A multi-objective fuzzy classification of large scale atmospheric circulation patterns for precipitation modeling

摘要

A multi-objective fuzzy rule-based classification (MOFRBC) technique is applied in order to cluster and classify daily large scale atmospheric circulation patterns (CPs) and analyze the relationship between the CPs and local precipitation. The methodology is illustrated by means of an Arizona case study. For this purpose, three indices are calculated to measure the information content of the clustering method in terms of predicted precipitation. A thorough sensitivity analysis is provided to gain more understanding on the robustness of MOFRBC model. Furthermore, it is shown that extending the daily premises to two-day and three-day sequences of CPs improves the information content of the classification. The results are also compared with the original subjective clustering. For the Arizona case study MOFRBC seems to be a competitive technique with the advantage that the physical aspects can be better represented by fuzzy rules (which tend to mimic the human way of decision making) than by objective methods.