Comparison of Operational Model Analyses with MPEX Dropsondes

The field phase of the Mesoscale Predictability Experiment (MPEX) was conducted over the Intermountain West and adjacent High Plains during May–June 2013. A goal of MPEX was to test the impact of increased observation density on short-range (6–24 hours) convective-scale forecasts by releasing approximately 30 dropsondes spaced ~75–200 km apart during the early morning hours using the NSF/NCAR GV aircraft. The evolution of convective-scale forecasts from high-resolution regional models, such as WRF-ARW, are influenced by the analysis fields used as initial conditions. Many of the short-range explicit convective numerical forecasts generated in real-time in operational and academic settings use operational model analysis fields, such as the NCEP Global Forecast System (GFS) analyses. The aim of this presentation is to use the additional subsynoptic-scale observations provided by the MPEX dropsondes to examine and identify any persistent errors in the operational model analyses during the MPEX field phase. We will then investigate how these model analysis errors may influence short-range explicit convective-scale WRF-ARW forecasts that use operational analyses as initial conditions.

Preliminarily, examination of a subset of MPEX cases reveals that the NCEP GFS analyses can be too moist in low- to mid-levels over western Texas and eastern New Mexico, represented on average as a ~+3–5 K anomaly in the 2–6 km layer mean equivalent potential temperature at 1200 UTC. For instance, inspection of a high-resolution ARW forecast initialized with the GFS analyses at 1200 UTC 28 May 2013 revealed that a moist error (+7 K equivalent potential temperature error in the 2–6 km layer) in the GFS analyses over west Texas may have contributed to an overdevelopment of convection over the Texas Panhandle in the 12–24 h forecast, where convection in the ARW forecast grew upscale to a mature mesoscale convective system, but remained as isolated cells in reality. In addition to this case, this presentation will show analysis from other cases during MPEX and will provide general comments on persistent errors in operational analyses.