High Spatiotemporal Resolution Quantitative Precipitation Estimation over the United Arab Emirates

Conventional rain gauge measurements have been widely used as reference information in weather and climatological studies. However, in-situ gauge efficacy is rather limited in capturing spatial variability of precipitation, particularly in remote, sparse, and unevenly gauge-distributed mountainous areas. To address these challenges, we implement two techniques to retrieve high spatiotemporal Quantitative Precipitation Estimation (QPE) at 4 km spatial and hourly temporal resolution using only gauge observation over the United Arab Emirates (UAE). The first is the Mountain Mapper (MM), which relies on Parameter-elevation Regressions on Independent Slopes Model (PRISM) to group gauges and interpolate their reports onto specific locations. The second is an end-to-end Neural Network (NN)-based model that in near-real-time incorporates rain-gauge measurements, spatiotemporal and other informative environmental variables to intelligently map QPE from the gauge locations to a gridded domain. This presentation demonstrates the applicability of both techniques and delivers comprehensive evaluations of gridded QPE outputs over the semi-arid UAE, a region with limited rain gauge coverage. By leveraging gauge-based precipitation products, these methods overcome the uneven distribution of gauges and substantially improve the representation of precipitation patterns across the terrain. This work marks one of the first attempts to establish a reference/ground-truth-gridded QPE over the region for operational weather forecasts and analyses.