Graduate Exam Abstract

Abstract: Precipitation estimation plays a vital role in weather forecasting and climate studies, offering essential insights into the distribution, amount, and timing of rainfall. Accurate precipitation estimation is also critical to understanding the water cycle and Earth ecosystems at different scales and to supporting human activities like agriculture, transportation, and energy production. Compared to other instruments such as rain gauges and satellites, radars have unique advantages in observing precipitation at high spatial and temporal resolutions. Recent studies have revealed that deep learning techniques can reduce parameterization errors and improve estimation accuracy compared to traditional radar rainfall estimation methods . However, developing a model that can be applied broadly across different precipitation regimes presents a significant challenge. To overcome this issue, this research utilizes crowdsourced data and a convolutional neural network equipped with residual blocks to improve polarimetric radar quantitative precipitation estimation (QPE). This approach allows for the transfer of knowledge from one location to others with varying precipitation characteristics. Experimental results confirm the effectiveness of this method, demonstrating its advantages over traditional fixed-parameter rainfall algorithms.
Such improvement provides considerable promise in enhanced hydrological and meteorological applications.

Adviser: Haonan Chen
Co-Adviser: N/A
Non-ECE Member: Haonan Wang, Department of Statistics
Member 3: V. Chandrasekar, Electrical and Computer Engineering
Addional Members: N/A

Publications:
Wang, L., H. Chen, R. Cifelli and Z. Li, 2023: Improving surface rainfall mapping in complex terrain regions through lowering the minimum scan elevation angle of operational weather radar. IEEE Geoscience and Remote Sensing Letters, 20, 1-5, Art no. 1001205

L. Wang and H. Chen, “Quantifying the hydrometeorological impacts of lowering operational weather radar scan elevation angle,” Progress In Electromagnetics Research Symposium (PIERS2021), Hangzhou, China, 21 -25 November 2021.

L. Wang, H. Chen, and Z. Li, “Impacts of WSR-88D supplemental lower elevation angles on quantitative precipitation estimation,” USNC-URSI National Radio Science Meeting, Boulder, CO, 4-8 January 2022.

Chen, H., L. Wang, G. Pratt, W. Liao, “Polarimetric radar-based rainfall estimation through adaptive learning with multi-source data from the NOAA meteorological assimilation data ingest system,” IEEE International Geoscience and Remote Sensing Symposium (IGARSS2022), Virtual and Kuala Lumpur, Malaysia, July 17-22, 2022.

L. Wang, and H. Chen, “Machine Learning for Polarimetric Radar Quantitative Precipitation Estimation,” 2023 USNC-URSI National Radio Science Meeting, Boulder, CO, USA, 10–14 January 2023.

H. Chen, L. Wang, S. Yao, J. Xu, D. Baral, and L. Xu, “Improving hydrometeorology applications of WSR-88D in a complex urban environment,” the 19th Asia Oceania Geosciences Society (AOGS) Annual Meeting, Online, 1-5 August 2022.

H. Chen, L. Wang, E. Thompson, “Interpretable Deep Learning for Polarimetric Radar Rainfall Estimation,” the 103rd AMS Annual Meeting, Denver, CO, USA, 8–12 January 2023.

H. Chen, L. Wang, and E. Thompson, 2023: Interpretable deep learning for polarimetric radar rainfall estimation. 40th Conference on Radar Meteorology, Minneapolis, MN, 28 August – 01 September 2023.

Program of Study:
Intro to Robot Program/Sim (ECE 455)
Fundamentls-Robot Mech&Contrls (MECH 564)
Electrical Energy Technologies (ECE 465)
Manycore Sys Machine Learning (ECE 558)
Principles-Digital Communictns (ECE 614)
Overview of Sys Engr Processes (ENGR 530)
Engineering Risk Analysis (ENGR 531)
Thesis (ECE 699)