WeatherStar 4000+

A live version of this project is available at https://weatherstar.netbymatt.com

About

This project aims to bring back the feel of the 90s with a weather forecast that has the look and feel of The Weather Channel at that time but available in a modern way. This is by no means intended to be a perfect emulation of the WeatherStar 4000, the hardware that produced those wonderful blue and orange graphics you saw during the local forecast on The Weather Channel. If you would like a much more accurate project please see the WS4000 Simulator. Instead, this project intends to create a simple to use interface with minimal configuration fuss. Some changes have been made to the screens available because either more or less forecast information is available today than was in the 90s. Most of these changes are captured in sections below.

What's your motivation

Nostalgia. And I enjoy following the weather, especially severe storms.

It's also a creative outlet for me and keeps my programming skills honed for when I need them for my day job.

Included technology

I've kept this open source, well commented, and made it as library-free as possible to help others interested in programming be able to jump right in and start working with the code.

From a learning standpoint, this codebase make use of a lot of different methods and technologies common on the internet including:

• The https://api.weather.gov REST API. (documentation).
• ES 6 functionality
  • Arrow functions
  • Promises
  • Async/await and parallel loading of all forecast resources
  • Classes and extensions
  • Javascript modules
• Separation between API code and user interface code
• Use of a modern date parsing library luxon
• Practical API rates and static asset caching
• Very straight-forward hand written HTML
• Build system integration (Gulp, Webpack) to reduce the number of scripts that need to be loaded
• Hand written CSS made easier to mange with SASS
• A linting library to keep code style consistent

Does WeatherStar 4000+ work outside of the USA?

This project is tightly coupled to NOAA's Weather API, which is exclusive to the United States. Using NOAA's Weather API is a crucial requirement to provide an authentic WeatherStar 4000+ experience.

If you would like to display weather information for international locations (outside of the USA), please checkout a fork of this project created by @mwood77:

• ws4kp-international

Deployment Modes

WeatherStar 4000+ supports two deployment modes:

Server Deployment (Recommended)

• Includes Node.js server with caching proxy for better performance (especially when running on a local server for multiple clients)
• Server-side request deduplication and caching
• Weather API observability and logging
• Used by: npm start, DIST=1 npm start, and Dockerfile.server

Static Deployment

• Pure client-side deployment using nginx to serve static files
• All API requests are made directly from each browser to the weather services
• Browser-based caching
• Used by: static file hosting and default Dockerfile

Run Your WeatherStar

Ensure you have Node installed. Clone the repository:

git clone https://github.com/netbymatt/ws4kp.git
cd ws4kp
npm install

Development Mode (individual JS files, easier debugging)

npm start

Development Mode without proxy caching

STATIC=1 npm start

Production Mode (minified/concatenated JS, faster loading)

npm run build
DIST=1 npm start

Production Mode without proxy caching (simulates static Docker deployment)

npm run build
STATIC=1 DIST=1 npm start

For all modes, access WeatherStar by going to: http://localhost:8080/

Key Differences

Development Mode (npm start):

• Uses individual JavaScript module files served directly
• Easier debugging with source maps and readable code
• Slower initial load (many HTTP requests for individual files)
• Live file watching and faster development iteration

Production Mode (DIST=1 npm start):

• Uses minified and concatenated JavaScript bundles
• Faster initial load (fewer HTTP requests, smaller file sizes)
• Optimized for performance with multiple clients
• Requires npm run build to generate optimized files

Docker Deployments

To run via Docker using a "static deployment" where everything happens in the browser (no server component, like STATIC=1):

docker run -p 8080:8080 ghcr.io/netbymatt/ws4kp

To run via Docker using a "server deployment" with a caching proxy server for multi-client performance and enhanced observability (like npm run build; DIST=1 npm start):

docker build -f Dockerfile.server -t ws4kp-server .
docker run -p 8080:8080 ws4kp-server

To run via Docker Compose (shown here in static deployment mode):

---
services:
  ws4kp:
    image: ghcr.io/netbymatt/ws4kp
    container_name: ws4kp
    environment:
      # Each argument in the permalink URL can become an environment variable on the Docker host by adding WSQS_
      # Following the "Sharing a Permalink" example below, here are a few environment variables defined. Visit that section for a
      # more complete list of configuration options.
      - WSQS_latLonQuery="Orlando International Airport Orlando FL USA"
      - WSQS_hazards_checkbox=false
      - WSQS_current_weather_checkbox=true
    ports:
      - 8080:8080 # change the first 8080 to meet your local network needs
    restart: unless-stopped

Serving a static app

There are several ways to deploy WeatherStar as a static app that runs entirely in the browser:

Manual static hosting (Apache, nginx, CDN, etc.): Build static distribution files for upload to any web server:

npm run build

The resulting files in /dist can be uploaded to any web server; no server-side scripting is required.

Docker static deployment: The default Docker image uses nginx to serve pre-built static files:

docker run -p 8080:8080 ghcr.io/netbymatt/ws4kp

Node.js in static mode: Use the Node.js server as a static file host without the caching proxy:

STATIC=1 npm start          # Use Express to serve development files
STATIC=1 DIST=1 npm start   # Use Express to serve (minimized) production files

What's different

I've made several changes to this Weather Star 4000 simulation compared to the original hardware unit and the code that this was forked from.

• Radar displays the timestamp of the image.
• A new hour-by-hour graph of the temperature, cloud cover and precipitation chances for the next 24 hours.
• A new hourly forecast display for the next 24 hours is available, and is shown in the style of the travel cities forecast. (off by default because it duplicates the hourly graph)
• The SPC Outlook is shown in the style of the old air quality screen. This shows the probability of severe weather over the next 3 days at your location.
• The "Local Forecast" and "Extended Forecast" provide several additional days of information compared to the original format in the 90s.
• The original music has been replaced. More info in Music.
• Marine forecast (tides) is not available as it is not reliably part of the new API.
• "Flavors" are not present in this simulation. Flavors refer to the order of the weather information that was shown on the original units. Instead, the order of the displays has been fixed and a checkboxes can be used to turn on and off individual displays. The travel forecast has been defaulted to off so only local information shows for new users.

Sharing a permalink (bookmarking)

Selected displays, the forecast city and widescreen setting are sticky from one session to the next. However if you would like to share your exact configuration or bookmark it, click the "Copy Permalink" (or get "Get Permalink") near the bottom of the page. A URL will be copied to your clipboard with all of you selected displays and location (or copy it from the page if your browser doesn't support clipboard transfers directly). You can then share this link or add it to your bookmarks.

Your permalink will be very long. Here is an example for the Orlando International Airport:

https://weatherstar.netbymatt.com/?hazards-checkbox=false&current-weather-checkbox=true&latest-observations-checkbox=true&hourly-checkbox=false&hourly-graph-checkbox=true&travel-checkbox=false&regional-forecast-checkbox=true&local-forecast-checkbox=true&extended-forecast-checkbox=true&almanac-checkbox=false&spc-outlook-checkbox=true&radar-checkbox=true&settings-wide-checkbox=false&settings-kiosk-checkbox=false&settings-scanLines-checkbox=false&settings-speed-select=1.00&settings-units-select=us&latLonQuery=Orlando+International+Airport%2C+Orlando%2C+FL%2C+USA&latLon=%7B%22lat%22%3A28.431%2C%22lon%22%3A-81.3076%7D

You can also build your own permalink. Any omitted settings will be filled with defaults. Here are a few examples:

https://weatherstar.netbymatt.com/?latLonQuery=Orlando+International+Airport
https://weatherstar.netbymatt.com/?kiosk=true
https://weatherstar.netbymatt.com/?settings-units-select=metric

Kiosk mode

Kiosk mode can be activated by a checkbox on the page. Note that there is no way out of kiosk mode (except refresh or closing the browser), and the play/pause and other controls will not be available. This is deliberate as a browser's kiosk mode it intended not to be exited or significantly modified. A separate full-screen icon is available in the tool bar to go full-screen on a laptop or mobile browser.

It's also possible to enter kiosk mode using a permalink. First generate a Permalink, then to the end of it add &kiosk=true. Opening this link will load all of the selected displays included in the Permalink, enter kiosk mode immediately upon loading and start playing the forecast.

Default query string parameters (environment variables)

When serving this via the built-in Express server, it's possible to define environment variables that direct the user to a default set of parameters (like the Permalink above). If a user requests the root page at http://localhost:8080/ the query string provided by environment variables will be appended to the url thus providing a default configuration.

Environment variables can be added to the command line as usual, or via a .env file which is parsed with dotenv. Both methods have the same effect.

Environment variables that are to be added to the default query string are prefixed with WSQS_ and then use the same key/value pairs generated by the Permalink above, with the - (dash) character replaced by an _ (underscore). For example, if you wanted to turn the travel forecast on, you would find travel-checkbox=true in the permalink, its matching environment variable becomes WSQS_travel_checkbox=true.

When using the Docker container, these environment variables are read on container start-up to generate the static redirect HTML.

Music

The WeatherStar had wonderful background music from the smooth jazz and new age genres by artists of the time. Lists of the music that played are available by searching online, but it's all copyrighted music and would be difficult to provide as part of this repository.

I've used AI tools to create WeatherStar-inspired music tracks that are unencumbered by copyright and are included in this repo. To keep the size down, I've only included 4 tracks. Additional tracks are in a companion repository ws4kp-music.

If you're looking for the original music that played during forecasts TWCClassics has thorough documentation of playlists.

Customizing the music

WeatherStar 4000+ supports background music during forecast playback. The music behavior depends on how you deploy the application:

Express server modes (npm start, DIST=1 npm start, or Dockerfile.server)

When running with Node.js, the server generates a playlist.json file by scanning the ./server/music directory for .mp3 files. If no files are found in ./server/music, it falls back to scanning ./server/music/default/. The playlist is served dynamically at the /playlist.json endpoint.

Adding your own music: Place .mp3 files in ./server/music/

Docker server example:

docker build -f Dockerfile.server -t ws4kp-server .
docker run -p 8080:8080 -v /path/to/local/music:/app/server/music ws4kp-server

Static hosting modes (default Dockerfile, nginx, Apache, etc.)

When hosting static files, there are two scenarios:

Static Docker deployment: The build process creates a playlist.json file with default tracks, but the Docker image intentionally removes it to force browser-based directory scanning. The browser attempts to fetch playlist.json, receives a 404 response with the X-Weatherstar header, which causes it to fallback to scanning the music/ directory.

Manual static hosting: If you build and upload the files yourself (npm run build), playlist.json will contain the default tracks unless you customize ./server/music/ before building.

For directory scanning to work properly:

• Your web server must generate directory listings for the music/ path
• Your web server must set the X-Weatherstar: true header (the provided nginx configuration does this)

Adding your own music: Place .mp3 files in music/ (or bind mount to /usr/share/nginx/html/music for Docker)

Docker static example:

docker run -p 8080:8080 -v /path/to/local/music:/usr/share/nginx/html/music ghcr.io/netbymatt/ws4kp

Subdirectories will not be scanned. When WeatherStar loads in the browser, it randomizes the track order and reshuffles on each loop through the playlist.

Music doesn't auto play

Ws4kp is muted by default, but if it was unmuted on the last visit it is coded to try and auto play music on subsequent visits. But, it's considered bad form to have a web site play music automatically on load, and I fully agree with this. Chrome and Firefox have extensive details on how and when auto play is allowed.

Chrome seems to be more lenient on auto play and will eventually let a site auto-play music if you're visited it enough recently and manually clicked to start playing music on each visit. It also has a flag you can add to the command line when launching Chrome: chrome.exe --autoplay-policy=no-user-gesture-required. This is the best solution when using Kiosk-style setup.

If you're unable to pre-set the play state before entering kiosk mode (such as with a home dashboard implementation) you can add the query string value below to the url. The browser will still follow the auto play rules outlined above.

?settings-mediaPlaying-boolean=true

Community Notes

Thanks to the WeatherStar community for providing these discussions to further extend your retro forecasts!

• Stream as FFMPEG
• Weather like it's 1999 Raspberry pi, streaming, music and CRT all combined into a complete solution.
• ws4channels A Dockerized Node.js application to stream WeatherStar 4000 data into Channels DVR using Puppeteer and FFmpeg.

Customization

A hook is provided as server/scripts/custom.js to allow customizations to your own fork of this project, without accidentally pushing your customizations back upstream to the git repository. A sample file is provided at server/scripts/custom.sample.js and should be renamed to custom.js activate it.

When using Docker:

• Static deployment: Mount your custom.js file to /usr/share/nginx/html/scripts/custom.js
• Server deployment: Mount your custom.js file to /app/server/scripts/custom.js

Issue reporting and feature requests

Please do not report issues with api.weather.gov being down. It's a new service and not considered fully operational yet. I've also observed that the API can go down on a regional basis (based on NWS office locations). This means that you may have problems getting data for, say, Chicago right now, but Dallas and others are working just fine.

Before reporting an issue or requesting a feature please consider that this is not intended to be a perfect recreation of the WeatherStar 4000, it's a best effort that fits within what's available from the API and within a web browser.

Note: not all units are converted to metric, if selected. Some text-based products such as warnings are simple text strings provided from the national weather service and thus have baked-in units such as "gusts up to 60 mph." These values will not be converted.

Related Projects

Not retro enough? Try the Weatherstar 3000+

Use

Linking directly to the live web site at https://weatherstar.netbymatt.com is encouraged. As is using the live site for digital signage, home dashboards, streaming and public display. Please note the disclaimer below.

Acknowledgements

This project is based on the work of Mike Battaglia. It was forked from his work in August 2020.

• Mike Battaglia for the original project and all of the code which draws the weather displays. This code remains largely intact and was a huge amount of work to get exactly right. He's also responsible for all of the background graphics including the maps used in the application.
• The team at TWCClassics for several resources.
  • A font set used on the original WeatherStar 4000
  • Icon sets
  • Countless photos and videos of WeatherStar 4000 forecasts used as references.
• The growing list of contributors to this repository

Disclaimer

This web site should NOT be used in life threatening weather situations, or be relied on to inform the public of such situations. The Internet is an unreliable network subject to server and network outages and by nature is not suitable for such mission critical use. If you require such access to NWS data, please consider one of their subscription services. The authors of this web site shall not be held liable in the event of injury, death or property damage that occur as a result of disregarding this warning.

The WeatherSTAR 4000 unit and technology is owned by The Weather Channel. This web site is a free, non-profit work by fans. All of the back ground graphics of this web site were created from scratch. The icons were created by Charles Abel and Nick Smith (http://twcclassics.com/downloads/icons.html) as well as by Malek Masoud. The fonts were originally created by Nick Smith (http://twcclassics.com/downloads/fonts.html).