Publications

Wang, YB; Jiang, N; Wu, YH; Xu, Y; Kaufmann, H; Xu, TH (2024). An Improved Model for the Retrieval of Precipitable Water Vapor in All-Weather Conditions (RCMNT) Based on NIR and TIR Recordings of MODIS. IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, 62, 5513412.

Abstract
This article aims to explore improving the accuracy of water vapor retrieval in all-weather conditions by combining the recordings of the near-infrared (NIR) and thermal infrared (TIR) bands of the Moderate-Resolution Imaging Spectroradiometer (MODIS). When analyzing different input parameters, we found that the NIR methods are more strongly influenced by a high cloud density than the IR methods. This motivated us to develop an improved model, named resilient combination model of NIR and TIR bands (RCMNT), which is based on machine learning algorithms and optimizes the accuracy of water vapor retrieval depending on varying cloud cover densities. The RCMNT uses high-precision precipitable water vapor (PWV) retrieved by the Global Navigation Satellite System (GNSS) as a reference for model training. Compared with GNSS PWV, the results show that the RCMNT is superior to the MOD05 in terms of spatiotemporal forecast accuracy. The RCMNT reveals a significant improvement in accuracy of 69.06% and a reduction in the root-mean-square error (RMSE) by 6.9335 mm under all-weather conditions compared to MOD05. The RCMNT also shows a substantial enhancement under cloudy conditions, with an increase in temporal prediction accuracy of 69.47% and an improvement in spatial prediction accuracy of 69.84%. The RMSE of the RCMNT decreased by 7.0251 mm, resulting in a 69.50% improvement in accuracy compared to MOD05 under cloudy conditions. Overall, the RCMNT shows its potential through a combined analysis of NIR and TIR data, achieving more accurate results for PWV retrieval across various spatiotemporal scales and weather conditions.

DOI:
10.1109/TGRS.2024.3381750

ISSN:
1558-0644