Multivariate, Temporal Visual Analytics for Climate Model Analysis

The left axis, GPP, is a model output and the other 8 axes are parameters used to run model ensembles (each line is an ensemble run). In both views, the upper range of GPP values are selected revealing an association with low values of crit_gdd and leafcn. New model runs will be executed using more constrained values for these parameters.


Exploratory Data analysis ENvironment (EDEN) enables exploratory data analysis for new DOE E3SM climate simulation and observational data using techniques that combine interactive data visualization and statistical analytics.

Significance and Impact

We are collaborating closely with Dan Ricciuto (ORNL) on his SciDAC BER application, “An Integrated System for Optimization of Sensor Networks to Improve Climate Model Predictions”.

EDEN gives climate scientists the ability to consider more variables from large scale, land model parameter sensitivity analyses and ultimately improve DOE model accuracy.


Research Details

  • The screenshots at left show one insight found during parameter sensitivity analysis for realistic values of GPP, a model output that measures photosynthesis in plants.
  • The plots helped scientists see that high values of GPP are associated with low leaf carbon to nitrogen ratio values (leafcn) and low critical growing degree days (crit_gdd).
  • Based on this insight, climate scientists will generate new ensembles covering smaller ranges of the leafcn and crit_gdd parameter space for more accurate surrogate models.


EDEN is an interactive visual analytics system for multivariate data analysis that extends the parallel coordinates information visualization technique providing new statistical analysis capabilities.

Last Updated: May 28, 2020 - 4:04 pm