Looking ahead for seasonal forecast trends

Daily checks are fine, but always watch for what's coming in the long-range picture.

By Karsten Shein
Comm-Inst, Climate Scientist

Bombardier CRJ lands at IAD during a snowstorm in Jan 2010. While long-range forecasts could not have predicted the timing or strength of a particular storm, they would have alerted the pilot to a better than usual chance of having to deal with adverse winter weather in the region.

Forecasts are an essential part of a pilot's preflight planning. We rely heavily on guidance from meteorologists and their computer models of the atmosphere to help us determine everything from possible convection to winds aloft and freezing levels anywhere from a few hours from now to 3 or 4 days out into the future.

While we often joke at the unreliability of some weather forecasts, we recognize their inherent value in setting risk probabilities. We accept even sizable uncertainty in a forecast over having no guidance at all about upcoming conditions.

This is because, more often than not, reliance on these weather forecasts helps to reduce our chances of running into adverse weather that may cost us time or money—neither of which we can normally afford to lose.

But what about long-range flight and operations planning? Is the persistent storminess at your airport likely to remain entrenched into next week? Will your department be faced with a heavy deicing bill this winter? What information is out there to help figure out what conditions might be like 2 or 3 weeks or even months from now?

The first tools to come to the minds of many might be to turn to the Farmer's Almanac, check with the local groundhog or measure the stripes on wooly bear caterpillars.

However, such arcane prognostic tools might not win you any credibility with the company's board when you ask for a 20% increase in your operating budget because you anticipate a colder, wetter than normal winter.

Fortunately, there are a few tools at a pilot's or flight department manager's disposal that are less based in folklore or secretive methodologies and are instead based soundly in our scientific understanding of the atmosphere.

Beyond the standard 7-day weather forecasts, several organizations, both governmental and private enterprise, produce forecasts anywhere from 7 days out to 12 months.

And while these forecasts are unlikely to tell you what the weather conditions are likely to be on Jan 5 at 0800, they are more than likely going to help reduce the uncertainty surrounding planning for winter weather conditions.

Weather forecast vs climate forecast

To understand what a long-lead forecast is, we first have to understand the similarities and differences between weather and climate forecasts. The forecasts we are used to seeing are weather forecasts. These are basically the best guess of computer models and meteorologists as to the most likely conditions of the atmosphere at a specific location and time.

Weather forecasts are generally the result of a relatively complex process. First, tens of thousands of data values from surface stations, weather balloons, radars and satellites are fed into computer models of the atmosphere.

The models then use equations for the physics of fluid flow to calculate the changes in the atmosphere from its present state out to several days into the future. For the free atmosphere—generally above a few thousand feet—the output of the models can be used more or less directly by meteorologists.

But for air in the surface layer, where topography and other local influences can modify the conditions greatly, a postprocessing technique called model output statistics (MOS) must be employed. MOS takes the model output and runs it through a set of statistical models to adjust for surface effects and generate the most probable values for a surface location.

These are the conditions you are most likely to see when you look at a TAF or other forecast for a specific location.

Precipitation outlook for Dec 2010–Feb 2011 (winter) over the US. Although unable to provide specifics, such products, issued in the middle of each month and going out 12 months in advance, give a good idea whether an area is likely to see above or below-normal precipitation for the period in question. Like weather forecasts, seasonal forecasts become less reliable the further ahead they look.

The final step in weather forecasting is for a meteorologist to examine the output and MOS from the several computer weather models running all the time. Each model differs in the way certain components—such as the influence of clouds—factor into the equations.

Local forecasters understand which models work best under certain conditions or during certain seasons, and use the information from all of the models to create a forecast. The certainty a meteorologist has in their forecast is often based on how well the models agree with one another.

In the case of the US National Weather Service (NWS), this certainty can often be found in the "forecast discussion" available on the NWS forecast page (weather.gov).
However, even the best forecasts are really just the most probable of a spread of values.

This spread increases and the probability of the most likely value occurring decreases as forecast time progresses. Each iteration of the model's calculations introduces a little random error into the mix, meaning an ever increasing distribution of possible solutions.

As can be seen clearly in forecasts of hurricane tracks, the most likely track for the storm is a single value, but the probability of that value occurring may drop from 90% on day 0 to 40% by day 3, while the range of alternative track locations expands greatly each day.

Beyond about 7 days, the probability of the most likely value is too low and the range of possible alternative values too great to produce a consistently meaningful weather forecast. This is where climatology takes over.

Climatology is normally associated with average or "normal" conditions, but it is much more than that. It is a study of the statistical properties of the atmosphere. Climatology looks at historical weather conditions and pieces together patterns and connections in both space and time. Like weather forecasting, climate analysis is now done primarily by high-speed computer models, but they actually share a common lineage.

In the centuries before weather observations and numerical models of the atmosphere, sailors and others routinely made accurate weather forecasts. They looked to the clouds and winds and, based on their knowledge of what similar conditions had brought in the past (climatology), could be reasonably certain that such conditions would again materialize.

When barometers were placed into use in the 1600s, the scientists who studied them were able to make a connection between changes in air pressure and the weather conditions that followed. Over time this knowledge allowed accurate short-term weather forecasts to be made.


1 | 2| next