Forecast variable types

For both users and practitioners of verification methods, the type of variable being verified is an important factor in the choice of verification method. There are two types of meteorological variables:

Continuous variable: The forecast is given as a specific value of the variable, in the units of the variable, for example, temperature in degrees Celsius. The forecast and verifying observation may take on any (or nearly any) value of the variable.

Categorical variable: The forecast is stated as the occurrence of a pre-determined range of values of the variable. Some variables are inherently categorical, while others may be redefined as categorical variables by selection of threshold values to delineate the categories. A special type of categorical variable is the dichotomous variable, consisting of two categories only.

Forecasts of continuous variables may be either deterministic or probabilistic. A deterministic forecast results in a specific forecast value of the variable which can then be compared with the corresponding observation. A probabilistic forecast of a continuous variable is usually in the form of a probability distribution of values, as might be obtained from an ensemble system. Special verification methods are needed for ensemble forecasts; these are not described in these modules.

Forecasts of categorical variables may be either categorical or probabilistic. The former is definitive – one of the categories is predicted to occur. A probabilistic forecast gives the estimated probability of occurrence of the categories of the variable.

The verification measures discussed in these modules apply to deterministic forecasts of continuous variables, categorical forecasts of categorical variables and probabilistic forecasts of categorical variables, three types in total.

Test your understanding of the different types of forecasts by completing the following exercise: Match the following forecasts with the correct type by dragging the tags to the appropriate location.

Now that you can classify your favourite weather forecast variable into one of the three types for verification, you are ready to move on to the verification module(s) of your choice.

Wind speed is forecast to be 25 knots

Probability of precipitation (POP) =60% overnight

Snow is expected tomorrow

Today’s maximum temperature is expected to be 6 degrees above normal

Rain at times heavy tonight, accumulations greater than 25 mm are expected

Correct. Forecast is for a wind speed of exactly 25 knots.

Incorrect. Try again.

Correct. Precipitation occurrence is a dichotomous variable, with the implied threshold at the minimum measureable amount.

Incorrect. Try again.

Correct. Precipitation type is inherently categorical. No amounts are specified.

Incorrect. Try again.

Correct. A specific temperature is forecast, but one needs to know what is "normal".

Incorrect. Try again.

Correct. Precipitation accumulation has been expressed as a categorical variable with a threshold.

Incorrect. Try again.

Good, this component is correctly located

That is not the correct location !

Please drag the component text labels to their correct position on surface chart.

You have completed the exercise