theweatherprediction.com
[--MAIN HOME--] [--ALL HABYHINTS--] [--FACEBOOK PAGE--]

FORECAST VERIFICATION

METEOROLOGIST JEFF HABY

It is important to have a forecast verification that is both fair to you and your viewers / clients. Forecast verification can be either an objective or subjective analysis of how well your forecast is compared to the actual conditions that occur. If a forecast does not have a fair verification it may be either because it is too general or too specific.

The "too general" forecast has a good chance of verifying because the temperature forecast and precipitation forecast has too large of a spread. The advantage of the too general forecast is the fact is has a good chance of being correct. The disadvantage is that is does not properly prepare the viewers / clients for the weather. An example includes: "The high will be in the 50's today with a 50% chance of precipitation". This forecast gives a 10-degree spread on temperatures and gives no characteristics to the rainfall such as intensity, amount, duration and timing.

In comparison, the "too specific" forecast does not have a good chance of verifying because there is little room for even the slightest error. The advantage of the too specific forecast is that you will look like a genius if it verifies. Forecast error is usually the result. An example includes: "The high will be 53 degrees today and it will rain 0.67 inches between 3:00 and 5:00 pm". There is no room for error on the temperature and the chance of having the precipitation forecast pan out is very unlikely. Other examples of too specific forecasts include predicting the exact county a tornado will occur before the severe thunderstorm forms and the exact time clouds will clear during the day.

A forecast that hits between the too general and too specific forecasts is best. An example of this is: "The high will be in the mid-80's with a 70% chance for afternoon convective thunderstorms. Rainfall amounts will range from 0.50 to 1.00 inches." Although exact temperatures can be used to get the point across on what temperature to expect, a small range such as the mid-80's has a much better chance of verifying. If the forecast says the high will be 75 degrees and the actual high is 74 degrees, is the forecast a bust? I think not, Clearly this is better than saying the high will be in the 70's especially when forecasting for a point location such as an airport. Mesoscale effects will often produce a temperature gradient over the forecast region. This makes it important to distinguish which regions will be cooler or warmer. This becomes a problem especially for large forecast regions.

It is wise to keep a forecasting journal to see how well your forecasts verify for temperatures, precipitation and other weather events you deem important. A high or low within 2 degrees on the today, tonight and tomorrow forecast is sufficient accuracy. Precipitation is more difficult to access forecast accuracy. Using both an objective and subjective analysis are beneficial. The objective analysis would include predicting the precipitation amount. The subjective / descriptive analysis would be describing the character of the precipitation (i.e. occurring around noon, heavy at first then becoming lighter, how the rainfall will change temperatures, precipitation type, and the intensity / duration).

Becoming a great forecaster series: http://www.theweatherprediction.com/greatforecaster/