Continuous Forecasts
When considering a forecast, one of the essential aspects is the actual value predicted by the forecast. For example, a forecast for 2 meter maximum air temperature of 75 degrees Fahrenheit is explicitly calling for one value to occur as the highest recorded temperature value for the entire day for that single point in space. If the maximum observed 2 meter air temperature was 77 degrees Fahrenheit for that point instead, then the forecast was incorrect: some might consider this the end of the story, another “blown” forecast. However, if the entire spectrum of 2-meter air temperature forecasts that could have been forecasted is considered, however, we can ask the question “how good or bad was that forecast, really”? After all, wasn’t the forecast only 2 degrees Fahrenheit off from what was observed? Continuous forecast verification is a category of verification that considers the entire real value spectrum that a forecast variable can take on, rather than individual discrete points or categories as is the case with dichotomous and multicategorical verification.
