A weekly utilisation series for named plants.
Vehicle counts, plume presence and thermal signal from satellite passes, combined per plant and delivered as CSV.
Spec table
- Platform
- VHR satellite / aerial
- GSD
- VHR, 0.5 m or finer
- Spectral
- RGB plus a thermal band
- Cadence
- weekly, as a target
- Delivery
- CSV, one row per plant per week
- Unit
- a single plant, or a sector rollup of a covered group of plants
- Signals
- vehicle count, plume presence, thermal signal
The files
- File
Plant utilisation series
- Format
CSV- What's in it
- One row per plant per week. Carries the week-ending date, the plant identifier you supplied, the utilisation value and the three signals it was built from. Weeks with no usable pass are left as gaps.
- File
Sector rollup
- Format
CSV- What's in it
- The utilisation series of a covered group of plants combined into one weekly series, on the sector grouping you agree at the start. A gap at plant level shows up as a lower plant count for that week, and nothing fills it.
- File
Signal columns
- Format
CSV columns- What's in it
- Vehicle count, plume presence and thermal signal are delivered next to the combined value, so you can weight them yourself or drop one from your model.
- File
Weekly feed
- Format
CSV, weekly- What's in it
- A new file each week for the pilot period, for your model to pick up. It lands once the week's pass has been processed, so the week-ending date on each row tells you which pass it came from and you can line it up against the release calendar you already keep.
How it works
- A
Supply the plants and the grouping
A list of plants with your own identifiers and coordinates, and the sector group you want rolled up. If you key plants to NACE or NAICS codes, send those too, so the rollup matches the breakdown in the official release you compare against.
- B
Weekly imagery
Each plant is imaged on a weekly pass with an RGB and a thermal band. Cloudy passes are marked as gaps.
- C
Read the signals
Vehicles are counted, plume presence is recorded, and surface heat signature is read at each site.
- D
Combine and deliver
The three signals are combined into one utilisation value per plant and week, rolled up by sector, and sent as CSV.
Limits
- Condition
Cloud cover leaves gaps
- Effect on the series
- A cloudy pass can leave a plant with no usable reading that week.
- What that means
- The series shows the gap. It is never interpolated.
- Condition
Surface heat, not process temperature
- Effect on the series
- The thermal band reads heat at the surface of the site.
- What that means
- It says a plant looks active. It says nothing about furnace or reactor temperature.
- Condition
Thermal is coarser than the visible bands
- Effect on the series
- The thermal signal is read for the site as a whole.
- What that means
- It cannot separate one stack or one unit from its neighbour.
- Condition
Activity, not output
- Effect on the series
- The series reads how busy a site looks.
- What that means
- It does not give tonnes, units or product type.
- Condition
0.5 m sets the vehicle floor
- Effect on the series
- At 0.5 m a car or truck is only a few pixels wide.
- What that means
- Counts are good for trend at a plant. They are not a headcount of vehicles.
- Condition
Plants in scope only
- Effect on the series
- Coverage is the plants and sectors agreed for the pilot.
- What that means
- It is not a global panel.
- Condition
No forecast
- Effect on the series
- The series describes past weeks.
- What that means
- It does not predict a print, GDP or a company's earnings.
Why this exists
You read the industrial cycle from the monthly production release and the surveys, both of which arrive after the month they describe and get revised. Between prints there is little below the level of the whole sector to look at.
When you want to know whether one cluster of steel or cement plants is running harder than the last print implies, you ring a contact or guess. This adds a plant-level weekly read to put beside the official series.
Who it's for
- Macro analysts and economists who cover industrial production and want a plant-level read between prints.
- Quant researchers on an alternative data team who will test the series against their own nowcast.
- Sector analysts who need a rollup for a named group of plants.
Who it's not for
- Teams that need tonnes, units or product mix. The series reads activity only.
- A forecast of GDP, official industrial production or company earnings.
- A replacement for official statistics. It sits beside them.
- Real-time monitoring. Weekly passes are the cadence.
Questions from evaluators
Does this replace official industrial production data?
No. It is a higher-frequency, plant-level complement that you line up beside the official releases.
What does the thermal band detect?
Surface heat signature consistent with a site running. It does not read internal process temperature.
Does cloud cover block a weekly read?
It can. A cloudy pass can leave a plant with no usable reading that week, and the file shows it as a gap.
Can it tell what a plant produces?
No. It reads activity signals, not product type or output volume.
Is coverage global?
Coverage follows the plants and sectors in scope for the pilot. Ask about specific names.
Can we test it against our own nowcast?
Yes. The signal columns and weekly dates are delivered so you can join them to your own series and check the fit yourself. We do not supply backtests.
Who owns the data we receive?
Use and ownership terms for the delivered files are written into the pilot agreement before it starts.
How is a pilot priced?
Scoped by number of plants, sectors and pilot length, and quoted in writing. Nothing is priced on this page.
Send the plant list and we will tell you what a pilot covers.
Start a pilot
Or ask about your sector in .