The Tianshan Mountains, one of the world's seven major mountain systems, are the farthest mountain range from any ocean and represent a typical alpine region in China. Known as the "Water Tower of Central Asia," this area holds strategic significance for Xinjiang and Central Asia. With advancements in remote sensing technology, satellite-derived precipitation has become a key tool for estimating mountainous precipitation. However, complex and heterogeneous terrain in mountainous regions leads to low accuracy in satellite precipitation retrieval products. To address this, this study developed a multi-source precipitation fusion dataset for the Tianshan Mountains. Using GSMaP satellite precipitation data as the initial field and incorporating ground-truth precipitation data from 1,065 regional stations, we developed an optimal interpolation-based fusion method for satellite-ground precipitation products, ultimately generating a daily precipitation dataset for the Tianshan Mountains (2000-2022). During development, strict quality control was applied to ground data, and daily fused precipitation data underwent rigorous evaluation. This dataset is expected to support water resource management and efficient utilization in complex terrain regions.
This dataset was primarily generated using GSMaP satellite precipitation and ground-based rain gauge data.
(1) GSMaP Satellite Precipitation: The Global Satellite Mapping of Precipitation (GSMaP) provides global hourly rainfall estimates at a 0.1° × 0.1° resolution. As part of the Global Precipitation Measurement (GPM) mission, GSMaP integrates passive microwave and infrared radiometer data from the GPM Core Satellite and auxiliary satellite constellations to produce global precipitation observations every three hours. GSMaP uses microwave datasets from low-Earth-orbit satellites and visible/infrared datasets from geostationary satellites as inputs for its retrieval algorithms, employing cloud motion vector and Kalman filter methods. It generates three products: GSMaP_NRT (near-real-time, using forward cloud motion vectors), GSMaP_MVK (standard product with bidirectional cloud motion vectors), and GSMaP_Gauge (adjusted using CPC global rain gauge data). All three products share a temporal resolution of 1 hour and a spatial resolution of 0.1° × 0.1°. Comparative evaluations show GSMaP_Gauge (gauge-corrected) has the highest accuracy among satellite products, making it the initial field for this fusion dataset. (2) Ground Precipitation: We selected hourly precipitation data from 104 solid precipitation stations (including 57 national stations, excluding 8 international exchange stations) and 961 regional automatic stations in the Tianshan Mountains. Ground data matched the temporal coverage of GSMaP and underwent quality control (climate extreme checks, station-specific extreme checks, and consistency tests) by the Xinjiang Meteorological Information Center. Solid precipitation stations use weighing-type gauges (measuring both rain and snow), while regional stations use tipping-bucket rain gauges (rain-only). Due to seasonal operational limitations (regional stations cease observations in cold seasons), the study focused on the warm season (May–September). A 10-fold cross-validation approach was applied: stations were stratified by elevation into 10 groups, with 90% (9 groups) used for modeling and 10% (1 group) reserved for independent validation, ensuring representativeness.
In this study, we carry out the development of a multi-source precipitation fusion dataset for the Tianshan mountainous region, using the GSMaP satellite precipitation data as the initial field and the live precipitation data from 1065 stations in the region during the same period, to develop an optimal interpolation-based fusion method of the star-earth precipitation products, and ultimately to generate a day-by-day precipitation product set in the Tianshan mountainous region for the period of 2000-2022.In this study, the optimal interpolation analysis takes the GSMaP precipitation as the initial estimation field, and the station real precipitation as the true value, and the final precipitation analysis value Ak on each grid point is equal to the initial valuation Fk of the point plus the deviation of the real observation value on the grid point from the initial valuation, which is weighted and estimated by the deviation of the known real observation value Oi from the initial valuation value Fi on n grid points within a certain range.
The dataset underwent strict quality control for ground data and comprehensive evaluation of daily fused precipitation. It is expected to enhance water resource management in complex terrains. Ground data from 104 solid precipitation stations and 961 regional automatic stations (aligned temporally with GSMaP) were quality-controlled by the Xinjiang Meteorological Information Center. Additionally, the fusion methodology was peer-reviewed and published in Journal of Hydrology, attesting to its credibility.
| # | number | name | type |
| 1 | 2023D01A17 | other | |
| 2 | 2023TSYCCX0079 | other | |
| 3 | 2021kf06 | other |
This work is licensed under a
Creative
Commons Attribution 4.0 International License.
CSTR:11738.11.NCDC.IDM.DB6553.2024
10.12072/ncdc.idm.db6553.2024