P2.5
Sensitivity of surface temperature analyses to background and observation errors
Daniel Tyndall, University of Utah, Salt Lake City, UT; and J. Horel
A two-dimensional variational surface temperature analysis is developed for a limited domain (4° latitude × 4° longitude) to evaluate an approach to evaluate analysis sensitivity to the specification of observation and background errors. This local surface analysis (LSA) uses a downscaled 5-km resolution 1-hr forecast from the Rapid Update Cycle (RUC) as its background field, as well as observations from METAR, RAWS, and various other mesonets obtained from the Meteorological Assimilation Data Ingest System.
Background and observation error variances are estimated through an observational method using all observations from the continental United States during the period 8 May – 7 June 2008. The ratio of observation to background error variance is found to be between 2 and 3. The background error covariance is specified as a function of spatial distance, which tends to remain strongly correlated over distance of approximately 80 km.
An approach to test the sensitivity to the specification of observation and background errors is evaluated using the LSA in a case study from 0900 UTC 22 October 2007 over the Shenandoah Valley of Virginia. Ten data denial experiments in which 10% of observations are randomly and uniquely withheld from each analysis are used to determine analysis error. Analysis error is estimated by the differences between the withheld observations and the corresponding analyses from which the observations are withheld. The analysis sensitivity to the withheld observations are computed from differences between the control analyses and the analyses from which the observations are withheld.
Poster Session 2, Poster session II
Wednesday, 19 August 2009, 2:30 PM-4:00 PM, Arches/Deer Valley
Previous paper Next paper