Abstract:

The global commissioning of BDS-3 has catalyzed extensive interest in applying GNSS technology for precipitable water vapor (PWV) retrieval. However, the accuracy of PWV estimation is frequently constrained by the absence of in-situ meteorological observations across many provincial CORS networks in China. Traditional approaches conventionally rely on the ERA5 reanalysis to derive meteorological parameters; nevertheless, their local applicability is somewhat limited by the sparsity of assimilated regional observations. In contrast, the newly released CRA40 atmospheric reanalysis dataset demonstrates a superior capability in reconstructing atmospheric background fields, leveraging its inherent algorithmic and data-assimilation advantages for domestic meteorological monitoring. Taking the Shaanxi CORS network as a case study, this paper investigates the feasibility of joint PWV retrieval utilizing BDS-3 and CRA40 in data-scarce regions. Based on the ZTD estimated from BDS-3, PWV was retrieved in combination with ERA5, CRA40, and NCEP2, respectively, and the performances were comparatively evaluated against in-situ measurements. Results demonstrate that the BDS-3 and CRA40 combined approach outperforms both ERA5 and NCEP2 in the Shaanxi region, yielding a BIAS, MAE and RMSE of－0.24, 1.35, 1.66 mm, respectively. Furthermore, a spatio-temporal autocorrelation analysis of the RMSE utilizing Moran's I verifies the robust adaptability of this strategy under the complex topographic conditions of Shaanxi province. Finally, the evaluation reveals that while the proposed method exhibits high precision and strong robustness on a national scale, the retrieval accuracy over regions with severe topography remains subject to further optimization. This paper provides a novel and high-precision PWV estimation paradigm for CORS networks lacking continuous meteorological observations.

Key words: BDS-3, CORS, PWV, global atmospheric reanalysis dataset, Moran's I