 49 basins. The multi-year mean ECR-B was 443.1 +/- 14.7 mm/yr and 431.6 +/- 21.2 mm/yr across China and Loess Plateau, respectively. Moreover, the annual ECR-B showed a significant upward trend during 1982-2017, and ECR-B increased rapidly before 1998, but decreased slightly after 1998 when considering 1998 as the turning point across China. Since the implementation of the large scale Grain for Green project on the Loess Plateau, nomalized difference vegetation index (NDVI) and precipitation increased significantly, while ECR-B showed no significant increases in both China and Loess Plateau. For the Loess Plateau, annual precipitation increased faster than annual evapo-ration, implying that water yield may increase. Overall, the findings would assist in clarifying the characteristics of evaporation across China and contribute to better management and sustainable utilization of regional water resources.