The uncertain art of long-range weather forecasting

By

Long-range weather forecasting can be an exercise in near frustration. Its practitioners claim only marginal skill. They stress that they lack a well-established scientific theory of cause and effect to guide their efforts. And they warn users of their forecasts to take their uncertain projections with caution. Nevertheless, every fall some of the most experienced meteorologists who practice this difficult art release estimates for the coming winter. They cover average conditions for the months of December, January, and February.

Both the United States National Weather Service (NWS) group at the predictions branch of the National Climate Analysis Center and the Jerome Namias/Dan Cayan forecasting team at the Scripps Institution of Oceanography in La Jolla, Calif., released their winter outlooks in late November. Now they are watching the weather as eagerly as anyone to see how well their projections are doing.

So far, the early cold and snows that have hit many parts of the United States are something of a comfort to the forecasters, even though they may challenge the rest of us.

Recommended: Default

Namias and Cayan call for a continuation of the cold, snowy weather initiated during the fall over the Far West. They add that some of this cold and snow will frequently penetrate into the central plains. On the other hand, they expect mild conditions with moderate rains east of the Appalachians and in the Gulf states. Frequent storms and windiness, with occasional sleet, are expected in the broad zone separating cold Western air masses from the Eastern warmth.

The NWS forecast is a little trickier to interpret. It's expressed in terms of probabilities, which reflect the confidence which people should place in the predictions.

As Donald Gilman, the team's chief, has explained, the NWS forecasters have no skill in predicting normal or near normal weather. Where they have no strong expectation of significant departures from normal conditions, they just give departures in either direction (higher or lower than normal) ``equal chances'' of occurring and leave it at that.

That may not seem too helpful. But at least it doesn't mislead people into thinking the forecasters can do better than cite the climate atlas.

However, Gilman's team does show some skill in forecasting above or below normal average temperatures and precipitation. In these cases, they adjust the climatological probability of extreme conditions to reflect their expectations.

For example, the climatology book assigns a 30 percent probability to abnormally high or low average temperature. This means that in the long run higher average temperatures occur 30 percent of the time and lower temperatures occur 30 percent of the time. If Gilman's team expects colder weather for a region, it may increase the probability of lower temperature to 45 percent.

That would drive down the probability of the opposite type of extreme weather by a corresponding amount. In this case, the probability of warmer temperature would drop from 30 percent to 15 percent.

The amount of the adjustment reflects the forecasters' degree of confidence in their prediction. Thus a 65 percent probability of below normal temperature is a more confident prediction than a 45 percent probability. This is the light in which the probabilities given in the accompanying NWS winter forecast chart should be taken. Remember that warm or cold average temperatures and light or heavy precipitation are four distinct weather categories, each of which has a natural probability of occ urring 30 percent of the time in the long run.

The basis for these long-range forecasts differs profoundly from that of the short-term predictions of TV weathercasters.

Mathematical models of the atmosphere, which are run on computers, underlie the latter. These models incorporate well-understood physical theory of how short-term weather operates. However, their predictive power fails for periods longer than a week to 10 days.

Namias, Gilman, and other long-range forecasters rely partly on climatological statistics, experience, and on factors that lend stability to long-term general weather patterns. These stabilizing factors include soil moisture, snow and ice cover, and the distribution of warm and cold surface water in the Pacific and Atlantic Oceans. It's what Gilman calls ``a kind of cook's mixture of empirical techniques'' which can give moderately useful results. His score card for last winter's forecast shows 7 points

out of a possible 17 for temperature and 4 points out of 8 for precipitation. Let the forecast user beware.

A Tuesday column. Robert C. Cowen is the Monitor's natural science editor.

Share this story:

We want to hear, did we miss an angle we should have covered? Should we come back to this topic? Or just give us a rating for this story. We want to hear from you.

Loading...

Loading...

Loading...