Tightly coupled tomography model for atmospheric water vapor based on multi-source remote-sensing and GNSS data

Accurate monitoring of water vapor content is of great significance to analyzing global warming. Presently, the amount of available atmospheric precipitable water vapor (PWV) data has exploded, mainly including all-weather GNSS PWV, multi-band spaceborne remote sensing PWV, global coverage reanalysis, etc. A three-dimensional water vapor density field has advantages in reflecting the vertical motion of water vapor. Therefore, based on the constraints built by historical reanalysis, we combined high-precision GNSS data and three types of remote sensing data (near-infrared: MODIS, long wave infrared: FY-4A, microwave: MIMIC) to build a tomography model. The retrieved three-dimensional water vapor density field from the model is validated by water vapor density from reanalysis (ERA5_WVD) and radiosonde data (ROS_WVD), respectively. The tomographic results are more consistent with ERA5_WVD. From the ERA5_WVD validation results, the prior-variance weighting strategy has slight advantages in solving the fused tomographic model than that of equal weighting strategy, especially for the top layers. Using weighting strategy, the RMSEs of the three tightly coupled tomographic models are less than 2 g/m ³ in most epochs, and their average RMSE is 1.59, 1.69, and 1.65g/m ³ , respectively. As the height rises, the RMSE value decreases gradually, but the relative RMSE value increases at first and then decreases. Compared with the single-GNSS tomographic model, the tightly coupled tomography model based on microwave remote sensing data (MIMIC_PWV) and GNSS data has the highest improvement in accuracy, followed by fusing with long-wave infrared FY4A_PWV and near-infrared MODIS_NIR_PWV. The comparisons of fusing different remote-sensing data contribute to selecting appropriate remote-sensing data to improve the tomographic water vapor density. Furthermore, it cannot be negatable that the tomographic model fusing the remote sensing data has a significant advantage in tomographic water vapor density near the ground, and the improvement ranges from 5%~20%. Therefore, adding the remote sensing data to perform tightly coupled tomography is significant for monitoring water vapor motion in the bottom atmosphere, which is essential to understanding and coping with climate change, water resources management, and ecological environment protection.