Summary

Nature of event	Precipitation bull's-eye, often with 6-hour amounts exceeding 3 inches. Model prediction has little skill.
History of problem	First noticed in operational Eta model in July 2001 Preceded 24 July 2001 Eta model upgrade and continued after implementation Occurs in Eta 10-km "threats" runs Occurs less frequently with the new Eta grid-scale precipitation parameterization implemented in late November 2001 Occurs in NSSL experimental Eta runs using Kain-Fritsch convective parameterization Long-standing problem in AVN/MRF model and AFWA MM5 model Event characteristics and evolution and impact on downstream forecast varies by model and is a function of model resolution
How to identify an event	Most of the heavy precipitation falls from the grid-scale scheme when CAPE exists (possibly elevated, not always surface based) Exceptionally high bull's-eye of upward vertical motion is present Remember, the convective parameterization does not directly change model vertical motion Precipitable water maximum coincides with the vertical motion and precipitation maxima Presence of Moist Absolutely Unstable Layer (MAUL) on model sounding MAUL might not be seen at the particular locations/times in model output. Presence of a MAUL is a caution flag, but its absence is not necessarily meaningful
Conditions favoring events in operational 22-km Eta	Strong mesoscale to regional-scale convergence of high-CAPE, high-dewpoint air, particularly along a warm front Scenario favors real MCS with heavy rain
Forecast challenge with operational 22-km Eta	MCS with heavy rain is often expected Model has virtually no skill in locating its heavy rain target Feedback during model event focuses mesoscale forcing in model fields toward model's event location, and thus away from region where MCSs actually do develop Model seems to have some skill in identifying whether or not an MCS will form within a few hundred kilometers of the model's event, but the level of this skill is not clear without further study and forecasters should not rely on it for their precipitation forecast
Forecast impacts in 22-km Eta	During event like the case shown here: A low forms in the lower to middle troposphere along the warm front A mesohigh may form in the lowest model layers Strong vertical motion occurs over the warm front or other model boundary A grid-scale cloud forms on top of the vertical motion maximum and ascends together with the vertical motion maximum A precipitable water maximum forms from deepening of the moist layer A mid-level vorticity maximum forms, typically strongest around 700 hPa Temperatures at 500 hPa clearly show a warm core coinciding with the heaviest precipitation If the model heavy precipitation region is shaped along a line or arc, these features will have that shape rather than a circular shape After an event: The event runs its life cycle, which may end from outrunning the inflow of unstable air or from getting choked off by occluding or convective parameterization cooling in the lower part of the cloud Vertical motion and precipitation rate quickly reduce to moderate values Vorticity and associated vertical motion and precipitation continue to gradually weaken while advecting downstream over the next 12-18 hours, usually becoming insignificant within 24 hours
Forecast impacts in other models	Forecast impacts are scale-dependent, reflecting the extent to which the diabatic heating projects onto large-scale balanced modes In the 10-km Eta "threats" runs, local perturbations such as vorticity in immediate vicinity to the grid-point bull's-eye are more intense, but the larger scale forecast is virtually unaffected by an isolated grid-point blow-up In the T170 AVN, with its physics grid roughly at 80 km at CONUS latitudes, these events often generate surface lows and affect synoptic evolution
Cause/explanation in the 22-km operational Eta	Strong mesoscale convergence and lift of high-CAPE air creates role reversal of convective and grid-scale parameterizations: The model generates a grid-scale cloud which grows upwards, essentially a grid-scale cumulus tower with an updraft core as wide as the model grid spacing times the number of participating adjacent grid columns. A model running with a 200 m grid spacing can make a good, though even then imperfect, simulation of a towering cumulus this way, but it becomes highly unrealistic at 20 km grid spacing! The BMJ convective parameterization in the operational Eta model reproduces the thermodynamic effects of a convective system more so than of an isolated individual thunderstorm, including the important heat and moisture transports which occur in the stratiform region of an MCS A mesoscale model should make a convective system by having the convective parameterization handle narrow deep updrafts while the grid-scale scheme handles the resolved mesoscale components of the MCS At finer resolution, more of the mesoscale flow becomes resolved, so more of the precipitation becomes "grid scale" and less adjustment is needed from the convective parameterization