Verification of CSU WRF and NCEP model forecasts: 500-mb height (vs. raobs), October 2012

Switch to current verification statistics

-- Forecasts are verified against radiosonde observations. For verification against the NAM analysis, click here.
-- BCRMSE = Bias-corrected root mean square error. Reflects the forecast errors not associated with the bias. Lower numbers are better.
-- RMSE = Root mean square error. Lower numbers are better.
-- Bias = Difference between the average forecast height and the average observed height. A value of zero reflects an unbiased forecast.
-- The second-from-bottom graph compares the area-averaged ensemble spread against the BCRMSE. In the long run, these should be approximately equal for a well-tuned ensemble.
-- Verification is done on the outer (36-km) grid for the CSU WRF runs, shown on this page.
-- Scores are calculated using the Meteorological Evaluation Tools software package.

Note for October 2012: Because of a power outage at CSU, the CSU WRF members were not run for the 0000 UTC 12 October cycle. As such this date is missing from the verification.