This paper is concentrated on the evaluation of circulation type classifications (CTCs) from the European Action COST733 catalogue version 2.0 in terms of their ability to capture winter precipitation, expressed as percentage, over Spanish Iberia and the Balearic Islands. The explained variation, the pseudo-F statistics and the Brier skill score are used to quantify the explanatory power of circulation classifications. As secondary aims, the impact of using different number of circulation types, additional types of variables and 4-day sequences in the generation of classifications is analysed.
Although no optimal method has been found, nevertheless, the results suggest that the use of CTCs based on optimization algorithms are, in general, performing better than those which are based on other algorithms (i.e. leader algorithms). Distinct variations in skill exist not only among classifications from different groups of basic methods but as well between classifications from the same method group; being remarkable the behaviour of the optimum random centroid method.
Results are very dependent on the metric, for the explained variation, and the Brier skill score, the larger the number of circulation types, the better the performance; contrary to the behaviour of the pseudo-F statistic. The inclusion of 500 hPa vorticity in the generation of classifications improves results while a general deterioration is observed when considering 4-day sequences. These results are only valid for the selected season and cannot be transferred to other locations and seasons.