A correlation analysis pattern recognition technique called TREC (for tracking radar echoes by correlation) is developed which is capable of determining motions within storms. The limitations and capabilities of TREC were explored using two approaches. First, the properties of real radar echoes (reflectivities, reflectivity gradients, and temporal and spatial decorrelations) were studied for their effect on the results of TREC. Radar echoes contain a variety of scales of reflectivity patterns which persist temporally and spatially quite well; useful results are possible to about 100 km or more, depending upon the motion scale size and radar resolution available. Second, flow fields and reflectivity fields were modeled to examine those properties of TREC that could not be studied using real storms. This study showed that the success of TREC depends upon the interrelationships between the flow and reflectivity field characteristics as well as the parameters of TREC processing. A comparison of TREC results with data from a triple-Doppler radar system showed that TREC vectors are apparently often generated either above or below the level being examined. A number of suggestions are given for applying TREC to other storm situations and other radars. Finally, the advantages and disadvantages of TREC and the overall conclusions are summarized.
