Looking ahead for seasonal forecast trends

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

Shading on seasonal outlooks from the US NWS varies according to the probability that the average temperature or total precipitation will fall within the top, middle or bottom third of all values that have been observed in the past.

In the past 2 centuries, however, the routine collection of weather observations from around the world has allowed climatologists to develop several tools that can help us in understanding the atmosphere and, from this understanding, predict likely conditions several months in advance where traditional numerical weather forecasts would have too great an uncertainty.

Based on weather observations, we know the "average" atmospheric conditions for a particular hour, day, month or season. We also know that weather patterns over one part of the planet can greatly affect the general flow of the atmosphere and alter the prevailing conditions over another—a process known as a teleconnection.

For example, when an El Niño event is present in the tropical Pacific Ocean, we know that there is an increased likelihood of a cooler and wetter than normal winter in the southeast US.

Beyond 7 days

While the ability of standard weather forecasts to predict the timing and magnitude of a particular variable such as temperature, precipitation, pressure or wind is extremely low beyond a week, our knowledge of climatology means that our ability to forecast the likelihood of conditions differing from average over a span of time and over a larger area is reasonably good.

If we focus not on predicting that 15 inches of snow might fall on Buffalo NY 12 days from now—a forecast that would probably have very low accuracy—and use instead the forecast models to determine that the week after next will likely have above normal snowfall in upstate New York, we can gain a great deal of confidence in the outlook.

In this way, forecasters can use short-range weather models to provide some measure of forecast to a medium-range time frame. However, for longer lead times—for example, a seasonal forecast 3–6 months or even a year from now—more specialized climate models are used.

These take into account factors such as teleconnections and average seasonal conditions to enhance their evaluation of conditions relative to historical scenarios.
Several national meteorological agencies and organizations around the world produce medium and long range forecasts.

For example, the Australian Bureau of Meteoro­logy (www.bom.gov.au) produces a 3-month outlook for temperature and precipitation, and the European Centre for Medium Range Weather Forecasts (www.ecmwf.int) provides forecasts worldwide to 15 days as well as seasonal (3-month) outlooks up to a year in advance, although not all of the ECMWF's output is freely available.

In the US, both NWS and the Intl Research Institute for Climate and Society (IRI) produce long-lead seasonal forecasts up to a year in advance. However, NWS also produces 8 to 14-day and 1-month forecasts.

Medium and long-lead NWS forecasts are available from its Climate Prediction Center (CPC) (www.cpc.noaa.gov). Although CPC makes output of average 850-hPa temperature, 500-hPa height and total precipitation available in the form of maps, they can be difficult to interpret. However, the 500-hPa maps (www.cpc.ncep.noaa.gov/products fcst_eval /html/d+8_page.shtml) for 6–10 days (D+8) and 8–14 days (WK2) can provide reasonably good guidance for potentially significant midlatitude storm systems.

Where the CPC excels is in forecasts of temperature and precipitation. However, it takes a bit of work to understand what you are looking at. On most temperature and precipitation outlooks you will see shaded areas with A or B in them, or unshaded regions with N or EC.

The shaded regions may also have numbers with them. In a general sense, the A, B and N mean you can expect (respectively) above, below or normal temperature and precipitation conditions in that area. However, it is a little more involved than that, and if you use the information to make decisions you may want to know the details.

CPC divides all the historical observations of temperature and precipitation into thirds. The middle third are considered to be in the "normal" (N) range, while the upper third are considered "above" normal (A) and the bottom third are "below" normal (B)—ie, not simply above or below the historical average value.

If you see a shaded area on the map with an A, that means that temperature or precipitation during the indicated period of time has a higher probability of falling into the above-normal third than into the below-normal third (but there's almost always a 33.3% chance of the value falling in the normal range).

In addition to the basic A, B and N division, there may be several numbered contours within a shaded region. These correspond to the degree of probability that the average temperature or precipitation will fall within that category.

For example, an A50 in 3-month average temperature means that there is a 50% chance it will be in the upper third of the historical record, a 33.3% chance of being near normal, and (what's left) a 16.7% chance the average will be in the coolest third. It is only when the least likely category drops to just 3.3% that the near-normal category might get lowered as the most likely category gains probability—for example, a 90% probability of below normal precipitation means a 3.3% chance of above normal and a 6.7% chance of near normal precipitation.


There is a final category—EC—which you are likely to see over large parts of most maps. Contrary to popular belief, EC does not stand for "extremely clueless"—it stands for "equal chance," which simply means the model was not able to determine with any confidence into which third of the climate record the temperature or precipitation might fall.

Given the lead time in some of these outlooks, it is perhaps surprising that more areas don't fall under EC. What this does mean, though, is that when there is enough information to merit an A, B or N, there will likely be some skill behind it.

CPC maps are, however, limited to the US. Fortunately, IRI seasonal outlook maps (iri.columbia.edu) are available for all regions of the world and use more or less the same model information and probability scheme (thirds) as CPC. However, one feature CPC adds to its suite of tools is a probability-of-exceedance chart for each climate division (www.cpcncep.noaa.gov/products/predictions/long_range/ index.php).

These charts plot the model forecast probabilities of exceeding certain temperature averages or precipitation totals against the historical probabilities. Once you are used to looking at these charts, you will be able to put numbers with the probabilities at a glance, for example, identifying that there is a is a 50% chance of exceeding the 55°F average winter temperature in San Antonio TX this winter, relative to about a 30% chance based on historical records.

Translated, it means a good chance of a warmer than usual winter for southern Texas.
While long-lead forecasts will not necessarily help you plan an upcoming flight, they serve an ex­tremely useful purpose for those among us who must plan flights or budget operating expenses months in advance.

The skill of these outlooks may not rival the accuracy of 1 or 2-day weather forecasts, but if you can anticipate that the routes you fly this winter may be colder and wetter than usual, or that above nor­mal temperatures next summer might drive up density altitudes, you have gained knowledge that can reduce unexpected costs or delays.

Karsten Shein is a climatologist with the National Climatic Data Center in
Asheville NC. He formerly served as an assistant professor at Shippensburg
University. Shein holds a commercial license with instrument rating.


1 | 2|