CMADS is awarded national recognition

Share to:

In accordance with the requirements outlined in the "Outline for High-Quality Development of Meteorology (2022-2035)" (State Council Document No. 11), the "Guiding Opinions of the State Council General Office on Improving the Mechanism for Evaluating Scientific and Technological Achievements" (State Council Document [2021] No. 26), and the "Interim Measures for the Evaluation of Meteorological Scientific and Technological Achievements of China Meteorological Administration" (CMA Document [2021] No. 2), the China Meteorological Administration (CMA) has organized the evaluation of scientific and technological achievements in the meteorological industry since the 13th Five-Year Plan period. Led by Professor Meng Xianyong and Professor Wang Hao, with the participation of the National Meteorological Information Center of China Meteorological Administration, the China Institute of Water Resources and Hydropower Research, China Agricultural University, and other units, the "China Meteorological Assimilation Driving Datasets for the SWAT model (CMADS)" has been recognized as an outstanding achievement in the meteorological industry during the 13th Five-Year Plan period.


The China Meteorological Assimilation Driving Datasets for the SWAT model (CMADS) is jointly funded by the National Natural Science Foundation of China's "Key Technologies and Datasets for Land Surface Reanalysis in the Qinghai-Tibet Plateau" and the "Key Technologies Research and Demonstration of Operation and Management for the Middle Route of the South-to-North Water Diversion Project during the 12th Five-Year Plan period". The CMADS dataset aims to address and provide downstream users in hydrology, nonpoint source pollution, and other fields with unified specifications (latitude, longitude, elevation), assimilate more data sources, and deliver reliable model meteorological input data assimilated and corrected using a large number of observation stations. Through a series of complex processes, including variational assimilation and optimized interpolation using convolutional neural networks, the regional reanalysis data for nearly 40 years (1979-2018) in China has been extensively corrected. This has led to the successful secondary production of localized grid meteorological products tailored to the needs of downstream users, such as hydrology, thereby significantly enhancing the accuracy and efficiency of scientific research data for users.


Atmospheric driving field quality is one of the key factors influencing the output results for downstream users. Poor atmospheric driving fields can lead to increased uncertainty in the output results of hydrological and other downstream models through error propagation. Given China's vast territory and complex terrain, traditional meteorological observation stations are no longer sufficient for driving research on large-scale water and energy balance processes. At present, coupling gridded atmospheric reanalysis data with hydrological and other physical models through the form of "virtual stations" is imperative.

The CMADS dataset is the earliest and only dataset in China that applies gridded products through big data technology to drive models in downstream fields such as hydrology and the environment. In terms of technological innovation, firstly, the background field of the CMADS gridded dataset adopts advanced multiple grid variational (STMAS) assimilation algorithms, combined with data from China's surface automatic weather observation stations, CMORPH precipitation, MERRA2 precipitation, and ERA5/ERA-Interim background fields (where the assimilation sources of the CMADS dataset from 2008 to 2016 use the CLDAS forcing field from the National Meteorological Information Center), ultimately establishing a 40-year CMADS climate gridded dataset (CMADS-GRID).

Secondly, based on the CMADS-GRID, the CLM 3.5 land surface model is driven, and after validation using national soil temperature observation stations, a 40-year ten-layer soil temperature land surface component dataset (0.7 cm, 2.8 cm, 6.2 cm, 11.9 cm, 21.2 cm, etc.) is generated.

Finally, based on the previous 40-year atmospheric gridded dataset (CMADS-GRID) and soil temperature dataset (CMADS-ST), combined with the team's published machine deep learning big data technology in Nature magazine, convolutional neural network optimized interpolation, and other technologies, a large amount of work has been completed, including quality control, data sampling, format conversion, and physical consistency tests of water and heat balance. This has led to the establishment of a 40-year CMADS series dataset covering the entire East Asia region, with unified specification virtual stations (latitude, longitude, elevation), with a temporal resolution of daily and spatial resolutions of 1/3° (58,500 stations), 1/4° (104,000 stations), 1/8° (416,000 stations), and 1/16° (1,664,000 stations), providing elements such as daily average temperature, pressure, relative humidity, specific humidity, wind speed, 24-hour cumulative precipitation, and soil temperature. CMADS further enhances the accuracy of meteorological data for grassroots scientific research users from the source.


In recent years, the CMADS dataset has entered a comprehensive application phase and has been widely utilized by over a thousand institutions, including the China Meteorological Administration, the Ministry of Ecology and Environment, the Ministry of Water Resources, and universities. Notably, in terms of supporting major national-level projects, the CMADS dataset has been employed in the Second National Pollution Source Census, providing crucial meteorological data support for this initiative. Overall peer evaluations indicate that the performance of hydrological models such as SWAT and VIC, driven by CMADS, surpasses that of other reanalysis datasets such as TRMM in various basins across China and East Asia. Presently, the CMADS dataset has gained widespread recognition and usage within both domestic and international academic circles. According to incomplete statistics, the user base of CMADS data has exceeded 1.15 million individuals, with tens of thousands of data requests received. Moreover, it has facilitated the publication of nearly 700 articles and has contributed to the implementation of 150 National Natural Science Foundation projects and 70 National Key Research and Development Program projects.