A comparison of WRF 3.2 and WRF 3.1.1 forecasts of seven nor'easter events

The recently released WRF model version 3.2 contains a number of additional parameterizations and bug fixes, including a potentially significant change to a commonly used radiation scheme. This study analyzes the effect of these changes in comparison to the previous WRF version 3.1.1 in the context of seven nor'easter events. Both versions of the WRF model were used to create 180-hour forecasts of significant nor'easter events from recent years. Each model was identical aside from the version, using the same GFS model forecasts as model input and identical settings for physical parameterizations. We quantitatively and qualitatively measure the forecast differences between WRF 3.2 and WRF 3.1.1 by comparing WRF model output from seven simulated nor'easter cases that occurred during the cold seasons of 2006–2010. Additionally, the accuracy of each WRF forecast is quantified by comparing WRF model output to GFS model analyses and Stage IV precipitation data, which we consider to be “ground truth.” Preliminary results indicate that both sea-level pressure and 500-hPa geopotential height fields associated with the developing nor'easter lag behind the ground truth similarly in both WRF 3.2 and WRF 3.1.1. No consistent pattern of differences is found between the two WRF versions in the context of these fields. Most of the observed differences can be attributed to varying degrees of phase shift between synoptic-scale weather patterns. At higher resolutions, our data suggest that WRF 3.1.1 is slightly wetter than WRF 3.2 on average.