Quantitative Precipitation Estimation with Phased Array Radars

Abstract

While substantial progress has been made in ground-based quantitative precipitation estimation (QPE) over the past two decades with the introduction of dual-polarized scanning radars, some key questions remain unresolved: (1) how can we accurately capture the rapid temporal and spatial variability of precipitation, especially in convective storms? and (2) how can radar systems better resolve microphysical and dynamical processes that directly influence rainfall estimation accuracy? Addressing these questions to make QPE estimates more reliable has implications for applications including weather and flood forecasting and water-resource management. To this end, the 2017 decadal survey identified key priorities that include quantifying global precipitation rates and phase (rain and snow/ice) at convective and orographic scales sufficient to capture flash flooding and larger-scale impacts, as well as characterizing hydrometeor microphysical processes through measurements of hydrometeor distributions and precipitation rates with an accuracy of approximately 5%. Research using conventional mechanically scanning dish radars with dual polarization has shaped our current approaches for hydrometeor classification, particle size distribution estimation, precipitation typology, and rainfall estimation. However, important gaps remain for capturing rapidly evolving precipitation features, storm structure, and microphysical processes with sufficient spatiotemporal resolution. Since 2022, the Advanced Radar Research Center (ARRC) at the University of Oklahoma (OU) has deployed a prototype all-digital phased array radar (PAR) as part of an effort supported by the National Oceanic and Atmospheric Administration’s (NOAA) National Severe Storm Lab (NSSL) to demonstrate the benefits and practical use of PAR technology for future weather observations. The goal of this work is to examine PAR QPE and to enhance our understanding of the benefits of PAR’s spatiotemporal information content to QPE and the estimation of precipitation process rates. In support of the decadal survey goals, this work presents a supervised learning framework using PAR's denser sampling of the atmosphere that enhances precipitation quantification at convective scales relevant to flash flooding, demonstrates the improvement of radar variable estimation across multiple meteorological timescales, introduces a method for retrieving microphysical process rates, and provides an approach for quantifying uncertainty in both radar variable and rainfall rate estimates. In the pursuit of PAR benefits for global precipitation observation coverage, PAR's ability to bridge satellite observation gaps and target novel insights with spaceborne and ground-based approaches is explored within convection. Multiple radar types and scanning strategies have been used to obtain data with high temporal and/or spatial representativeness in incremental steps throughout this work from 2021 to 2025 to validate the benefits of PAR relevant to QPE. Prior to 2022, micro-rain radar (MRR) observations enabled vertically pointed radar reflectivity retrievals at 24 GHz. Beginning in 2023, Horus began taking Range Height Indicator (RHI) observations of storms, which provided high temporal resolution observations at a single azimuth, yielding RHIs with revisit times of 2 seconds to sample convective precipitation events. A new QPE retrieval framework using supervised learning was developed to integrate proposed altitude and time derivatives of radar variables (referred to hereafter as PAR variables) with information content from the vertical profiles of radar variables afforded by these measurements. These RHI observations, as well as RHI observations taken in 2024 wherein the temporal revisit time was reduced to 0.76 seconds, allowed for the characterization of decorrelation times and temporal sampling errors for two convective events, leading to recommendations for the required sampling timesteps for weather radar QPE. The dense spatiotemporal PAR sampling is explored through the continuity equation applied to condensed water content as a rigorous theoretical framework for the interpretation of precipitation process rates. Observations taken in 2024 further allowed for the exploration of three-dimensional flux terms to create estimates of sources and sink rates (process rates) for liquid water content observed by the radar through rotated imaging scans capable of completing full radar-volume coverage in 24 seconds. The data analysis in this work focuses on exercising PAR variables for the improvement of QPE, characterizing temporal sampling errors afforded by faster temporal updates, and demonstrating the estimation of atmospheric process rates by using three-dimensional radar variables available due to PAR's high spatial sampling. Overall, this work demonstrates the need for fast sampling from PAR to meet radar quality requirements given temporal scales of applications and weather variability. Novel quantification of precipitation process rates has been explored in convective storms, opening a door for interfacing radar and numerical weather modeling. These steps lay the groundwork for key Radar NEXT system specifications and for creating a roadmap for operational QPE that takes into account horizontal and vertical atmospheric fluxes.