Estimates of atmospheric predictability at medium range

Publisher Website
Google Scholar
Abstract
Recent studies based upon the output of the ECMWF operational forecasting model indicates that if, after the first day of a forecast, a perfect model could be substituted for the present model, forecasts as good as those presently produced at seven days would be realized at ten days. These studies do not reveal how much improvement in one‐day forecasting is possible. We hypothesize that if all other imperfections in the forecasting procedure could be removed, the inevitable initial uncertainties in observing the small‐scale features would, after D days, lead to error fields with amplitudes and spectra resembling those of the errors in present one‐day forecasts. The appropriate value of D is highly dependent upon the spectrum of actual atmospheric motions. Estimates with a crude model place D at about four days, thereby implying that the present forecasting success at one week may some day be realized at nearly two weeks.
Keywords

This publication has 0 references indexed in Scilit: