GIS Case Study
A public GIS portal visualising a national network of 300+ automatic weather stations in real time.
Domain
Weather Observation Network / Public Portal
Role
Full-Stack GIS Developer

At A Glance
Live observations from 300+ automatic weather stations across the country
Sixteen switchable metrics: temperature, humidity, rain rate, wind, pressure and more
Satellite, terrain and 3D basemaps with a full layer and boundary tree
Auto-refreshing observations with a live data-freshness indicator
A national network of more than 300 automatic weather stations was reporting observations every few minutes into an internal system that only the meteorological service could see. The public, the media, and downstream users such as agriculture and aviation got summaries hours later. The data existed in near real time; the access to it did not. The portal was commissioned to publish the network live: every station, every observation, on a public map, with the metric switching and context a general user needs to make sense of it, and reliable enough to stay up when a weather event sends traffic surging.
The front end is React with Mapbox GL JS, and station observations are served as vector tiles for fast rendering across the whole network. Sixteen metrics are switchable, temperature, dew point, wet bulb, humidity, rain rate, pressure, wind and the rest, and station markers are colour-graded against a live legend keyed to the selected metric. A network-wide minimum and maximum readout names the extreme stations, which is the headline most visitors are looking for. Observations auto-refresh and a data-freshness indicator shows when the map was last updated, because a public weather map that might be stale is worse than none. Station search, sorting and a nearby-station lookup by latitude and longitude help users find their location. A layer tree covers map type, administrative boundaries, satellite imagery and terrain, with 3D basemaps available. Everything is tuned for the traffic pattern of a public weather service: quiet most days, and then everyone at once.
The meteorological service now publishes what it observes as it observes it. Journalists pull current extremes from the map; farmers check the nearest station; the service's own forecasters use it as a quick network overview. Traffic spikes during storms have been absorbed without incident, which is the test a public portal either passes or does not. The interesting engineering was in freshness and load rather than cartography. A live map has to be honest about its own latency and has to survive its busiest day. That combination shows up in most public-facing web application development we do for agencies with real-time data.
Capabilities
Metric switcher across stations, temp, dew point, wet bulb, heat index, feels like, wind chill, humidity, solar radiation, wind speed, wind gust, rain rate, daily / monthly / yearly rain and pressure
Colour-graded station markers with a live legend keyed to the active metric
Network-wide min / max readout naming the extreme stations
Station search, sorting and nearby-station lookup by latitude / longitude
Layer tree covering map type, boundaries, satellite imagery and station visibility
Satellite basemap, 3D terrain and bearing controls
Light, auto and dark interface themes
Technologies
Product



Questions
Stations report every few minutes and the map auto-refreshes to match. A freshness indicator shows the time of the last update, so a user always knows how old the picture is rather than assuming it is live.
Sixteen, including temperature, dew point, wet bulb, humidity, rain rate, pressure, wind speed and direction. Switching metric recolours every station marker against a legend for that metric and updates the network-wide min/max readout.
Yes. Observations are served as vector tiles behind a cache, and the front end is static, so load scales without hitting the observation database directly. The portal has absorbed storm-day traffic spikes without degradation.
Yes. Station search, sorting and a nearby-station lookup by latitude and longitude are built in, and the map can be filtered to a region using the administrative boundary layer.
Contact Us
Have a project in mind? Tell us what you're building and we'll get back to you within 12–24 hours with a clear plan.
🔒
100% Confidential
⚡
12–24 Hr Response
🛡️
60-Day Bug Fix
✅
Free Consultation
💬
Start Your Project
Fill in the details below or book a call