Barbara
G. Brown, Randy Bullock, Chris Davis, Kevin Manning, Rebecca Morss, and Cindy
Mueller
National
Center for Atmospheric Research
Boulder
CO
A number of new verification approaches for QPFs and convective forecasts have recently been under development, in an effort to provide more meaningful information from verification studies. Traditional methods suffer numerous flaws, which minimize their usefulness. The new approach described here characterizes forecast and observed precipitation/convective regions in a natural way as objects with particular types of characteristics. This approach allows direct comparison of fundamental attributes of the forecasts and observations, such as the centroid location, axis orientation, shape, and the underlying distribution of intensities. Advantages of this approach include the following: (a) it is diagnostic (i.e., it provides information that can be used to understand and improve the forecasts); (b) it is sensitive to – and can measure – location and timing errors; (c) it can be tailored to provide information needed by both forecast developers and forecast users; (d) it can be applied to a variety of different types of QPFs, including NWP output and mesoscale nowcasting systems; and (e) it is less sensitive than standard verification approaches to issues of scale and other attributes of the observations. Issues that were faced in the process of developing the approach (e.g., setting rules for combining objects, selecting methods to define objects) will be described. These issues are demonstrated through a few examples of QPFs produced by the Weather Research and Forecasting (WRF) model in July-August 2001.