ABSTRACT:
The numerical forecast using ALADIN model in Meteorological and Hydrological Service of Croatia is run operationally since July 2000. Over the years, various methods of validation and verification of the operational forecast have been applied. The classical methods using root mean square error and mean absolute error would often renalize the high resolution ALADIN when compared to a low resolution global model forecast due to double penalty paradigm. Therefore, the model was mostly evaluated by plotting the forecast and the measurements to allow subjective comparison, especially in weather situations that have high impact on the living and traffic conditions in Croatia. Here we show an overview of validation and verification products created operationally. These products intended for subjective validation in real time can help the forecaster in the decision if to rely on a particular forecast run more or less than to another. Statistical verification scores provide information on model bias and root mean square error but suffer from missing data due to automatic procedures used quality check and filtering of the measured data.
