Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Department of Land Surveying and Geo-Informatics | en_US |
| dc.contributor.advisor | Liu, Zhizhao (LSGI) | en_US |
| dc.creator | Xu, Jiafei | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/13420 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | en_US |
| dc.rights | All rights reserved | en_US |
| dc.title | Developing novel machine learning-based algorithms to precisely measure atmospheric water vapor greenhouse gas from multi-satellite remotely sensed observations under all weather conditions | en_US |
| dcterms.abstract | Atmospheric water vapor is the largest natural greenhouse gas and plays a crucially important role in the Earth’s climate and weather systems. It also has a critical role in energy transport, water circulation, and radiation absorption. Water vapor varies considerably in both spatial and temporal dimensions. It is important to measure and monitor the spatial variation and temporal change of the water vapor, both locally and globally. Precise water vapor observations in the representative spatial-temporal domains play a vital role in climate change monitoring, numerical weather prediction, and other related applications. | en_US |
| dcterms.abstract | There have been a large number of observational techniques to gather water vapor from ground-based and satellite-based data. Unlike ground-based observations, satellite remotely sensed measurements provide a unique technique to observe the spatiotemporal distribution of atmospheric water vapor with proper spatial and temporal resolutions in both local and global areas. However, satellite-sensed water vapor data have two limitations: (1) the accuracy is low compared to ground-based observation techniques such as radiosonde and Global Navigation Satellite System (GNSS); (2) the accuracy is even degraded when observations are made under cloudy sky conditions. For these reasons, the water vapor data records are not continuous in the temporal domain and/or spatial domain, only limitedly available under clear sky conditions. This considerably limits the application of satellite-based water vapor data in climate change, weather forecasting, and many other applications. It is thus crucial to technically improve the performance of satellite-derived operational water vapor data using retrieval/calibration methods, particularly for all weather conditions. Little research has been reported on enhancing all-weather water vapor satellite observations, while the previous work made several efforts to improve satellite-sensed water vapor data under clear sky conditions. | en_US |
| dcterms.abstract | We propose a Novel Neural Network-based Calibration Method (N3CM) to improve the observational performance of operational all-weather water vapor products from Aura-borne Ozone Monitoring Instrument (OMI)-sensed visible measurements. The N3CM-estimated water vapor data significantly outperform operational OMI-derived water vapor estimates under all weather conditions, showing R2 = 0.83, Root-Mean-Square Error (RMSE) = 5.32 mm, and Mean Bias (MB) = -0.19 mm with GNSS measurements and R2 = 0.79, RMSE = 5.19 mm, and MB = -0.97 mm with radiosonde measurements. The RMSE of operational OMI-derived water vapor estimates is overall reduced by ~90%, illustrating the capability and effectiveness of the N3CM. The newly proposed N3CM method could be a promising technique to calibrate the quality of OMI-sensed water vapor observations, which could be extended to other OMI-like sensors. | en_US |
| dcterms.abstract | In addition, a practical retrieval algorithm based on machine learning is developed to retrieve enhanced all-weather water vapor estimates from near-infrared observations of the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor, which is different from previously published algorithms focusing on clear sky conditions. The weighted mean water vapor retrievals present a better performance than the water vapor data calculated using one single absorption channel, which exhibit a diminution in RMSE of ~57% and ~47% for all-weather and confident-clear conditions, respectively. The retrieval algorithm can considerably improve the water vapor retrieval accuracy from MODIS near-infrared bands, regardless of the sky weather conditions. | en_US |
| dcterms.abstract | We also propose a novel machine learning-based calibration scheme to improve the observational quality of MODIS-sensed operational infrared water vapor measurements under all weather conditions. The newly calibrated water vapor data exhibit a better consistency with reference water vapor measurements (R2 = 0.88–0.94, RMSE = 2.79–4.08 mm, and MB = 0.16–0.52 mm), compared with operational water vapor observations (R2 = 0.83, RMSE = 4.93 mm, and MB = -1.14 mm). The RMSE of all-weather water vapor from MODIS and GNSS is reduced by 44.95% in the daytime and 42.12% in the nighttime. The newly proposed machine learning-based calibration scheme could also be suited to other satellite-derived infrared water vapor products. | en_US |
| dcterms.abstract | On the other hand, a feasible retrieval algorithm based on machine learning is presented to derive all-weather water vapor estimates over land from the Global Precipitation Measurement (GPM) Microwave Imager (GMI)-sensed microwave observations, without the employment of surface emissivity and radiative transfer that are traditionally utilized in previously published retrieval approaches. The GMI-derived land-region water vapor data agree well with reference water vapor estimates from GNSS and radiosonde, with R2 = 0.78, RMSE = 5.67 mm, and MB = 0.07 mm (GNSS-based reference) and R2 = 0.86, RMSE = 5.89 mm, and MB = -0.36 mm (radiosonde-based reference). The retrieval accuracy of newly derived water vapor data records is comparable to operational clear-sky near-infrared water vapor satellite observations, such as the MODIS near-infrared water vapor product. | en_US |
| dcterms.abstract | In summary, we have done several works to technically enhance the all-weather performance of satellite-based operational water vapor estimates using retrieval/calibration methods, based on machine learning. The enhanced all-weather water vapor satellite measurements could play a more important role in the Earth’s climate, weather, and other related fields. | en_US |
| dcterms.extent | xxi, 178 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2024 | en_US |
| dcterms.educationalLevel | Ph.D. | en_US |
| dcterms.educationalLevel | All Doctorate | en_US |
| dcterms.LCSH | Water vapor, Atmospheric | en_US |
| dcterms.LCSH | Satellite meteorology | en_US |
| dcterms.LCSH | Remote sensing | en_US |
| dcterms.LCSH | Machine learning | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.accessRights | open access | en_US |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/13420