Nowcasting Cloud and Precipitation Fields Using the Remote Atmospheric Processingand Information Display (Rapid) System

摘要

The Remote Atmospheric Processing and Information Display (RAPID) System has beendeveloped at the Geophysics Directorate of Phillips Laboratory to provide an environment for the creation and testing of image processing techniques of remotely sensed data. The first objective of RAPID was to provide nowcasts of cloud and precipitation fields. This is done for cloud fields by tracking and extrapolating contours of infrared bright temperatures from a geostationary satellite. Precipitation fields are forecast by tracking and extrapolating radar reflectivity contours. Currently, there are three techniques in RAPID to extrapolate the future position and shape of contours. The three techniques are the Whole Contour, Segmentation, and Statistical Extrapolation methods. They are all similar in that the contours are represented mathematically, the mathematical features are extrapolated out in time, and forecasted features are used to construct the forecast contour. Tests of the three techniques were conducted using data from the GOES satellite (IR) and the Phillips Laboratory's 10-cm Doppler weather radar. The initial results indicate that, for both satellite and radar data, all three methods do show skill with respect to persistence and produce forecasts that are comparable to each other.