AN ANALYSIS OF THE ACCURACY OF 120-H PREDICTIONS BY THE NATIONAL-METEOROLOGICAL-CENTERS MEDIUM-RANGE FORECAST MODEL

Citation
Ma. Bedrick et al., AN ANALYSIS OF THE ACCURACY OF 120-H PREDICTIONS BY THE NATIONAL-METEOROLOGICAL-CENTERS MEDIUM-RANGE FORECAST MODEL, Weather and forecasting, 9(1), 1994, pp. 3-20
Citations number
NO
Categorie Soggetti
Metereology & Atmospheric Sciences
Journal title
ISSN journal
08828156
Volume
9
Issue
1
Year of publication
1994
Pages
3 - 20
Database
ISI
SICI code
0882-8156(1994)9:1<3:AAOTAO>2.0.ZU;2-6
Abstract
An assessment was made of the 120-h predictions by the medium range fo recast (MRF) run of the National Meteorological Center's (NMC's) globa l spectral model. The ability of the model to forecast surface cyclone s and anticyclones was evaluated and compared to shorter range (48 h) forecasts by the NMC Nested Grid Model (NGM) verifying at the same tim e. The study covers a period from lj April 1991 through 20 December 19 91 over a domain extending from 30 degrees N to 50 degrees N and from 60 degrees W to 130 degrees W. Results showed that the MRF at the 120- h range tended to underforecast the frequency of surface cyclones and anticyclones in the Rocky Mountain region and, in general, to forecast anticyclones and cyclones farther away from the observed systems than did the 48-h NGM verifying at the same time. The forecasts for centra l sea level pressures averaged within 1.5 hPa of the forecasts of pres sure by the NGM. Averaged over regions and seasons, the 120-h MRF erro rs in many cases were statistically indistinguishable from the 48-h fo recast errors. To illustrate model performance, an example of an accur ate 120-h forecast is contrasted with a less accurate 120-h forecast f ollowing similar initial conditions. In addition, the accuracy of each 120-h MRF prediction for sea level pressure over a domain between 30 degrees N and 60 degrees N from 60 degrees W to 150 degrees W was subj ectively determined for the period 14 March 1991 through 20 March 1992 . Initial conditions of these forecasts were inspected for common feat ures that could potentially be used to anticipate the accuracy of the 120-h MRF predictions. Typically, the best forecasts were characterize d by below-normal 500-hPa heights at 40 degrees N, 50 degrees W in the initial conditions, corresponding to relatively weak 500-hPa geostrop hic midlatitude (40 degrees N-60 degrees N) westerlies at 50 degrees W . By comparison, the initial conditions of the worst forecasts feature d, in most cases, above-normal 500-hPa heights at 40 degrees N, 50 deg rees W corresponding to relatively stronger 500-hPa geostrophic midlat itude westerlies at 50 degrees W.