Open AccessVerification of Convection-Allowing Model Ensemble Analyses of Near-Storm Environments Using MPEX Upsonde Observations
Author(s) -
Christopher A. Kerr,
David J. Stensrud,
Xuguang Wang
Publication year - 2017
Publication title -
monthly weather review
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.862
H-Index - 179
eISSN - 1520-0493
pISSN - 0027-0644
DOI - 10.1175/mwr-d-16-0287.1
Subject(s) - storm , mesoscale meteorology , inflow , convective storm detection , predictability , ensemble kalman filter , environmental science , data assimilation , climatology , outflow , convective available potential energy , meteorology , convection , ensemble forecasting , severe weather , atmospheric sciences , geology , kalman filter , statistics , geography , mathematics , extended kalman filter
The Mesoscale Predictability Experiment (MPEX) conducted during the spring of 2013 included frequent coordinated sampling of near-storm environments via upsondes. These unique observations were taken to better understand the upscale effects of deep convection on the environment, and are used to validate the accuracy of convection-allowing (Δ x = 3 km) model ensemble analyses. A 36-member ensemble was created with physics diversity using the Weather Research and Forecasting Model, and observations were assimilated via the Data Assimilation Research Testbed using an ensemble adjustment Kalman filter. A 4-day sequence of convective events from 28 to 31 May 2013 in the south-central United States was analyzed by assimilating Doppler radar and conventional observations. No MPEX upsonde observations were assimilated. Since the ensemble mean analyses produce an accurate depiction of the storms, the MPEX observations are used to verify the accuracy of the analyses of the near-storm environment. A total of 81 upsondes were released over the 4-day period, sampling different regions of near-storm environments including storm inflow, outflow, and anvil. The MPEX observations reveal modest analysis errors overall when considering all samples, although specific environmental regions reveal larger errors in some state fields. The ensemble analyses underestimate cold pool depth, and storm inflow meridional winds have a pronounced northerly bias that results from an underprediction of inflow wind speed magnitude. Most bias distributions are Gaussian-like, with a few being bimodal owing to systematic biases of certain state fields and environmental regions.