Webviz-ert

Introduction

Webviz-ert is an alternative visualisation tool in ERT. Our main goal is to provide our users with a more rich interface with the common, yet simple visualisations needed. Advanced users / needs would still be pointed in the direction of ResInsight or equivalent.

Feedback regarding stability, bugs and features is very much appreciated.

How-to

With a komodo release sourced, webviz-ert is currently launched from the terminal.

cd /<path>/<to>/<ert-project>
ert vis <ert-config-file>

Expect approx 60-90 seconds of initialization-time before firefox is launched and webviz-ert is up and running.

To end the session just close firefox and Ctrl+C in the terminal.

Views

Currently webviz-ert consists of 4 different pages available from the left menu. In all the pages, the user is provided with the option to select ensembles (one or more) based on which the corresponding plugin will populate its response and / or parameter list boxes.

Plot viewer

We can select both responses (simulation results) and parameters, given in the selected ensemble(s). Each selection opens a stand-alone plot. In case of responses, the data is shown as a function plot, where x-axis represents time and y-axis represents function values. We can choose to render either all realizations (Function Plot) or as a statistical curves (Statistics), via the provided radio button. When showing all realizations individual realizations can be turned off / on by clicking on the corresponding realization id on the right hand side. When multiple ensembles are selected each plot will be overlayed on top of each other transparently, where each ensemble gets its own color.

Observation analyzer

We can choose a single response key, for which the computed misfits are visualized. One can choose to show misfits grouped by time (Univeriate) as boxplots showing misfits statistics per a single time point or to render misfits as a histogram aggregating misfits over the entire temporal axes. When multiple ensembles are selected each graph will be overlayed on top of each other transparently, where each ensemble gets its own color.

Parameter comparison viewer

Every selected parameter is visualized in the parallel coordinates plot, where each vertical axis represents a single parameter. Multiple ensembles are distinguished by different colors.

Response correlation viewer

We can select both responses (simulation results) and parameters, where the goal of the correlation viewer is to visualize how big is the influence of the selected set of parameters on the selected set of responses. Or in other words, it describes how the response behaves if there is some change in a parameter. One can choose from two correlation methods: spearman and pearson. Once we select at least one response and one parameter, we are provided with the following 4 sub-views: 1) Response overview, which visualizes the currently selected response. By clicking along the time axis we can select a new index / timestep (a dashed black line) for correlation analysis. 2) Scatterplot depicting the selected response against the selected parameter. It is accompanied by distribution plots for both selections. 3) Heatmap based representation of correlation within interval (-1, 1) across all selected responses and parameters. Each response has its own column, where each row represents one of the selected parameters. We can select the currently active response and parameter by clicking directly on a heatmap. 4) Correlation BarChart for the single selected response (from heatmap) and all selected parameters in a descending order according to the correlation.