Abstract:In order to evaluate the contribution of each process from numerical forecast, objective method to subjective forecast products to the accuracy of weather forecast, the Meteorological Evaluation Program Library (MetEva) is developed. Aiming at the whole process coverage of the verification algorithm and the comparability of the evaluation results, MetEva adopts a hierarchical architecture including basic layer and functional layer, and designs a modular inspection and calculation process based on a unified data structure. The program library provides over 400 functions around the steps of data reading, data merging and matching, sample selection, sample grouping, inspection calculation and result output for verification. MetEva provides 54 evaluation methods in five categories, covering most of methods recommended by the World Meteorological Organization and algorithms in domestic specifications. By using matrix calculation in each module and providing parallel scheme for verification algorithms, the operation efficiency is improved. Taking the evaluation of temperature and precipitation forecast as an example, this paper briefly explains the application MetEva, and shows its value in verification. The program library has been released as open source, which can effectively support meteorological departments at all levels to carry out the evaluation of the whole process of weather forecast.