Sensitivity of Convective Initiation Prediction to Near-Surface Moisture When Assimilating Radar Refractivity: Impact Tests Using OSSEs

Abstract

he Advanced Regional Prediction System (ARPS) three-dimensional variational (3DVAR) system is enhanced to include the analysis of radar-derived refractivity measurements. These refractivity data are most sensitive to atmospheric moisture content and provide high-resolution information on near-surface moisture that is important to convective initiation (CI) and precipitation forecasting. Observing system simulation experiments (OSSEs) are performed using simulated refractivity data. The impacts of refractivity on CI and subsequent forecasts are investigated in the presence of varying observation error, radar location, data coverage, and different uncertainties in the background field. Cycled refractivity assimilation and forecasts are performed and the results compared to the truth. In addition to the perfect model experiments, imperfect model experiments are performed where the forecasts use the Weather Research and Forecasting (WRF) model instead of the ARPS. A simulation for the 19 May 2010 central plain convection case is used for the OSSEs. It involves a large storm system, large convective available potential energy, and little convective inhibition, allowing for CI along a warm front in northern Oklahoma and ahead of a dryline later to the southwest. Emphasis is placed on the quality of moisture analyses and the subsequent forecasts of CI. Results show the ability of refractivity assimilation to correct low-level moisture errors, leading to improved CI forecasts. Equitable threat scores for reflectivity are generally higher when refractivity data are assimilated. Tests show small sensitivity to increased observational error or ground clutter coverage, and greater sensitivity to the limited data coverage of a single radar.