This case discussed the development and maintenance of GFS grid-scale convection and its effect on the forecast for 12z 20 August 2002 (at T170L42 resolution). Grid-scale convection (a.k.a. a precipitation "bomb") occurs in environments conducive to the development of MCS's; low-level convergence, significant vertical shear, plentiful low-level moisture, and conditional instability. The scale of this convection is on the order of a few adjacent physics grid boxes, or as much as 200-km or more in horizontal extent for the largest convective events. They can last a day or more and travel over tens of degrees of longitude before dying out or being absorbed into larger circulations. Additionally, they may affect the evolution of the larger-scale circulation in such a way as to degrade the larger-scale forecast as well.
Unlike in NCEP's mesoscale models, the horizontal dimensions of these systems in the GFS can result in significant remote impacts on the forecast. This is because the grid-scale convective system may feed back onto the larger mesoscale and even to the synoptic scale, as we saw with the precipitation bombs moving from SD and MN through northern MI into Ontario. The erroneous circulations set up by grid-scale convective systems can result in further deepening through baroclinic processes, thus compounding the initial forecast errors resulting from the spin-up of vorticity beneath the vertical velocity maximum created by the grid-scale convection.
Because the environment is conducive to MCS development, the existence of precipitation bombs in the GFS does give the forecaster some information on QPF. The location and timing of the QPF is often incorrect, however, as we saw here. Heavy precipitation and flash flooding occurred in the Quad Cities, IA CWA 24 hours prior to the grid-scale convective development in the forecast. Additionally, the distortion of the large-mesoscale and synoptic scale environment as a result of the anomalously large diabatic heating resulting from grid-scale convection may render the GFS forecast useless without considerable adjustment to account for the effects of the grid-scale diabatic heating on development of warm frontal waves and on the spin-up of vorticity in the middle troposphere.
It was also shown that this model phenomenon occurs in the new T254L64 GFS that was put into operations on 29 October 2002.