The joint probabilistic density functions (JPDF) of wind speed and direction are important prerequisites for directional wind energy assessment (DWEA). Based on the beta boundary kernel and the optimal bandwidth algorithm in the R programs, a nonparametric copula model (NP-copula) for the JPDF of wind vector data is proposed for DWEA in China. Eight parametric models, including five parametric copula models, two angle-linear (AL) models, and the anisotropic Gaussian model, are introduced for comparison. The three-year daily-average wind vector forecast data of mainland China (total 6019 nodes) is adopted for comparisons at the regional scale. The comprehensive metric value reaches 4.9669 (full score 5), indicating that the NP-copula model has the superior performance in fitting JPDF of wind vector data. Subsequently, the DWEA for China, including the estimations of direction-related wind power density (WPD) and wind turbine power output (WTPO), is carried out using the proposed NP-copula model. The estimated values of WPD and WTPO have good consistency with the reference values, indicating that the DWEA based on the NP-copula model is reliable. It is found that the regions with the most abundant wind resources are concentrated at the southeast coastal region, some western provinces, and the central and eastern regions of Inner Mongolia. The average values of WPD and WTPO could reach (or exceed) 240 and 5 GWh, respectively. Besides the average values, the direction-related WPD and WTPO are also identified based on the NP-copula model. For example, the wind resources at the southeast coastal region are concentrated at the S and SW directions. For the southern Xinjiang and western Gansu provinces, wind resources with SW and W directions are dominant. These results might be useful for the wind farm site selection, as well as the design and condition monitoring of wind turbine systems in China. (C) 2020 Elsevier Ltd. All rights reserved.