Challenges in comparing realistic, high-resolution spatial fields from convective-scale grids

An outstanding problem in storm-scale numerical prediction is the determination of useful methods to compare high resolution spatial fields, such as comparing forecast and observations or multiple analyses resulting from an ensemble system. In this paper, several objective techniques will be applied to high-resolution, detailed meteorological fields. Two types of ensemble model forecasts/analyses are generated using data from the 8-9 May 2003 central plains tornado outbreak. First is an ensemble Kalman filter analysis and forecast using a 25 member ensemble covering the central plains. Six-hour, high resolution (2.5 km horizontal resolution) forecasts are employed to explicitly predict the convective evolution. The high resolution runs are generated from an EnKF data assimilation of surface data using the MM5 model at 30 km resolution. Predicted and observed composite radar reflectivities and will be compared. A second perspective focuses analyzing the details of one individual storm. High resolution Doppler and reflectivity data from the NSSL KOUN data are used to create an EnKF analysis of the 8 May 2003 Oklahoma City (Moore) tornadic storm. Here model analyses are compared to an independent set of Doppler and reflectivity data available from the Terminal Doppler Weather Radar system in southwest Moore. The proximity of the TDWR to the storm permits a detailed comparison of model versus actual storm evolution.

Several traditional and new techniques will be used to analyze the "goodness" of the model output versus observations. The availability of ensemble model output for each case presents new opportunities and challenges. New object-oriented techniques will be used to compare the forecast and observed characteristics of a variety of features. A general framework for object-oriented comparison involves object identification, characterization, and comparison. We will report on ongoing research related to meaningful, objective comparisons of spatial fields that contain realistic, high-resolution detail for these emerging ensemble-derived model analyses and forecasts.