Abstract
A quantitative investigation of the relationship between satellite-derived cloud-top temperature parameters and the detection of intense convective rainfall is described. The area of study is that of the Cooperative Convective Precipitation Experiment (CCOPE), which was held near Miles City, Montana during the summer of 1981. Cloud-top temperatures, derived from the GOES-West operational satellite, were used to calculate a variety of parameters for objectively quantifying the convective intensity of a storm. A dense network of rainfall provided verification of surface rainfall. The cloud-top temperature field and surface rainfall data were processed into equally sized grid domains in order to best depict the individual samples of instantaneous precipitation.
The technique of statistical discriminant analysis was used to determine which combinations of cloud-top temperature parameters best classify rain versus no-rain occurrence using three different rain-rate cutoffs: 1, 4, and 10 mm h−1. Time lags within the 30 min rainfall verification were tested to determine the optimum time delay associated with rainfall reaching the ground.
A total of six storm cases were used to develop and test the statistical models. Discrimination of rain events was found to be most accurate when using a 10 mm h−1 rain-rate cutoff. Use parameters designated as coldest cloud-top temperature, the spatial mean of coldest cloud-top temperature, and change over time of mean coldest cloud-top temperature were found to be the best classifiers of rainfall in this study. Combining both a 10-min time lag (in terms of surface verification) with a 10 mm h−1 rain-rate threshold resulted in classifying over 60% of all rain and no-rain cases correctly.
