| description abstract | Advances in remote sensing from earth- and spaceborne systems, expanded in situ observation networks, and increased low-cost computer capability will allow an unprecedented view of mesoscale weather systems from the local weather office. However, the volume of data from these new instruments, the nonconventional quantities measured, and the need for a frequent operational cycle require development of systems to translate this information into products aimed specifically at aiding the forecaster in 0- to 6-h prediction. In northeast Colorado an observing network now exists that is similar to those that a local weather office may see within 5?7 years. With GOES and TIROS satellites, Doppler radar, wind profilers, and surface mesonet stations, a unique opportunity exists to explore the use of such data in nowcasting weather phenomena. The scheme, called LAPS (the Local Analysis and Prediction System), objectively analyzes data on a high-resolution, three-dimensional grid. The analysed fields are used to generate mesoscale forecast products aimed at specific local forecast problems. An experiment conducted in the summer of 1989 sought to test the use of a preconvective index on the difficult problem of convective rain forecasting. The index was configured from surface-based lifted index and kinematically diagnosed vertical motion. The index involved a number of LAPS-derived meteorological fields and the results of the test measured in some sense the quality of those fields. Using radar reflectivity to verify the occurrence or nonoccurrence of convective precipitation, forecasts were issued for three time periods on each of 62 exercise days. The results indicated that the index was significantly better than persistence over a range of echo intensities. Skill scores computed from contingency tables indicated that the index had substantial skill in forecasting light convective precipitation with 1- to 3-h lead time. Less skill was shown for heavier convective showers. The skill of the index did not depend strongly on the density of surface data, but was negatively influenced by mountainous terrain. | |
