Quantification of the Effects of Data Denial and Limitation in MM5 Initialization on Forecast Accuracy
Abstract
Using the 3-Dimensional Variational Analysis data assimilation scheme and the (MM5), input observations were denied in three different categories: total, upper air, and surface observation denial. Two control groups were run using all available data as received by AFWA. The main control group used a 6 hours old first guess as a baseline. The data denied test cases and the secondary control group used a 30-hour old first guess because it was not possible to deny data from the first guess. The secondary control group was used to estimate errors resulting from the use of different first guess forecasts between the main control group and the test cases for all forecast times. The analyses show statistically significant differences between the main control group and test cases in almost every instance. However, the surface comparisons provide little evidence of significant meteorological differences due to the relatively small magnitude of the differences in root mean square error and bias. In the upper air analyses, the largest differences emerged at the model initialization time. The magnitude of the differences rapidly diminishes as the solution of the test cases converge toward that of the main control group throughout the forecast period. Errors associated with the use of different initial and boundary conditions account for most of the resulting differences.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 26, 2002
- Accession Number
- ADA404582
Entities
People
- Robert W. Evans
Organizations
- Air Force Institute of Technology