Author: | Gong, Yangzhao |
Title: | Assimilating multiple-sensor water vapor data into Numerical Weather Prediction (NWP) model to improve the weather forecasting performances |
Advisors: | Liu, Zhizhao George (LSGI) |
Degree: | Ph.D. |
Year: | 2024 |
Department: | Department of Land Surveying and Geo-Informatics |
Pages: | xxv, 145 pages : color illustrations |
Language: | English |
Abstract: | Water vapor accounts for a small part of atmosphere but plays a crucial role in the formation and evolution of different weather processes. As a key meteorological parameter in Numerical Weather Prediction (NWP) model, assimilation of external accurate water vapor data can improve the initial field of NWP model, and further lead to a better forecasting performance. In assimilation community, the satellite remote sensing water vapor data are routinely assimilated in the type of radiance. Additionally, most previous studies assimilated water vapor radiance data from infrared (IR) and microwave bands, while the impact of assimilating near-infrared (NIR) data was rarely explored. In this thesis, water vapor data in the type of Precipitable Water Vapor (PWV) retrieval from the ground-based Global Navigation Satellite System (GNSS) and two satellite-based NIR sensors, i.e., the Ocean and Land Colour Instrument (OLCI) onboard Sentinel-3 satellites and the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard Terra satellite, are assimilated into the state-of-the-art mesoscale NWP model, i.e., the Weather Research and Forecasting (WRF) model. The contributions of various water vapor data to weather forecasting have been comprehensively investigated and analyzed. The impact of assimilation of GNSS PWV on the performance of WRF model over the South China region for April 01, 2020 to May 31, 2020 is firstly investigated. PWV observations derived from 213 GNSS stations are assimilated into the WRF model. Additionally, meteorological profiles measured by 23 radiosonde stations are also incorporated into the WRF model to further improve weather forecasting capacity. PWV observations derived from 170 independent GNSS (having not been used in assimilation) and rainfall data recorded by 648 surface meteorological stations are used to evaluate the WRF forecasting performance in the aspects of PWV and rainfall, respectively. The results indicate that, for the first 6 h after data assimilation, assimilation of both GNSS PWV and radiosonde profiles improves the PWV forecasting accuracy by up to 14.8%. The equitable threat score (ETS) values of 6-h accumulated rainfall forecast are improved by up to 0.036, 0.057, 0.047, 0.044, and 0.039 for rainfall thresholds of 0.1 mm, 5 mm, 10 mm, 15 mm, and 20 mm, respectively, after assimilation. The benefits of assimilating satellite-based NIR water vapor have also been examined. Two assimilation schemes are adopted to assimilate Sentinel-3 OLCI NIR clear-sky PWV and all-sky PWV, respectively, in the South China area. Additionally, a no data assimilation scheme is adopted as the control scheme. For each scheme, we have carried out 14 forecasts in March 2020 and 14 forecasts in June 2020. The weather condition in June 2020 is much wetter than March 2020. Generally, assimilating Sentinel-3 OLCI PWV improves the WRF forecasting performance, particularly for June 2020. Assimilation of all-sky PWV outperforms assimilation of clear-sky PWV. The comparison results with radiosonde profiles show that assimilating Sentinel-3 OLCI PWV appreciably corrects the bias of WRF water vapor mixing ratio forecasting results for June 2020. The rainfall validation results indicate that both assimilation schemes show a positive impact in June 2020, but a neutral impact in March 2020. For June 2020, assimilating Sentinel-3 OLCI all-sky PWV improves rainfall forecast ETS score by 0.018, while the rainfall ETS score improvement is 0.007 after assimilating OLCI clear-sky PWV. Additionally, assimilating Sentinel-3 OLCI PWV can modify the WRF moisture field, which further improves the rainfall spatial pattern. Better weather forecasting results are expected when more accurate water vapor data are assimilated. To improve the satellite-based NIR PWV accuracy, the Sentinel-3 OLCI and Terra MODIS NIR PWV are calibrated by GNSS PWV using the Back Propagation Neural Network (BPNN) method. The advantages of assimilating calibrated PWV have been demonstrated for a heavy rainfall case study of May 20 to 21, 2020 at Guangzhou. The results indicate that, for the area within 500 km to Guangzhou, after assimilating OLCI PWV, the first 12-h accumulated rainfall forecast ETS scores are improved by up to 0.029, 0.008, 0.007, 0.022, and 0.020 for rainfall thresholds of 0.1 mm, 5 mm, 10 mm, 20 mm, and 30 mm, respectively. After assimilating MODIS PWV, the ETS score improvements are up to 0.004, 0.035, 0.010, 0.011, and 0.042, for rainfall thresholds of 0.1 mm, 5 mm, 10 mm, 20 mm, and 30 mm, respectively. Additionally, for OLCI, assimilation of calibrated PWV shows higher rainfall forecast scores than assimilation of raw PWV for rainfall thresholds of 5 mm, 10 mm, and 30 mm. For MODIS, calibrated PWV shows a larger positive impact on rainfall forecast than that for raw PWV when rainfall thresholds are set to 5 mm and 10 mm. Finally, the tropospheric wet delay corrections retrieved from WRF water vapor forecasts are applied to augment GNSS Precise Point Positioning (PPP) performance during two observation periods with two different weather conditions, i.e., period 1: March 01 to 14, 2020 (average PWV: 23.5 kg/m2) and period 2: June 02 to 15, 2020 (flooding weather with average PWV: 55.6 kg/m2), over the South China area. PWV data from 277 and 263 GNSS stations are assimilated into the WRF model to enhance the WRF water vapor forecasting capability for period 1 and period 2, respectively. Wet delay corrections from two different WRF schemes, i.e., WRF no assimilation and WRF with assimilation of GNSS PWV, are used to augment the PPP. Compared with traditional PPP, for static mode, WRF-enhanced PPP schemes improve the vertical positioning accuracy by 14.6% and 33.7%, for period 1 and period 2, respectively. The corresponding static PPP convergence time can be shortened by 41.8% and 25.0% for period 1 and period 2, respectively. For kinematic mode, the positioning accuracy improvements in the up component are 13.8% and 19.0% for period 1 and period 2, respectively. WRF-enhanced kinematic PPP schemes shorten convergence time in the up component by up to 8.2% for period 1. |
Rights: | All rights reserved |
Access: | open access |
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/200/13077