Climatological aspects of observed and simulated precipitation entities

Precipitation remains one of the most important weather events to the general public. Forecasters have long relied on numerical weather prediction (NWP) models to provide guidance concerning both rainfall intensities and locations. Much emphasis has been placed on the NWP guidance as well as the associated verification of these forecasts. In particular, model skill is often assessed solely by the verification scores of the precipitation forecasts. The verification can be performed at relatively fine granularities in time which allows users, as well as model developers, to assess situations where the model may be performing very well or very poorly. Traditional verification methods treat NWP forecasts as gridded fields that are verified either at station locations, or at a series of grid points without ever isolating synoptic or even mesoscale features that may exist. Techniques are beginning to emerge that allow precipitation forecasts to be viewed instead as a set of objects, each with their own attributes, which can be evaluated differently than could be done in the past.

This study will focus on evaluating gridded fields of precipitation from a slightly different view than traditional forecast verification; one based on entities, or objects, found in the data. Further, this study will not focus on verifying a set of forecasts against a matched set of observations for the sake of calculating traditional verification meausres. Data from three sources will be compared in a climatological setting to learn more about the behavior of precipitation objects in observational and NWP model data. The data sources will include the Stage IV analyses, the 20 km RUC, and the 12 km Eta for the entire year of 2004. These data sources will allow us to compare the two models against one another, to determine if the modeled behavior changes quite significantly between the two, given that they each utilize markedly different grid resolutions and convective parameterizations from each other. The models will also be compared to observed precipitation objects and systematic differences cataloged and interrogated. Such a comparison will prove useful to forecasters who must utilize NWP guidance on a regular basis.