Premium
Debates—Does Information Theory Provide a New Paradigm for Earth Science? Causality, Interaction, and Feedback
Author(s) -
Goodwell Allison E.,
Jiang Peishi,
Ruddell Benjamin L.,
Kumar Praveen
Publication year - 2020
Publication title -
water resources research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.863
H-Index - 217
eISSN - 1944-7973
pISSN - 0043-1397
DOI - 10.1029/2019wr024940
Subject(s) - causality (physics) , pairwise comparison , perspective (graphical) , computer science , multivariate statistics , earth system science , meaning (existential) , complex system , management science , data science , range (aeronautics) , epistemology , artificial intelligence , ecology , machine learning , engineering , philosophy , physics , quantum mechanics , biology , aerospace engineering
The concept of causal interactions between components is an integral part of hydrology and Earth system sciences. Modelers, decision makers, scientists, and other water resources stakeholders all utilize some notion of cause‐and‐effect to understand processes, make decisions, and infer how systems react to change. However, there are different perspectives on the meaning of causality in complex systems and, further, different frameworks and methodologies with which to detect causal interactions. We propose here that information theory (IT) provides a compelling framework for the detection of causality and discuss approaches for several levels of analyses that capture interactions that range from pairwise to multivariate in nature. We illustrate these types of analyses with an example based on weather station time series variables, in which variables may interact pairwise or jointly and on short to long time scales. In general, many unsolved or even unanticipated questions in the hydrologic sciences could benefit from this perspective.