GFS vs. GFS Parallel Model Verification for the May 1st–2nd Nashville, TN Flood

The Global Forecast System (GFS) and GFSP (parallel version of GFS with latest upgrades) models are examined for a major flood event that occurred in Nashville, TN on May 1st – May 2nd 2010. The new GFSP is higher horizontal resolution (from 35-km to 27-km) and contains an upgrade in the model physics package. In this study, we examine the advantages and disadvantages of the new GFSP model and evaluate both models ability to forecast the synoptic forcing from the day 7 to day 1 forecast period as well as compare the verification of 48-hr rainfall totals that capture the full extent of this rare event. In this flood event, many locations in western Tennessee and southern Kentucky received 8-10 inches of rainfall and the loss of life (24 deaths) is unprecedented in the history of Tennessee flood events. Precipitation statistics are also used to verify the model runs. The results show that the GFSP provided a more accurate forecast of the areal locations that would be affected by the heavy rainfall compared to than the GFS. However, precipitation statistics show that the GFS performed better for this event due to the old physics package producing more convection in Southwest Tennessee.