CTRL LOGICTesela · User manual

Chapter 16

FAQ, glossary and limits

This chapter brings together the most common questions, the meaning of the technical terms you will see in Tesela and the limits of the system, stated plainly from the start. Knowing what can be seen and what cannot is what allows you to trust what can be seen.

Frequently asked questions

Does Tesela tell me whether an activity is illegal?

No. Tesela detects indications: alterations of the territory or activity consistent with something, and prioritizes them to decide where to look first. Every detection requires verification. The conclusion is reached by the competent institution through its own procedure. It is a technical support tool; it does not replace administrative procedures or field inspection.

Does detection use artificial intelligence?

No. Detection is mathematical and explainable: spectral indices are calculated on satellite images and thresholds are applied. You can see from which scenes, with which parameters and when each result was produced. Artificial intelligence is used only in the “Ask the map” assistant, which only reads and queries.

Why does an alteration that I know exists not appear?

It may be due to the resolution of the images (a very small intervention is not visible), to clouds during the chosen period, to a threshold that is too high, or because the alteration occurred outside the dates compared. Try other periods or adjust the thresholds; the “Honest limits” section explains each cause.

Why does an analysis take a while?

Analyses are processed in the background and request the imagery from the configured provider. The sheet or the monitoring panel shows progress by stages. A long period or a large area takes longer.

What is the difference between a live layer and a snapshot?

The live layer changes when new data arrives and is for exploring. The snapshot is a frozen, immutable copy, with a record and SHA-256 fingerprint, and is for delivering or citing. See “Snapshots and sources”.

I do not see a tool that another person sees. Why?

Each tool can be public, signed-in only or hidden (administrator, “Viewer options”), and some additionally depend on a permission in your profile: “Field”, “Assistant”, “Monitoring”. Ask the administrator if you need it.

What do I do if the assistant shows “The request cannot be processed right now”?

Contact the administrator. The assistant may not be configured, or there may be a problem with the provider; the detail stays in the server log and the person asking never sees it.

Can I cross-reference the results with official concessions or cadastres?

Only if the institution that owns the data authorizes it through an agreement. Official third-party data is always used in read-only mode. The protected areas in the demonstration come from OpenStreetMap and are indicative, not official.

Can I use Tesela on an iOS phone?

The “Tesela Field” field app is available for Android. The iOS version is on its way.

Glossary

TermMeaning
IndicationA detected signal that suggests an alteration or activity and that requires verification. It is neither proof nor a determination.
Requires verificationNotice that accompanies every finding: before acting, someone must check it against other sources or on the ground.
Sentinel-2Satellites of the European Copernicus programme that image the Earth’s surface with 10 to 20 m pixels. They are the source of Tesela’s scenes (level L2A, atmospherically corrected).
SceneA satellite image of a period, or a composite of several dates, from which the indices are derived.
NDVIVegetation index: high values indicate dense, active vegetation; a drop between two dates may indicate loss of cover.
NDWIWater index: highlights water bodies and moisture.
NDSISnow index: highlights snow-covered surfaces.
BSIBare soil index: highlights surfaces without vegetation.
NBRBurn index: combines infrared bands to highlight surfaces affected by fire.
dNBRDifference in NBR between before and after a fire. It is used to estimate severity; without local calibration it may underestimate it.
SeverityClass estimated from the dNBR (low, moderate-low, moderate-high, high). It is an estimate that requires verification.
ThresholdHow much an index must vary to be considered a change. It is calibrated with real cases.
Heat hotspotThermal anomaly detected by satellite (NASA FIRMS). It is consistent with burning or other heat sources; each hotspot is an indication.
SnapshotFrozen, immutable copy of the state of the data on a date, published as a layer with its record.
SHA-256 fingerprintDigital signature of the content: if the data changes, the fingerprint changes. It makes it possible to verify that what was delivered is what was published.
ISO 19115 recordStandard metadata record that describes a layer: what it is, where it comes from, how it was produced and under what conditions it can be used.
WMSStandard service that delivers maps as images.
WFSStandard service that delivers vector features with their attributes.
GeoServerMap server that publishes Tesela’s layers as WMS and WFS.
COGCloud-optimized GeoTIFF: the format in which rasters are stored so they display quickly on the map.
GSDSize on the ground covered by one pixel of a photo, in centimeters per pixel. The lower the GSD, the more detail.
OverlapPercentage by which consecutive photos (front) or neighboring lines (side) overlap in a drone mission.
Live layer / reference layerLive: updated with new data. Reference: context data such as boundaries or protected areas.
Profile and rankThe profile is a named set of permissions; the rank orders the profiles and limits whom each person can manage.

Honest limits

  1. Resolution. Sentinel-2 pixels measure 10 to 20 m. Interventions down to fractions of a hectare are detected, but not small pits or activity under cover.
  2. Clouds. In tropical areas and in the rainy season there may be no clean image for weeks. This is mitigated by compositing several dates. Radar (Sentinel-1), which penetrates clouds, is planned but not yet available.
  3. Thresholds. How much an index must vary to be an alert cannot be guessed: it is calibrated with real cases. Without local calibration, severity may be underestimated.
  4. Official data. Cross-referencing the results with concessions or cadastres of another institution requires an agreement with whoever owns the data.
  5. Protected areas. The demonstration layer is indicative and non-exhaustive: its boundaries are not the official ones. The official layer depends on the source that each institution decides.
  6. Artificial intelligence. Detection does not use it. The assistant is optional, read-only and depends on the provider and key that the institution configures.

Cautious language

Tesela is written so as not to state what it cannot know. A satellite image shows that something changed, not why or whether it was permitted. That is why the system and this manual speak this way, and we ask that your reports do so too:

Instead of…Write…
“Illegal mining”“Anomaly detected” or “territorial alteration”
“Criminal activity”“Activity consistent with…”
“Confirmed exploitation”“Possible activity…”, “requires verification”
“Proof of a crime”“Indication that requires verification”
“Detects offenders”“Prioritizes areas for inspection”

Summary