Processes and predictability in recent widespread heavy rain and flood events

This study examines several recent instances of widespread heavy rainfall over 5--7 day periods in the central United States. All of these events resulted in destructive flooding and had numerous impacts on local and regional scales. First, the predictability of these events by global forecast ensembles is assessed by analyzing how far in advance the ensembles identified the threat of large-scale heavy rains, and by identifying the atmospheric processes responsible for limiting the predictability in some cases. For example, in an event with strong synoptic forcing that occurred in mid-March 2008 in parts of the Mississippi and Ohio valleys, the ensemble forecasts strongly suggested the possibility for a long-lived, widespread rain event in the correct area over five days before the event began. In contrast, the record rainfalls in the southern Plains in late June 2007, which were related to mesoscale forcing, had substantially shorter lead times for successful prediction in the ensemble forecasts.

The important processes in this event, which occurred from 25 June--2 July 2007, will then be discussed in greater detail. The situation began as a mesoscale convective vortex (MCV) intensified and grew upscale into a long-lived warm-core vortex over Oklahoma and Texas. The lifting associated with this circulation helped to initiate deep convection each evening, which in turn reinvigorated the vortex overnight; this process continued for several days. The circulation remained remarkably stationary from a large-scale perspective for over a week, and the deep convection associated with it produced in excess of 100 mm (4 in) of rain over much of southern Kansas, Oklahoma, and northern Texas, with smaller regions receiving over 400 mm (16 in) over this 7-day period. Observations and WRF simulations of this event show that deep convective activity was integral to the intensification, maintenance, and movement of the mesoscale circulation and associated rainfall. Deep convection led strong divergence and ridging aloft, which reduced the already-weak shear over the vortex and created conditions favorable for it to persist. These upper-level effects of the convection also altered the large-scale flow by deflecting an approaching upper-level wave, which allowed the vortex to remain nearly stationary for an extended period. Ensemble forecasts show that once the mesoscale vortex is represented in the model initial conditions, the members converge toward an accurate solution, but at earlier initialization times when the vortex has not yet developed, model forecasts provide little indication that a long-lived extreme rain event will take place.