Michael Baldwin
Cooperative Institute for
Mesoscale Meteorological Studies,
University of Oklahoma. Also affiliated with NOAA/NSSL and
NOAA/NWS/NCEP/SPC.
1313 Halley Cir
Norman, OK 73069
(405)579-0732
fax:(405)579-0700
email: Mike.Baldwin@noaa.gov
Spatial fields, such as rainfall observed by radar or predicted by a sophisticated NWP model, may contain high-amplitude, small-scale features. In verifying such fields, small forecast errors in phase, displacement, or timing can produce very large differences between forecast and observed variables at specific locations. This could lead to poor traditional measures of forecast accuracy, particularly when compared to forecasts that lack spatial detail. Despite the potential for large errors at specific points in space and time, forecasts that do contain realistic spatial structure may be of considerable value to certain users. Consequently, the value of "realistic" forecasts will likely not be accurately expressed when using traditional "point-to-point" verification methods. In order to obtain more useful information on the accuracy of forecasts that contain high-amplitude, small-scale features, the paradigm of "point-to-point" verification is expanded to the verification of "events" or "objects", which are defined as meteorological features of phenomena. The main challenge is to develop objective methods to characterize forecast and observed fields by identifying regions within the spatial fields that contain similar attributes. Collection of other information related to the characterization of the forecast, which could be determined subjectively (e.g., forecaster confidence) or objectively (e.g., veg. fraction), can also be used to increase the usefulness of verification information. New measures of accuracy can then be obtained by examining, for example, the similarity of forecast and observed events.