Sensitivity of Surface Air Temperature Analyses to Background and Observation Errors

Abstract A two-dimensional variational method is used to analyze 2-m air temperatures over a limited domain (4° latitude × 4° longitude) in order to evaluate approaches to examining the sensitivity of the temperature analysis to the specification of observation and background errors. This local surface analysis (LSA) utilizes the 1-h forecast from the Rapid Update Cycle (RUC) downscaled to a 5-km resolution terrain level for its background fields and observations obtained from the Meteorological Assimilation Data Ingest System. The observation error variance as a function of broad network categories and the error variance and covariance of the downscaled 1-h RUC background fields are estimated using a sample of over 7 million 2-m air temperature observations in the continental United States collected during the period 8 May–7 June 2008. The ratio of observation to background error variance is found to be between 2 and 3. This ratio is likely even higher in mountainous regions where representativeness errors attributed to the observations are large. The technique used to evaluate the sensitivity of the 2-m air temperature to the ratio of the observation and background error variance and background error length scales is illustrated over the Shenandoah Valley of Virginia for a particularly challenging case (0900 UTC 22 October 2007) when large horizontal temperature gradients were present in the mountainous regions as well as over two entire days (20 and 27 May 2009). Sets of data denial experiments in which observations are randomly and uniquely removed from each analysis are generated and evaluated. This method demonstrates the effects of overfitting the analysis to the observations.