Application of Cluster Analysis to Climate Model Performance Metrics

• DOI: 10.1175/2011jamc2643.1
• Publication Date: 2011-03-29T17:01:11Z
• Authors: Satoru Yokoi; Yukari N. Takayabu; Kazuaki Nishii; Hisashi Nakamura; Hirokazu Endo; Hiroki Ichikawa; Tomoshige Inoue; Masahide Kimoto; Yu Kosaka; Takafumi Miyasaka; Kazuhiro Oshima; Naoki Sato; Yoko Tsushima; Masahiro Watanabe
• Keywords: 13. Climate action; 01 natural sciences; 0105 earth and related environmental sciences

Abstract The overall performance of general circulation models is often investigated on the basis synthesis a number scalar metrics individual that measure reproducibility diverse aspects climate. Because physical and dynamic constraints governing climate, model’s in simulating certain aspect climate sometimes related closely to another aspect, which results significant intermodel correlation between metrics. Numerous correlations may cause problem understanding evaluation synthesizing One possible way alleviate this group correlated beforehand. This study attempts use simple cluster analysis 43 Two clustering methods, K -means Ward yield considerably similar results, several are found be physically dynamically reasonable. Furthermore, averages lower than These suggest helpful obtaining appropriate grouping. Applications also discussed.