State of the Art of Predicting Short Period Climatic Variations

Abstract

This paper offers an attempt to summarize, compare and critically examine methods of long range weather forecasting and to suggest avenues where greater progress might be found.

It starts out with some general remarks about the extremely complex nature of the prediction problem on time scales of months, seasons and years, which explains in large part the low degree of skill presently achieved. Also discussed are the poor quality and quantity of available global data, a deficiency which is frequently and erroneously advanced as the principal cause of forecast failures. But it is concluded that the root of the problem and the main obstacle in the quest for better long range forecasts is a lack of understanding of the physics governing these large scale phenomena.

There follows a treatment of methods used which involve statistical, synoptic and physical (conceptual) procedures.

Among the methods are:

• 1. The use of analogues (similar past situations which can be used as guides for the future).

• 2. Long period trends in atmospheric data.

• 3. Teleconnections, or interactions between remote anomalous weather-producing circulations.

• 4. Regression equations and adaptations of empirical orthogonal functions and other devices designed to arrive at significant lag relationships.

• 5. Climatological contingencies indicating probabilities of monthly or seasonal surface temperature and precipitation following given observed conditions.

• 6. Use of the anomalous character of surface boundary conditions like snow cover, sea surface temperature and soil moisture to determine probable forcing and stabilization of anomalous atmospheric patterns (including storm tracks, etc.).

In practice the above methods are usually used jointly, with a weighting which changes according to situation. This weighting, largely subjective, is often the reason for both forecast success and forecast failure.

Some evaluations of forecast skill are presented which can be viewed as encouraging or discouraging, depending on the receptivity of the reader.

Finally, an optimistic view of the future is presented wherein greater understanding of the physics of large space and time scales may permit numerical (dynamical) long range simulations and predictions for months or seasons and conceivably open the door to predictions of anomalous climate regimes for years and decades. This happy event defies setting a time frame, so that prediction of the state of the art years hence becomes as difficult as a long range weather forecast a season in advance is at present. However, some recent encouraging dynamical predictions out to a month or so give some cause for optimism.