Abstract
A 3-h intermittent data assimilation system (Mesoscale Analysis and Prediction System—MAPS) configured in isentropic coordinates was developed and implemented in real-time operation. The major components of the system are data ingest, objective quality control of the observation, objective analysis, and a primitive equation forecast model, all using isentropic coordinates to take advantage of the improved resolution near frontal zones and greater spatial coherence of data that this coordinate system provides. Each 3-h forecast becomes the background for the subsequent analysis; in this manner, a four-dimensional set of observations can be assimilated.
The primary asynoptic data source used in current real-time operation of this system is air-craft data, most of it automated. Data from wind profilers, surface observations, and radiosondes are also included in MAPS.
Statistics were collected over the last half of 1989 and into 1990 to study the performance of MAPS and compare it with that of the Regional Analysis and Forecast System (RAFS), which is run operationally at the National Meteorological Center (NMC). Analyses generally fit mandatory-level observations more closely in MAPS than in RAFS. Three-hour forecasts from MAPS, incorporating asynoptic aircraft reports, improve on 12-h MAPS forecasts valid at the same time for all levels and variables, and also improve on 12-h RAFS forecasts of upper-level winds. This result is due to the quality and volume of the aircraft data as well as the effectiveness of the isentropic data assimilation used. Forecast fields at other levels are slightly poorer than those from RAFS. This may be largely due to the lack of diabatic and boundary-layer physics for the MAPS model used in this test period.
