Short course on AI/ML for Radar Quantitative Precipitation Estimation and Nowcasting
Haonan Chen (Colorado State University)
V. Chandrasekar (Chandra) (Colorado State University)
Slides from the short course can be found HERE
It is long believed that polarimetric radar observations are rich in information content and conventional methods of utilizing them have only been able to extract part of the information due to the nature of the analytical tools. For example, we still rely on empirical algorithms such as Z-R relations or other polarimetric relations to estimate rainfall. These parametric relations are generally derived using power- law fitting between rainfall rates and radar observables, and they often need to be adjusted since raindrop size distribution (DSD) varies in different precipitation regimes or even within a single storm system. In addition, it is challenging to remove the inherent parameterization error in such “fixed” parametric relations, even if the relations are fine-tuned with local DSD data. The multi-parameter multi-dimensional radar observations are ideal candidates for modern artificial intelligence (AI) applications in extracting quantitative precipitation information. This short course will familiarize participants with fundamentals of AI and machine learning. A particular emphasis will be placed on quantitative precipitation estimation and short-term prediction (nowcasting) and different aspects of weather radar data sciences.