Scholarlink

<p>Promax and Infomax rotated T-mode PCA are applied to ten data batches (subsets) to develop a 500&#xa0;hPa daily circulation catalogue for Iran for January 1948 to December 2019. The ten subsets yield statistically insignificantly different classifications; therefore, one subset is selected and its PCs projected onto the full dataset. An alternative to the batch technique, Common Principal Components (CPCs) is applied to a matrix comprised of PCs of all ten subsets, identifying the common PC loadings for projecting onto the full dataset. For the batch Promax method, a congruence coefficient (G) ≥0.92 is used to ensure an accurate classification, selecting four PCs. However, owing to their single-phase sign of these four Promax PCs, only four circulation types (CTs) could be defined, too few to capture the varied circulations. Therefore, an alternative approach, using an oblique Infomax rotation, was applied to both the batch and CPC approaches. In contrast to Promax, Infomax supports eight rotated PCs, all with G ≥ 0.96, and PCs exhibiting large values for both phase signs, providing sixteen meaningful daily CTs. The batch and CPC Infomax solutions performed similarly for classification, with five cold season patterns, six warm season patterns and five transitional patterns. Further, Infomax rotated TPCA from batch and CPC applications classifications improve the within-group similarity compared to Promax, classifying ~ 75% more cases with good to excellent matches (G≥0.92) and exhibiting a 60% decrease in cases matched less satisfactorily (G≤0.92). The batch and Infomax approaches were found to be roughly equivalent in accuracy.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A daily atmospheric circulation pattern catalogue for Iran and adjacent areas: application of T-mode principal component analysis (TPCA) using batch and common principal component approaches with oblique Promax and Infomax rotations

  • 期刊论文

  • 发表日期：

    2025年4月22日

查看全文
  • Tayeb Raziei,
  • Michael Richman

摘要

Promax and Infomax rotated T-mode PCA are applied to ten data batches (subsets) to develop a 500 hPa daily circulation catalogue for Iran for January 1948 to December 2019. The ten subsets yield statistically insignificantly different classifications; therefore, one subset is selected and its PCs projected onto the full dataset. An alternative to the batch technique, Common Principal Components (CPCs) is applied to a matrix comprised of PCs of all ten subsets, identifying the common PC loadings for projecting onto the full dataset. For the batch Promax method, a congruence coefficient (G) ≥0.92 is used to ensure an accurate classification, selecting four PCs. However, owing to their single-phase sign of these four Promax PCs, only four circulation types (CTs) could be defined, too few to capture the varied circulations. Therefore, an alternative approach, using an oblique Infomax rotation, was applied to both the batch and CPC approaches. In contrast to Promax, Infomax supports eight rotated PCs, all with G ≥ 0.96, and PCs exhibiting large values for both phase signs, providing sixteen meaningful daily CTs. The batch and CPC Infomax solutions performed similarly for classification, with five cold season patterns, six warm season patterns and five transitional patterns. Further, Infomax rotated TPCA from batch and CPC applications classifications improve the within-group similarity compared to Promax, classifying ~ 75% more cases with good to excellent matches (G≥0.92) and exhibiting a 60% decrease in cases matched less satisfactorily (G≤0.92). The batch and Infomax approaches were found to be roughly equivalent in accuracy.