Attribution and Causality Analyses of Regional Climate Variability

Abstract

A two-step attribution and causality diagnostic is designed by employing singular spectrum analysis to unfold the attributed climate time series into a trajectory matrix and then subjected to an empirical orthogonal function analysis to identify the evolving driving forces, which can finally be related to major climate modes through their independent frequencies by wavelet analysis. Application results from the arid and drought-prone southern Intermountain region of North America are compared with the climate or larger scale forcing diagnosed from slow feature analysis using the sources of the water and energy flux balance. The following results are noted: (i) The changes between the subsequent four 20-year periods from 1930 to 2010 suggest predominantly climate-induced forcing by the Pacific Decadal Oscillation and the Atlantic Multidecadal Oscillation. (ii) Land cover influences on the changing land cover are of considerably smaller magnitude (in terms of area percentage cover) whose time evolution is well documented from forestation documents. (iii) The drivers of the climate-induced forcings within the last 20 years are identified as the quasi-biennial oscillation and the El Niño–Southern Oscillation by both the inter-annual two-step attribution and the causality diagnostics with monthly scale-based slow feature analysis.

Keywords

• large-scale climate forces
• causality diagnostics
• slow feature analysis
• water and energy flux balance