Publications

Gao, ZR; Jiang, N; Xu, Y; Xu, TH; Zeng, RB; Guo, A; Wu, YH (2023). A Spatial PWV Retrieval Model Over Land for GCOM-W/AMSR2 Using Neural Network Method: A Case in the Western United States. IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING, 16, 2954-2962.

Abstract
Precipitable water vapor (PWV) is an important and active part of the atmosphere. As known, microwave PWV retrieval is well applied for the ocean but challenging over land. In this article, we creatively established a microwave PWV retrieval spatial model over land using the backpropagation neural network to combine the high precision ground-based global navigation satellite system (GNSS) data and high spatial continuity satellite-borne data. Three-year data from 167 GNSS stations located in western America were utilized for training the network, and 40 untrained sites were selected as the test set. Root-mean-square error (RMSE) of the test set can reach 3.90 and 3.88 mm in the ascending (As) and descending (De) orbits, respectively. Then, we analyzed the influence of land cover types on the model over land. As a result, we found that stations located in the area with a single large-scale continuous land cover type had higher retrieval accuracy. In contrast, stations with diversified land cover types had lower precision. Furthermore, after fully considering the impact of land cover type, we performed an improved model with 61 stations on the single large-scale continuous grasslands, and the results of 8 stations as the test set showed that the RMSE could reach 3.41 and 3.31 mm in the As and De orbits, respectively. Compared with the spatial model established previously, the accuracy had been improved by about 13%. We think this is due to the stable physical properties (such as microwave emissivity) of the single large-scale continuous land cover type.

DOI:
10.1109/JSTARS.2023.3255259

ISSN:
2151-1535