Comparison of sea–level pressure errors between the U.S. East and West Coasts during the cool seasons of 2000–2005

It is well known that numerical weather prediction (NWP) models have short-term forecast problems along the U.S. West Coast given the relatively data sparse Pacific Ocean. For example, it has been shown that 48-h sea-level pressure (slp) forecasts for mid-latitude cyclones over the northeastern Pacific sometimes exceed 20 mb, with position errors greater than 500 km. However, NWP models also have difficulty in predicting mid-latitude cyclones along the U.S. East Coast (nor'easters) even for relatively short lead times (24- to 48-h). To date, there have been no multi-year studies on the magnitude of these slp errors in NWP models for this region or a comparison of these errors with those typical found on the West Coast. In this study, 24-h and 48-h sea-level pressure forecasts from the National Centers for Environmental Prediction (NCEP) Eta model and a 12-km grid-spacing version of the PSU-NCAR MM5 are verified at six buoys along the East Coast and western Atlantic during the cool seasons (Nov 1-Mar 31) of 2000-2004. These results are then compared to NCEP Eta 24-h and 48-h slp forecasts verified at several West Coast buoys during the same period.

For the East Coast relatively large slp errors (greater than 8 mb) for the 48-h forecasts in the Eta and MM5 models occur on average between 4-6 times each winter offshore. The MM5 and Eta errors are comparable, although the Eta has significantly larger errors in the 2000-2001 cool season because of a well-documented problem in SST initializations along the coast that led to poor nor'easter predictions. Large-scale composites illustrate that the largest East Coast slp errors (> 2 standard deviation) are clearly tied to rapid cyclogenesis along the East coast. As a result, the interannual variability of the number of large errors each winter is dependent on the dominant large-scale flow pattern. For example, the slp errors are largest for the 2002-2003 season in association with several nor-easters. Interestingly, the published results from the West Coast illustrate that the 2002-2003 season had the smallest slp errors due to less cyclone activity on the West Coast. As a result, the East Coast slp errors and standard deviations for the 48-h period are more comparable to the West Coast on average for the 2002-2003 cool season, but for other years the West Coast slp errors are larger than the East Coast slp errors on average.