Weather Forecast Verification

Abstract

These notes are voted to stress the importance as well as the intrinsic difficulty of the weather forecasts verification, giving some hints to solve specific problems and some tools to face various situations. In general weather forecasts cannot be fully wrong but they cannot be neither fully right, this because they are trying to represent a future state of an extremely complex system, which is defined by too many aspects to be fully well described. There is a quite general confidence on the fact that it is at least possible try to quantify the amount of good and bad information that forecasters are trying to give on that future state. Nevertheless it is not possible to define in a unique way this quantification process, then different verifications procedures might give different results even if correctly realized. The standardization of definitions and of procedures is generally still poor and sometimes contraddictory. This fact makes, if possible, even more difficult to deal with the weather forecast verification. Facing the verification of rare weather events, as can be the case for the phenomena related to deep moist convection, extra difficulty arise by the fact that the powerful tool represented by statistics becomes less effective and the intepretation of results becomes in those cases even more tricky. In any case the verification of weather forecasts is an extremely important and structural aspect of the forcasting activity, that cannot be considered complete without it. Moreover the verification of weather forecasts can be an important opportunity to have a different look to the atmospheric aspect toward which we are pointing our attention and, for whom it might interest, to have a different look at the forecasters mind.