Preliminary site screening · 37 counties · 10 states
GridCite screens large parcels for solar, storage, industrial and data-centre siting. It ranks what is worth pursuing, says what could kill each site, and — the part that costs real money to get wrong — keeps nobody measured this firmly apart from this is fine.
Free to search and screen every county we cover. The county count on its own would overstate what most of them can answer: outside California no utility publishes circuit-level interconnection data, so most of the footprint answers a strictly smaller question. Coverage & data gaps says which, county by county.
How the work goes
Four questions in order, and the product is organised around them rather than around the datasets underneath. The datasets are the means; the decision is the thing you are buying.
Find sites worth pursuing.
Rank every parcel in a county against your project's own requirements — acreage, slope, grid distance, zoning, environment — rather than panning a map and guessing. The answer is a ranked list with the reasons attached, not a heat map you have to interpret.
Kill bad ones early.
A site that fails a hard requirement is out before anybody is briefed on it. What the screen refuses to do is fail a site on missing data: absence holds a parcel for review, because a wrongly rejected site is never looked at again and nobody ever learns it was good.
See what nobody has checked.
Every finding carries how well established it is, in its own channel — separate from the finding itself. Where two sources disagree, both readings are kept and neither is used to pass or fail a rule. Where nothing was measured, the page says nothing was measured.
Know what has to happen next.
A development memo per site: what could kill it, what is still unknown, which questions to take to the county and the utility, and what has changed since the last time you looked at it.
Measured, with the date on it
Every figure below was measured in a named county on a named day, and each one is recorded in this product's own working files. None is an estimate or an industry average — and where a figure comes from a sample rather than a whole county, the card says which sample, because a number that hides its own population is the kind of number this product exists to replace.
366of 445
A hybrid battery sits inside the solar project and shares its gen-tie, so the standalone storage rule is the wrong test for it. Run the standalone test anyway and four out of five qualifying sites disappear for a reason that does not apply to them.
Kern County, California — the 15,872-parcel valley-floor window we hold · measured 2026-08-10 · PR #491, re-derived independently
0substations
Neither county holds a substation our grid layer can measure to — re-derived against the live transmission service, with two other counties as controls. Distance to grid there is unknown: not far, not zero, and never a blank cell that reads as a clean site. No parcel total is quoted alongside it: two of our own build records disagree about Sterling's by eight, and the parcel extract in this repository holds only the Kern window, so there is nothing here to settle it against. A number we cannot check ourselves does not belong on this page.
Glasscock and Sterling Counties, Texas · measured 2026-08-10 · docs/findings/no-substation-counties.md
23.8%
22.4% of those at 30 acres and above have none either — and that is the finding. The rate barely eases with size, running from 26.5% at 10-30 acres to 17.5% above 200, so this is not a small-lot problem that utility-scale land escapes. Measured over all 15,872 parcels against the live TIGER road service. It is reported as a question to resolve and never as proof that land is landlocked, because TIGER carries no easements.
Kern County, California — all 15,872 parcels we hold · measured 2026-08-10 · supersedes the sampled figure in PR #499
133of 133
Not a threshold artifact: Midland records no earlier event at any magnitude at all, and neither do Martin or Loving. Kern County, California, over the same catalogue, is 186 of 1,164 — the opposite ratio. Counts and dates are what is reported; whether any individual event was induced is a determination this product does not make.
Midland County, Texas · Kern County, California — county bounding box, not the padded window the map layer draws · measured 2026-08-10 · USGS ComCat, retrieved 2026-08-10 · scripts/lib/seismic.mjs
2,302of 27,709
One tract carries 72. Every one of them has to sign, which makes an otherwise identical site a materially worse acquisition — and a product holding one owner field per parcel would have shown you a single name and called it the owner. The fleet's four other Texas counties publish no interest column at all, so their owner counts are unknown rather than one, and they are not in this total.
The eleven Texas counties whose rolls publish an ownership interest column · measured 2026-08-10 · docs/findings/fractional-ownership.md
The same discipline, on ground nobody mapped
California's Geological Survey publishes mapped liquefaction hazard zones. Across the bounding boxes of all eight California counties GridCite covers, those layers map zones in two quadrangles, in one county — Yolo — and no earthquake-induced landslide zone anywhere in the eight.
Wire that layer into a map on its own and it prints a clean reading over seven counties. A developer sees no hazard zone and moves on.
The reason is in the Survey's own companion layer — the one that maps the areas nobody has evaluated for liquefaction or seismic landslide. Over the same eight bounding boxes it returns 681 unevaluated quadrangles, between 45 and 192 per county.
The state has largely not looked. That is a different answer from “no hazard”, and it is the answer that should reach the reader.
Measured against the live California Geological Survey services over each county's own bounding box on 2026-08-10, and independently re-derived: Kern 88, Fresno 192, Tulare 108, Kings 48, Colusa 45, Yuba 49, Yolo 52, Merced 99 unevaluated quadrangles. Counted by quadrangle rather than by polygon on purpose — the mapped zones are three polygons of 146, 0.09 and 0.045 km², so a polygon count would make 146 km² of mapped ground sound like nothing. Recorded in docs/legal/DATA_RIGHTS.md and PR #496.
Zone layer alone
Liquefaction zone: none
True to the layer. Read as a clean site. Wrong.
GridCite
Not evaluated by the state
No hazard mapping has been performed over this quadrangle. That is a question for a geotechnical scope, not a clean screen.
No blended score
A site can be excellent land, ordinary permitting and a complete unknown on grid — and any single grade averages that third fact away. GridCite reports each part separately, names the part that is unresolved, and gives the reason next to it.
Illustration of the shape of a verdict — not a reading of a real parcel
The category names and the verdict words above are the product's own vocabulary. What you will not find anywhere in GridCite is a time to notice-to-proceed or a dollar value for a site: both are predictions, this product sources everything it prints, and neither can be sourced.
Two ways in
Developers, originators and site-selection teams for solar, storage, industrial and data-centre projects. Open a county, set the project's requirements, and work the ranked list.
Free to search and screen. What you get · Coverage & data gaps
If you have acreage and want to know what energy or industrial uses it might qualify for, that is a different conversation and it has its own front door. Tell us about the parcel and we will look.
Runs as GridSpan — the same screening, written for landowners rather than developers.