Operational air quality dashboard and decision-support system

The problem

Many agencies and programmes already have pieces: a few reference monitors or a low-cost sensor network, access to satellite products, maybe a forecast run by a partner institute. These live in separate spreadsheets, FTP folders and portals. Nobody can answer "what is the air like today, and what is coming tomorrow?" in one view, and pilot dashboards often stop updating once the project team moves on. The task is an integrated, automated system with a clear data model that the in-house team can operate.

What you get

• Requirements and data inventory: stations, satellite products, model outputs, users and decisions supported
• PostgreSQL/PostGIS air quality database with documented schema for sites, parameters, measurements and QA flags
• Automated ingestion jobs for station feeds, satellite products and model output, with logging and failure alerts
• Gridded data services (THREDDS/ncWMS or GeoServer) for model and satellite layers
• Web dashboard: station time series, AQI categories, map layers and forecast panels
• Deployment on the client's Linux server or cloud, an operations runbook and hand-over training

How it works

1. 1

   Scope

   Map the users, the decisions they make (public advisories, school closures, enforcement, reporting) and every data source with its format, latency and licence. Agree which AQI scale and pollutants to show.

2. 2

   Data model and pipelines

   Design the database and write Python ingestion and QA/QC scripts for each source. Range, spike and flat-line checks are applied as flags. Raw data is never overwritten.

3. 3

   Services and dashboard

   Publish gridded layers through standard OGC/OPeNDAP services and build the dashboard against them. Review it with users through iterative demonstrations.

4. 4

   Deploy and harden

   Deploy on Linux with scheduled jobs, monitoring and backups, then run in parallel with existing reporting until the outputs are trusted.

5. 5

   Hand over

   Provide documentation, a runbook and hands-on training for the operations team, plus optional retainer support.

Tools

• Python (Pandas, xarray)
• PostgreSQL/PostGIS
• THREDDS Data Server
• ncWMS
• GeoServer
• Tethys Platform / Django
• React or Vue.js
• Linux