Seasonal Forecasting Perils by Sean Sublette
Beyond the typical medium range weather forecasting, which goes 7-14 days, there are two big questions that get asked of meteorologists every year.
1.What's the winter going to be like?
2.Is the hurricane season going to be bad?
Going back about 30 years, there was little idea on how to make such a forecast. The atmosphere is a turbulent and chaotic place. Forecasting specifics on this large of a time scale is, at best, difficult.
However, there is a public demand for this information. So our assignment, as meteorologists, is to look for clues. We look at definable and repeatable patterns in the atmosphere. Some relations are physical, some are statistical, and some are both. But these are big things that we rarely discuss in the typical 3-minute weather forecast. Some examples:
El Nio/Southern Oscillation
Atlantic Multidecadal Oscillation
Pacific Decadal Oscillation
North Atlantic Oscillation
Madden Julian Oscillation
You can use your favorite search engine to look up the definitions. But by analyzing these large patterns, and looking for similarities in them over years and decades, we hope to shed light on what future conditions will be like. It is not perfect. This type of forecasting still has its growing pains. Most meteorologists refer to these as outlooks rather than forecasts. And the more responsible among us will point out where they believe uncertainty is the greatest.
And remember that uncertainty is not the same as unknown. There are certain things we know well about these patterns, but precisely how they will affect the weather in one specific spot is also dependent on many other influences.
What troubles me is to read headlines that broad bush the entire meteorological community after a seasonal forecast does not work out. You have probably read them. Those ALL CAPS headlines that read something like: WEATHER FOLKS DEAD WRONG ON HURRICANE SEASON.
Excuse us for trying. We never claimed to know everything about everything. There is a limit to what we know, but we made our best guess. And we all have disagreements about this type of forecasting.
A more painful example occurred within the last few months. The consensus was that this past December would be colder than average across eastern North America. That obviously did not come to pass. Most of us will begrudgingly admit our mistakes, but we get very testy about having our information misrepresented in the press.
There have been times when we forecast an active hurricane season, with respect to the total number of tropical cyclones (that includes tropical storms and hurricanes), and it has indeed been active. In this case, we define active as above normal. But during that one particular season, the number of tropical cyclone hits on the United States was very small. So the public perception is that it was not active. After all, out of sight, out of mind.
To be sure, there is very limited, and some would argue, questionable skill in forecasting where and when tropical cyclones will hit several months in advance. But society implores us to try. The energy industry. The insurance industry. The agriculture industry. They all want us to try. So we try. Seasonal forecasting is one of the great meteorological puzzles to solve.
Like all good science, when presented a new challenge such as this, we use the knowledge that we do have to attack it and see if our new ideas work. If part of it works, great. If part of it does not work, we go back to the drawing board and try something else. We learn from our mistakes and (hopefully) get better. It is a slow process. One that can take years to show progress. But that is how new ground is broken, and that is how the science advances.
It beats sitting around and hoping the Farmer's Almanac will be right.