WK 6 - CausalNet: A Modern Approach for Understanding Ecosystem Interactions on a Global Scale

Thursday, August 18, 2022

11:45 AM – 1:15 PM EDT

Location: 513D

Organizer(s)

ES

Erik Saberski, Saberski

Scripps Institution of Oceanography
La Jolla, California, United States

Co-organizer(s)

EM

Ewa Merz

Swiss Federal Institute of Aquatic Science and Technology, Switzerland

Speaker(s)

EM

Ewa Merz

Swiss Federal Institute of Aquatic Science and Technology, Switzerland

Co-organizer(s)

ED

Ethan Deyle

Boston University, United States

Speaker(s)

ED

Ethan Deyle

Boston University, United States

Co-organizer(s)

SM

Steve Munch

University of California, Santa Cruz, United States

Speaker(s)

SM

Steve Munch

University of California, Santa Cruz, United States

Co-organizer(s)

GS

George Sugihara

University of California-San Diego
La Jolla, California, United States

Speaker(s)

GS

George Sugihara

University of California-San Diego
La Jolla, California, United States

Session Description: Ecosystems are largely structured by the interactions between species and with their environments. Despite decades of research considering ecosystems as closed geographic regions, interactions are not necessarily bound to one place. For example, the global circulation of water masses influences and connects nearly all marine ecosystems. Here, we offer an introduction to a new open-access tool for analyzing interaction networks on a global scale.

CausalNet is a new software that aims to connect time series variables across different regions. The power of this software is in the synthesis of 1) its large, queryable database of global time series data combining many publicly available datasets encompassing millions of time-series, 2) harnessing empirical dynamic modeling (EDM) to automatically identify relationships in these datasets and 3) analysis of the many millions of EDM-driven relationships to find general patterns among variables.

In this workshop, we will first introduce EDM methods as an approach for identifying non-linear relationships in time-series data. Next, we will teach attendees how to use CausalNet to 1) analyze patterns in datasets already hosted within the app and 2) input their own datasets to be analyzed in conjunction with the extensive, publicly available global database already curated within CausalNet. CausalNet will be a webapp, so all the attendee needs to bring is a computer with internet connection to have immediate access to these tools.