Medium- and Long-Range Forecasting

A. James Wagner Climate Analysis Center, NMC, NWS, NOAA, Washington, D.C.

Search for other papers by A. James Wagner in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

In contrast to short and extended range forecasts, predictions for periods beyond 5 days use time-averaged, midtropospheric height fields as their primary guidance. As time ranges are increased to 3O- and 90-day outlooks, guidance increasingly relies on statistical techniques using autocorrelation fields rather than numerical weather prediction (NWP) products as the primary prediction tool.

The basis for the medium-range 6- to 10-day forecast is a mean 500 mb height and anomaly field for the forecast period, derived from a mix of output from two different numerical models, with some statistical and subjective modification added if desired.

The monthly or 30-day outlook is based on a subjectively constructed mean 700 mb prognostic map based on available NWP mean height and anomaly fields out to 10 days, the appropriate 700 mb 1-month lag auto-correlation field, and subjective use of teleconnection and empirical orthogonal function patterns for consistency.

A quantitative midtropospheric height and anomaly map is not constructed for the seasonal (90-day) outlook, but statistically significant height indications are obtained from a series of seasonal 700 mb lag autocorrelation fields going back as far as 2 1/1 years. Numerical weather prediction products do not enter into the seasonal forecast, but boundary forcing by sea surface temperature anomalies, particularly in the Pacific, is considered during the seasons these factors have been shown to have a significant effect on the mean circulation. Extensive use is made of teleconnections to obtain a consistent overall qualitative concept of the expected pattern.

Mean surface temperature and precipitation anomalies expressed either in terms of probabilities or categories are the main forecast products. The skill varies regionally and seasonally, is considerably less than for short-range forecasts, and declines slowly with increasing length of the forecast period.

Abstract

In contrast to short and extended range forecasts, predictions for periods beyond 5 days use time-averaged, midtropospheric height fields as their primary guidance. As time ranges are increased to 3O- and 90-day outlooks, guidance increasingly relies on statistical techniques using autocorrelation fields rather than numerical weather prediction (NWP) products as the primary prediction tool.

The basis for the medium-range 6- to 10-day forecast is a mean 500 mb height and anomaly field for the forecast period, derived from a mix of output from two different numerical models, with some statistical and subjective modification added if desired.

The monthly or 30-day outlook is based on a subjectively constructed mean 700 mb prognostic map based on available NWP mean height and anomaly fields out to 10 days, the appropriate 700 mb 1-month lag auto-correlation field, and subjective use of teleconnection and empirical orthogonal function patterns for consistency.

A quantitative midtropospheric height and anomaly map is not constructed for the seasonal (90-day) outlook, but statistically significant height indications are obtained from a series of seasonal 700 mb lag autocorrelation fields going back as far as 2 1/1 years. Numerical weather prediction products do not enter into the seasonal forecast, but boundary forcing by sea surface temperature anomalies, particularly in the Pacific, is considered during the seasons these factors have been shown to have a significant effect on the mean circulation. Extensive use is made of teleconnections to obtain a consistent overall qualitative concept of the expected pattern.

Mean surface temperature and precipitation anomalies expressed either in terms of probabilities or categories are the main forecast products. The skill varies regionally and seasonally, is considerably less than for short-range forecasts, and declines slowly with increasing length of the forecast period.

Save