Extended Abstract (1.4M)P1.41
Improvement of LAPS wind analysis by including background error statistics
Ok-Yeon Kim, Pukyung National University, Busan, Korea; and C. Lu, S. Albers, J. A. McGinley, and J. H. Oh
The NOAA Earth System Research Laboratory’s Local Analysis and Prediction System (LAPS) has been ported and used in many numerical weather forecast centers around the world. The system has consistently been producing favorable and realistic meteorological analyses (with initial diabatic and cloudy atmospheric conditions) and ensuing forecasts. Over the years, LAPS has continued to evolve and upgrade to include more and better schemes and capabilities. This study reports a new work that has been done to include an ensemble-based estimate of background error statistics and to improve the weighting of the model background information in the LAPS data assimilation system.
The background error statistics is calculated from a set of model forecasts initialized at different times, but validated at the same time. This set of forecasts forms a time-phased ensemble system, and can capture “flow-dependent” (time-evolving) background error statistics. Although both error variance and covariance for all forecast variables can be calculated using these ensemble members, we currently only ingest background wind error variance in the LAPS data assimilation system to see if wind analysis is improved.
We verify the wind analysis by comparing the current LAPS analysis (with constant weighting of background errors) with this upgrade LAPS analysis (with ingested background wind variance as the weighting), and with both against the withholding RAOBs. The verification did show a significant improvement of wind-speed analysis for most pressure levels, and also an improvement of wind-direction analysis in lower levels. The analysis for wind direction at mid- and upper-levels showed a mixed picture: some levels were improved and some were deteriorated. We are still investigating this problem.
Poster Session 1, Monday Poster Viewing
Monday, 25 June 2007, 4:35 PM-6:30 PM, Summit C
Previous paper Next paper