Methodology
A score you can take apart.
The sources are public. The work is in making 26 of them agree with each other: one schema for 2.1M+ water rights filed under 17 different state systems, one hex grid every layer is measured against, and a score whose every point can be traced to a distance, an overlap or a count. This page publishes the weights, thresholds and caps the scoring engine runs on, and says plainly what the score does not measure.
1. Normalization
Water rights arrive from each state engineer through the Western States Water Council's WaDE data exchange. Every state files differently — flow in CFS, gallons per minute or acre-feet, priority dates in different formats, beneficial uses under different vocabularies. Each record is mapped to a single schema: priority date, flow normalized to cubic feet per second, beneficial use, source type, legal status, owner of record and the WaDE identifier that ties it back to the filing.
Owner names are indexed for related-rights lookup, so every right held by the same owner of record can be found from any one of them. Points of diversion and places of use are kept as separate mapped sites with their own county and hydrologic unit.
Every other layer keeps its federal or state steward and is loaded on that steward's cadence: WaDE (Western States Water Council) and state engineers, USGS, USBR and state water agencies, Colorado Division of Water Resources, USGS and state well registries, EPA and USGS, US Drought Monitor, NRCS, EPA, State water agencies, EIA, HIFLD, FCC, PeeringDB, USFWS, BLM. The full list, with refresh cadence and the plan that includes each layer, is on the data catalog.
2. The grid
Sites are scored on H3 resolution-6 hexagons, each roughly 36 km². Hexagons have uniform neighbor distances, so "nearby" means the same thing in every direction, and the same cell id is used by every layer — no re-projection between datasets.
A land mask removes open water before scoring. A lake is excluded, not scored zero, because an excluded cell and a zero-scoring cell mean different things when a list is sorted.
3. Site Score components
The composite runs 0–100. The five weights below total 100.
Land Access
30 ptsMeasures: Share of the cell under private ownership, from PAD-US Fee Managers. Measures acquirability, not parcel boundaries.
Scoring: Full points at 60% or more private land in the cell, zero at 5% or less, linear in between. Cells with no land-ownership data score zero here and are flagged NO_LAND_DATA — an unknown owner is not a willing seller.
Transmission Proximity
25 ptsMeasures: Straight-line distance to the nearest mapped transmission line and substation. Measures adjacency, NOT available capacity.
Scoring: Full points within 2 miles of a mapped transmission line, zero beyond 25 miles. Up to 4 bonus points for a substation of 230 kV or more within 10 miles, capped at the component weight.
Fiber Proximity
15 ptsMeasures: Distance to the nearest internet exchange facility. A proxy for connectivity — long-haul route data is not yet loaded.
Scoring: Full points within 10 miles of an internet exchange, zero beyond 100 miles. Up to 3 bonus points for 3 or more reported fiber providers, capped at the component weight.
Water Infrastructure
15 ptsMeasures: Count of mapped points of diversion and use within the search radius. Measures infrastructure density, NOT water available to a new user.
Scoring: Rises with the count of mapped points of diversion and use nearby and plateaus at 1,000. Above 3,000 the cell is flagged WATER_APPROPRIATION_RISK: in a prior-appropriation state, that density usually means a fully claimed basin, not spare water.
Environmental Constraints
15 ptsMeasures: Overlap with designated critical habitat, wetlands, and protected areas. A screening flag, not a jurisdictional determination.
Scoring: Starts at full points. Wetland overlap removes 35% and protected-area overlap removes 35%, additive and capped at 60% so correlated designations are not double-counted.
4. Hard constraints and flags
Some conditions are gates, not trade-offs. Rather than clamp every constrained cell to the same number, the engine scales the composite into a lower band, so a constrained region can still be ranked internally. Every score also carries flags naming what the data could not establish.
- CAPACITY_UNVERIFIED
- On every score until interconnection-queue evidence is loaded. Distance to a line is not a statement about what the utility can serve.
- EXCLUDED_RISK
- Inside designated critical habitat. The composite is scaled into 0–40 rather than clamped, so constrained cells stay rankable among themselves.
- PUBLIC_LAND_DOMINANT
- More than 80% public land. The composite is scaled into 0–45, so unacquirable ground never outranks a buildable private parcel.
- LAND_ACCESS_CONSTRAINED
- Less than 50% of the cell is private.
- TRIBAL_LAND_PRESENT / DOD_LAND_PRESENT
- Material tribal or Department of Defense land in the cell — effectively closed to acquisition, for different reasons.
- WATER_APPROPRIATION_RISK
- Diversion density high enough that senior appropriation is near certain.
- WETLAND_OVERLAP / PROTECTED_AREA_OVERLAP
- The cell clips a mapped NWI wetland or a PAD-US protected area. A screening flag, not a jurisdictional determination.
- NO_LAND_DATA / NO_POWER_DATA / NO_FIBER_DATA / NO_WATER_DATA
- A component had no evidence. It scores zero, never full marks, and the score's confidence drops to low.
5. Confidence
Confidence comes from the evidence behind each component, not from the score itself. A high score built on a missing dimension is not a medium-quality high score — any component with no evidence caps the whole result at low confidence. Every stored score is tagged with its model version, and a ranking that would mix versions says so instead of silently blending them.
6. What the score does not tell you
- How much power a utility can deliver. That number comes from an interconnection study. Distance and voltage class narrow which substations are worth studying.
- Whether water is available to a new user. Diversion density measures infrastructure; availability depends on basin closure and the seniority of the specific rights you can acquire.
- Parcel-level ownership outside Utah, Nevada and Arizona. Elsewhere the land component scores zero and the score is flagged NO_LAND_DATA.
- Private or fee severed mineral ownership. Mineral layers cover federal claims, federal leases and USGS-recorded occurrences; private mineral title lives in county records.
- Anything a title search, a jurisdictional wetland delineation or legal counsel is needed for. The score is a screen that decides where to spend diligence.