Skip to content
/ Nimbus Public

Visualization package for large-eddy simulations

License

Notifications You must be signed in to change notification settings

CliMA/Nimbus

Repository files navigation

CLiMA – NIMBUS Prototype


Core team:
Martin Bernard (CS Lead, ArtCenter College of Design)
Rachel Rose Waterhouse (Design Lead, Frontend Developer, ArtCenter College of Design)
Barbara Alonso (Design Lead, Harvard School of Engineering)

CliMA / Caltech collaborators:
PI: Tapio Schneider
Akshay Sridhar
Lenka Novakova
Yair Cohen

Project Mentors:
Scott Davidoff, NASA Jet Propulsion Laboratory
Santiago Lombeyda, Caltech + ArtCenter College of Design
Maggie Hendrie, ArtCenter College of Design
Hillary Mushkin, Caltech


Info

This project was created with React, bootstrapped with create-react-app see documentation, and d3.js. The project can deployed as a standalone desktop application via Electron, but this would require some additional configuration.

Setup

Make sure you have installed Node & NPM. One way to do this is with Homebrew.

Clone the repo, cd into the directory, and then run npm install to install the required dependencies for the project.

App Structure

As mentioned above, this project was initially created with create-react-app. Therefore, Webpack and Babel are preconfigured. If you wish to create a custom configuration, you will need to run npm run eject, which will extract the configuration files for you that you can then change.

The entry point of the application is index.js. That is where we import the App React component, which contains the child components SlicesContainer and DiagnosticPlotsContainer. These correspond in general to the top and bottom views of the prototype.

Similarly, these child components also import other components - for example, SlicesContainer uses the HorizontalSlice and VerticalSlice functional components.

The data visualizations are created with d3, and use both SVG and Canvas for rendering.

DEVELOPMENT - Web

In a terminal window, run node server.js to start the server (running on port 8080). In another terminal winow, run npm run start and open a browser to localhost:3000.

The start script is run from the react-scripts module as part of create-react-app see here, which uses its own webpack config. We have not changed that. To make a custom config, you must first eject (see aforementioned link to available scripts above). Additionally, the create-react-app has a /public folder. In development, src/index.js is looking for the index.html inside /public, as it appends everything within <div id="root"></div>.

Calls to the node server running on 8080 are done via proxy so that calls can be made from port 3000 - this is set up in package.json.

PRODUCTION - Web

To create a production build for the web, run npm run build in terminal. This will create the bundled js files in /build > /static > /js. The index.html from the /public is copied to the build folder that is generated. Once that is complete, you should be able to run the server again with node server.js, which will then serve the build directory. You should be able to open a browser tab to localhost:8080.

PRODUCTION - Electron

See package.json scripts section and script notes to create a production Electron app. In particular, see the package script which will generate executables for both Mac and Windows inside of the ./dist directory See notes that we followed here.

Notes

This is a prototype that was built to visualize BOMEX LES data provided by the CLIMA team. It can be extended to other types of data.

During the prototyping process we used Observable notebooks to develop each individual feature included in the final design. The notebook where this all happened is open to anyone who has the link.

For the project, we converted some of the data we were given to JSON so that we could quickly consume the data in this environment. Ideally, we would be able to directly import .nc files or have a pipeline that would convert those files to JSON. We used Julia notebooks to convert these files, and that code is also included in this repository, along with some of our very early Julia prototypes that access the .nc files directly.

The variables available in the contour plot dropdowns, the diagnostic plots, and the timeline variable scrubber (the TLV dropdown in the UI, default variable shown is TKE) are hardcoded - we chose a subset for the purposes of the demo. Ideally, this list of variables would be determined based on the data uploaded.

For the purposes of this prototype and demo, we use a subset of the data that we were provided, and is imported in the DiagnosticPlotsContainer and the SlicesContainer components. Potentially this would be moved up a level, so that the parent App.js component would have support for uploading the data (which at time of writing is not implemented/supported), then passing it to the respective child components. We have not tested this with a full set of simulation data, but recognize that there is additional work that would need to be done to support this.

Responsiveness has not been implemented in detail - there are some variables that we set that attempt to account for window size, but not resize. As a result, you may need to refresh on screen resize.

About

Visualization package for large-eddy simulations

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •