Verification of Near-Storm Environment Forecasts Using MPEX Upsonde Observations

Convection-allowing models (CAMs) have been shown to accurately depict convective storm mode and intensity after radar and surface data assimilation.  However, these observations are not sufficient for verification of surrounding mesoscale near-storm environments.  The Mesoscale Predictability Experiment (MPEX) conducted during the spring of 2013 included upsonde operations to sample these near-storm environments.  Several regions of the environments were sampled including inflow, outflow, downstream, and anvil.  A four-day sequence of convective events from 28-31 May 2013 in the southern Great Plains is the focus of this study.  During this period, 81 MPEX upsondes were released. 

Each event is simulated by a 36-member nested ensemble created with physics diversity using the Weather Research and Forecasting (WRF) model while observations, including radar and surface, are assimilated using the Data Assimilation Research Testbed (DART).  The MPEX upsonde observations are not assimilated.  Short-term ensemble forecast (30-60 minutes) errors within near-storm environments are presented.  A previous study of ensemble analysis errors reveals systematic biases of storm inflow wind speed and cold pool depth.  Using these upsonde observations, forecast sensitivities to environmental initial condition errors will be assessed as well.