Eric Loken-March 31

Resolution and Physics Sensitivities in Convection-Allowing Models

Convection-allowing models (CAMs), which typically have horizontal grid spacing of 4-km or less, do not require convective parameterization to resolve the dominant circulations in mid-latitude mesoscale convective systems. While previous research has generally agreed that CAMs provide greater quality and value relative to non-CAMs, the benefits of reducing the horizontal grid spacing of a CAM beyond 4-km are less clear. Moreover, it is unknown how the forecast quality and value of a CAM ensemble compares with that of a finer-resolution deterministic CAM.

The first part of this talk uses data from 63 days of the 2010 and 2011 NOAA Hazardous Weather Testbed Spring Forecasting Experiments (HWT SFEs) to compare next-day probabilistic severe weather forecasts from three Advanced Research Weather Research and Forecasting (ARW-WRF) model configurations: a 1-km deterministic CAM; an equivalently-configured 4-km deterministic CAM; and an 11-member, 4-km CAM ensemble. For all three model configurations, next-day severe weather forecasts are produced by spatially smoothing large values of forecast updraft helicity (UH). Relative operating characteristic (ROC) curves, attributes diagrams, and performance diagrams are analyzed for the aggregate 63-day dataset as well as individual days. No significant difference exists between the 4- and 1-km deterministic forecasts’ aggregate area under the ROC curve (AUC), while a weakly significant difference exists between the 4-km deterministic and 4-km ensemble aggregate AUC. These results suggest that creating a CAM ensemble may improve next-day severe weather forecasts more than reducing the horizontal grid spacing of a deterministic CAM from 4-km to 1-km.

The second part of this talk uses data from the 2016 HWT SFE and Community Leveraged Unified Experiment (CLUE) to compare the skill of mixed- and single-physics CAM ensembles’ next-day severe weather forecasts. No significant difference is found between the mixed- and single-physics ensembles’ AUC or fractions skill score, suggesting that a mixed-physics ensemble does not produce significantly more skillful UH-derived next-day probabilistic severe weather forecasts compared to a single-physics ensemble.