From the Climate Prediction Center


The maps show the likelihood of the occurence of each of three categories of the three-month (1 month) mean of temperature (Below normal, Near normal, Above normal) and that of total precipitation (Below, Moderate, Above) during the target season (month). These are designated by the letters B, N or M, and A. At a given location we define the limits for each of these classes or categories based upon 30 years (1961-90) of observations so that each category occurs, by our definition, one-third of the time.

Regions in which we recommend the use of climatological probabilities (e.g., 33 1/3 percent for each category) are designated by "CL". Regions in which our forecast tools indicate that the likelihood of a category equals or exceeds CL are denoted by numbers (i.e., probability anomalies) labeling each contour. Note that the zero anomaly contours also correspond to CL. The zero contours separating regions of opposite category may be far apart, close together or may even join or branch. Please keep in mind that even a confident forecast for B, N or A does not imply that the categories favored by the forecast are expected to occur on each day, week or month of the three month period.

I'm planning an outdoor activity for 2 1/2 months from now. Do you have a forecast for that date?

No. While our 90-day forecast does include that day, no conclusions can be drawn about a particular day. Rather, the forecast applies to the average temperature and precipitation for the entire period.

What good is climatological information and where can I obtain it for my region?

Simply knowing the climate for your region can be suprisingly valuable. For example, you can determine, for a particular month (say August) and place (say Helena, Montana) the mean number of days that the sky is clear (13.1) partly cloudy (10.7) or overcast (7.1) from sunrise to sunset. For the same location and month the average number of days with thunderstorms is 9; the average number of days that the temperature is above 90 degrees, Farenheit, is 6.7; the normal high temperature is 81.3 F, while the minimum is 50.3 F.

This information is published in the "Local Climatological Data", which is updated with a new year ever year. The publication is available from the National Climatic Data Center, Federal Building, Asheville, N.C. 28801. You can obtain climatological information from their dial-in service if you have a touch-tone phone. The number is 704-271-4800.

I have to make a decision involving millions of dollars. Weather is an important input to my decision process but it is not the only input. How reliable are these forecasts, i.e., how much weight should I give them in my decision making process?

Your decision can be weighted appropriately by combining the known skill of the forecast (available from your RCC) with the appropriate mathematical decsion-making tools. These are certainly available to a company such as yours. You may also wish to consult a certified consulting Meteorologist. The American Meteorological Society publishes a list of these.

Are the forecasts more reliable for different times of the year?

Yes. The skill also varies regionally. Your RCC or a consulting meteorologist can provide you with detailed information about these aspects of the skill of the forecasts.

I live in Idaho. I understand that these forecasts are less reliable in my area. Can you explain why?

The accuracy of any forecast varies both regionally and seasonally. In some places, e.g., coastal areas, the ocean exerts a stabilizing influence on the climate, making it less changeable, or less "noisy", with respect to atmospheric disturbances at seasonal time scales. In those regions, the atmosphere is also more forecastable on long-ranges for a larger portion of the year, since slow climate variations which produce relatively small, but persistent anomalies, stand out better in the data.

Regions which are far from coasts, and especially in the mountainous western section of North America, including large portions of Idaho, are more subject to rapid and repeated short-term climate variations. Unfortunately, many of these variations are, by nature, not forecastable. Rather, they contribute to the climate noise, often making any small, persistent anomalies which do occur indistinguishable from the climate noise.

Are these forecasts produced using the same techniques as the 30- and 90-day forecasts we are accustomed to seeing?

No. The techniques and our methods for using them are different. The techniques we use are both statistical, e.g., the canonical correlation analysis and the optimal climate normals, and dynamical, e.g., two sets of coupled ocean-atmosphere models.

Even more importantly, we strictly require that a forecast of climatological probabilities be used in regions where the techniques we use have marginal accuracy. Furthermore, we do not use technique for which a long-term record of its accuracy does not exist.

This is the first forecast operation ever to apply such rigorous requirements for the use of skill. The principal effect of this rule is to greatly reduce the uncertainty which might accumulate at various stages of the forecast process. It also provides users with and unprecedented opportunity to make applications of the forecasts to their unique requirements.

Is there any difference in the reliability of temperature versus precipitation forecasts?

Yes. Temperature forecasts are inherently higher in accuracy, over larger regions than precipitation. This is because temperature is continuous, i.e., at each point there is always a temperature value, while the temperature varies uniformly from place to place. Precipitation, on the other hand, is discontinuous, i.e., at each point there is either zero precipitation or some amount, while precipitation does not vary uniformly from one place to the next. is the U.S. government's official web portal to all federal, state and local government web resources and services.