Machine Learning‐Based Model for Real‐Time GNSS Precipitable Water Vapor Sensing. Issue 3 (28th January 2022)

• Record Type: Journal Article
• Authors: Zheng, Yuxin; Lu, Cuixian; Wu, Zhilu; Liao, Jianchi; Zhang, Yushan; Wang, Qiuyi
• Is Part Of: Geophysical research letters. Volume 49:Issue 3(2022)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2022-01-28
• Subjects: GNSS -- precipitable water vapor -- machine learning -- real‐time inversion; Geophysics -- Periodicals; Planets -- Periodicals; Lunar geology -- Periodicals; 550
• DOI: 10.1029/2021GL096408
• Languages: English
• ISSNs: 0094-8276

Abstract

Abstract: Global Navigation Satellite Systems (GNSS) provide a promising opportunity for real‐time precipitable water vapor (PWV) sensing. However, relying on meteorological information restrains the implementation of real‐time inversion from tropospheric zenith total delays (ZTD) into PWV. In this study, a stacked machine learning model for mapping ZTD into PWV in the absence of meteorological parameters is developed. The model performance is assessed and validated with information offered by fifth generation European Centre for Medium‐Range Weather Forecasts reanalysis (ERA5) and radiosondes. An accuracy of better than 2.5 mm is achievable for the PWV values. Compared to the physical model which applies GPT3‐derived meteorological parameters, the proposed model reveals an enhanced performance, especially in the high‐latitude regions, with improvements of 28.1% and 22.2% when validated with ERA5 and radiosondes. This model is capable of fulfilling the demands of time‐critical meteorological applications and is also promising for real‐time PWV retrieval of other techniques that own the capability to sense ZTD. Plain Language Summary: The measurement of precipitable water vapor (PWV) in the atmosphere is very important to weather monitoring and forecasting. Global Navigation Satellite System observations provide a good opportunity for such measurement but are kept from practical due to a lack of timely meteorological observations. We developed a machine learning method to solve this problem by extracting weather patterns from the existing data sets and use it to create maps of the PWV without using meteorological information. The developed model reveals good performance and shows promising prospects for other techniques involved in atmospheric water vapor sensing. Our method also has the potential for further development of a real‐time water vapor product. Key Points: A machine learning model for mapping the zenith total delays (ZTD) to precipitable water vapor in the absence of meteorological quantities is developed Accuracy better than 2.5 mm is achievable for model predictions, which fulfills the demands of time‐critical meteorological applications This model is promising for real‐time water vapor inversion of other techniques that own the capability to sense the ZTD