Functional Data Analysis for Rain Rate Statistics

Detailed rain rate frequency statistics (“How many minutes per year does this site receive more than x mm/hr of rain?”) have global applications but are only actually measured at a small number of sites worldwide. State-of-the-art (SOTA) estimation models are based on curve-fitting by expert working groups and involve years of effort to update. This project explores the use of functional data analysis (FDA) as a potential alternative; we suggest that non-parametric functional regression (NPFR) can be combined with satellite radar data to provide similar or better accuracy for suitable applications.