Data Sources & Methodology

Live data status

PM2.5 smoke forecast: FMI SILAM

All PM2.5 (fine particulate smoke) values on this dashboard come from FMI SILAM (System for Integrated modeLling of Atmospheric coMposition), the operational atmospheric composition model run by the Finnish Meteorological Institute. SILAM publishes a real global operational forecast once a day, each run covering a 7-day horizon, on a public THREDDS data server.

This app pulls the cnc_PM2_5 variable (surface-level PM2.5 concentration) from the silam_glob_v6_1_sfc global model dataset — a 0.2°×0.2° grid covering the whole planet, published hourly. This is a live numeric grid, not the PNG map images SILAM's public viewer shows; PNGs cannot be used to compute a province-level statistic, so this app queries SILAM's NetCDF Subset Service (NCSS) directly for the raw numbers.

How a province's daily PM2.5 number is calculated

1. A bounding box is computed around every Sub-Saharan African country, then the cnc_PM2_5 grid for that box and the forecast horizon (5 days by default) is downloaded from SILAM's THREDDS server.
2. Units are converted from kg/m³ (SILAM's native unit) to µg/m³.
3. Hourly grid values are aggregated into one daily-mean and one daily-maximum value per grid cell.
4. For every province polygon (from the real admin-1 boundaries below), every grid cell whose center falls inside that province is identified, and the province's daily mean/max PM2.5 is the average/maximum of those cells (a "zonal statistic"). Very small provinces that don't contain a full grid cell fall back to the nearest grid cell to their centroid.
5. The resulting province-day table is loaded into the dashboard's database.

If a scheduled refresh fails (network issue, upstream outage), the previous successfully ingested forecast is left in place rather than being overwritten with missing or broken data — the dashboard always shows the last real forecast that was successfully fetched, and the live status box above shows exactly when that was.

Administrative boundaries

Country and province outlines come from Natural Earth's 10m-resolution admin-0 (country) and admin-1 (province/state/region) cultural vector data, filtered to the 49 countries of Sub-Saharan Africa (the UN M49 statistical definition — all of Africa except Algeria, Egypt, Libya, Morocco, Tunisia, and Western Sahara).

Risk categories

Each province-day is classified from its daily-mean PM2.5 value:

A province is also upgraded to Very High if both its daily mean and daily maximum are elevated at once. These are communication categories for this proof of concept, not clinical or regulatory thresholds — before any operational use, they should be reviewed and approved by environmental health, maternal health, and meteorology stakeholders.