Methodology · traceable evidence
Understand what the map can tell you.
datacenter-geo separates observations, source claims, display models and unknowns so early site research can be examined and repeated.
Inspect the source dataFour things that must stay separate
| Information | What it supports | What it does not establish |
|---|---|---|
| Directory record | A source lists an entity at a location. | Current operation, building identity, spare capacity or a verified footprint. |
| Mapped infrastructure | A source records geometry or infrastructure attributes. | A utility connection offer, permission, ownership or usable capacity. |
| Modeled screening signal | A reproducible transformation of stated inputs. | Investment merit, expected return or development approval. |
| Evidence gap | A needed observation is unavailable or unverified. | A negative physical condition or a zero value. |
Substation proximity in Bhopal
The legacy proximity display samples the pilot at 2 km spacing, calculates straight-line distance to the nearest mapped substation centroid, and fades intensity over 10 km. Its display formula is max(0, 1 − distance_km / fade_km).
This is Tier D modeled context derived from mapped input geometry. A nearby substation can have no available connection capacity. A centroid is not a verified electrical connection point, and straight-line distance is not a feasible cable route.
Explainable MP and Bhopal research cells
MP uses a 10 km grid clipped to the mapped state boundary; Bhopal retains a separate 2 km grid clipped to its research extent. Each grid intersects mapped industrial land and water, and measures great-circle distance from a cell reference point to the nearest mapped substation centroid. Overlapping source polygons are unioned before measuring area. The source boundary and each grid's coverage are retained in the download.
- Any positive mapped water intersection triggers a review label.
- A missing input category remains unknown.
- At least 1% mapped industrial overlap, a mapped substation within 5 km, and no mapped water intersection identify a research lead.
- Other cells remain insufficient evidence; they are not declared unsuitable.
These thresholds are research display heuristics, not validated investment criteria. Every cell still lacks verified power capacity, land rights, flood, terrain, fiber and commercial evidence. Small water overlaps can round to zero; the explicit intersection flag preserves them.
The 10 km and 2 km grids summarize different areas. Use the finer view and exact parcel check to investigate a state cell; do not compare cell-level percentages across resolutions as if they describe equivalent sites.
National 0.5° cells contain directory records only. They are not clipped to India’s border and do not imply nationwide analytical coverage.
Read screening features and metadata →Evidence tiers describe how a claim was obtained.
| Tier | Meaning |
|---|---|
| A | A value retrieved from a versioned machine API. |
| B | A parsed published file, dataset or structured artifact. |
| C | A sourced statement interpreted from a web page or document. |
| D | A model, comparison or explicit assumption. |
A tier does not independently prove that a claim is current or correct. Review source authority, scope, date and contradictions as well. Existing directory snapshots retain their recorded tier; ingestion through an API does not independently verify a community-maintained record.
Exact boundary checks and comparison
The map can intersect your actual Polygon or MultiPolygon with mapped water and industrial land, preserving holes. It measures spherical area, coverage within the infrastructure source boundary, facility points inside the boundary, and distance from mapped substation reference points to the parcel. Saved parcels can be compared using the same snapshot and exported with geometry and source links.
These checks are source-limited. Missing categories stay unknown, overlapping source polygons are unioned, and partial geographic coverage can understate intersections. A zero water intersection is not proof of flood safety, and mapped industrial land is not proof of buildability or availability.
Before a site becomes a shortlist decision
- Resolve the parcel. Establish the actual boundary, location and responsible authorities.
- Verify feasibility. Obtain evidence for power delivery, land rights, access, water, hazards and permission pathways.
- Match the workload. Record target IT load, redundancy, cooling assumptions and opening date.
- Test uncertainty. Identify evidence gaps and facts that could reverse the comparison.
- Keep an audit trail. Preserve source dates, assumptions, contradictory claims and later corrections.
The browser's saved boundaries are drafts kept on that device. Drawing a polygon does not certify a cadastral boundary or create a completed analysis. Published CLI analysis runs are separate from map screening features.
Correct the evidence, then update the conclusion.
When a record is wrong or outdated, report its stable ID, supporting source and the field that needs correction. Please avoid submitting confidential land documents or personal contact details to public issues.
Report a data correction →