--- title: the ten hottest days in prineville canonical_url: https://ensurance.app/guide/the-ten-hottest-days-in-prineville markdown_url: https://ensurance.app/guide/the-ten-hottest-days-in-prineville.md subtitle: "the wells were built for the peak, not for an average year" category: how-to --- # the ten hottest days in prineville *the wells were built for the peak, not for an average year* The last week of August, the rimrock above Prineville holds the day's heat long after sunset, and the sprinklers come on before six so the water reaches the roots instead of the air. On an afternoon like that, the city's water system can move close to four million gallons. In January it moves less than one. Same town, same pipes, four times the water. That gap — not the annual total — is what Prineville's water system is actually built around. It is also why the city, Apple, and Meta spent the better part of a decade moving winter water underground. If you run a small municipal system anywhere in the interior West, the shape is familiar. What makes Prineville worth reading closely is that the peak-day problem here has been described out loud, in public, by the people who operate the system — and that two of those public sentences keep getting mashed into one. ## the day the system is built for Prineville counted 10,736 residents at the 2020 census and an estimated 11,994 by July 2025. That is a small system by any measure: a few thousand service connections on the Crooked River, in a basin ringed by juniper and rimrock, with the Ochocos holding the snow that eventually becomes everything. LimnoTech's August 2025 report for Meta puts the city's peak-day demand at 1 to 4 million gallons a day, and names that range as the thing the aquifer project was built to cover. The city's own 2018 announcement said it in plainer words: under a million gallons a day in winter, over four in summer. An annual average hides all of that. The planning question for a town this size is never "how much water did we use last year." It is "what happens on the tenth-hottest afternoon of the year, at the far end of the line, with a crew pulling from a hydrant and every lawn in town running at once." That is the design case. Ten days, more or less. Everything else is slack. ## two sentences about ten percent In March 2026, the City of Prineville's public-works lead told KTVZ that the three data centers in town use about 10 percent of the water used in Prineville, and set that against the lumber-mill economy the campuses replaced. The same piece notes the companies funded nearly the entire aquifer system and pay ordinary industrial rates and system development charges. The third operator was not named. Nearly two years earlier, in May 2024, the Bend Bulletin had already attached that same share to a narrower thing: peak days — about the ten hottest days of the year. Hold both sentences. Do not stack them. A share of annual municipal use and a share of the hottest-day load are two different facts about a system, and only one of them tells you anything about reliability. The version that travels fastest online — *data centers use a tenth of the town's water, every day* — is not what either source says. If you are a council member, a reporter, or a utility planner working from these numbers, the distinction is the work: annual share is a budget question, peak-day share is a capacity question, and capacity is what fails first. ## the winter water under town The fix Prineville built is unglamorous and good. In winter, when the town barely needs it, the city pumps shallow groundwater near the Crooked, treats it, sends it uphill, and injects it into the rock. In summer, it pulls that water back out. If you want the method itself, [managed aquifer recharge](/guide/managed-aquifer-recharge?from=guide) is its own guide; the point here is the calendar, not the technique. Water that falls and flows in the cold months is held for the ten days it is worth the most. The 2024 reporting describes 5,100 acre-feet of stored Prineville Reservoir water backing that winter pumping, so the river is not simply drawn down while the wells run. The cost share is public in pieces. Apple's grant for the first injection well was $8.7 million, per the city's own release. The City of Prineville holds the permit, operates the system, and does the monitoring and maintenance. Phase I has been injecting since 2021, and LimnoTech attributes 30.3 million gallons a year of increased recharge to Meta's share of it, through 2030. Add the 1.5 million gallons a year credited to a restored 15-acre wet meadow on Auger Creek in the Ochoco National Forest, and Meta's counted benefit in the Columbia chapter of its 2024 report is 31.8 million gallons. Those gallons are not a gift to the river in general. They are winter water, in the basalt, under a town that needs it in August. ## what the peak day reaches On a system this size, the hot week is where everyone meets. The sprinkler at the end of your street, the hose at the fairgrounds, the hydrant a crew pulls from when a fire starts east of town, and the cooling load that peaks on the same afternoons — all of it lands on the same pumps during the same ten days. The campuses know it. An Alliance for Water Stewardship note on Apple's Prineville campus says the global water-stress datasets scored Prineville as *low* baseline stress, and that it was local climate and stakeholder input that led the company to treat summer peak availability as the actual constraint. A global risk screen would have told Apple this town was fine. The ten hottest days told it otherwise. That is the quiet lesson for every other small system with a large customer: the risk that reaches you is not the one in the dataset. It is the one on the calendar. ## the other way the hot week goes wrong Peak-day supply is not the only thing that can fail in Crook County in August. About six miles of McKay Creek — the Crooked River tributary at Prineville, not McKay Dam over on the Umatilla — go dry each season under irrigation withdrawals. The proposed exchange would leave stored Prineville Reservoir water with the irrigators and keep the creek's rights instream, with steelhead named as the reason in the report. The reintroduction above Pelton Round Butte is co-owned by Portland General Electric and the Confederated Tribes of Warm Springs, downstream sovereigns, not buyers here. LimnoTech lists it at about 35 million gallons a year, anticipated, and it is not in the 2024 count. Two failures, one watershed, one calendar. A town can cover its peak day and still have a dewatered tributary six miles long. [That creek has its own post](/guide/mckay-creek-still-goes-dry?from=guide) — it is listed in Meta's appendix as anticipated, designed, and not yet wet. ## the number that is still anticipated Phase II of the aquifer system — another injection and recovery well, a horizontal source well, metering, an oxygen-system upgrade — is listed at an anticipated 150 million gallons a year. As of the August 2025 report, it still sits in the appendix for projects not yet generating benefits. Say it that way. A city planning for a peak day can only spend water that is actually in the ground, and the difference between *counted* and *anticipated* is the difference between a plan and a press release. Prineville's own documents keep the two apart. Anyone writing about this town should too. :::johnson **the peak day is the unit, not the year.** Average annual use will never tell a council, a ratepayer, or a campus whether the system holds on the tenth-hottest afternoon. Winter water in storage will. [see how a city funds its source watershed →](/solutions/governments?from=guide&topic=crooked-river) ::: ## three things worth copying If you operate or oversee a small system with a large industrial customer, Prineville's record gives you three moves that cost nothing to adopt. 1. **State demand as a peak-day range, not an annual average.** "1 to 4 million gallons a day" is a sentence a council, a reporter, and a customer can all act on. "X gallons per year" is not. 2. **Label every gallon counted or anticipated.** 30.3 million a year is operating. 150 million is anticipated. 35 million on the creek is anticipated. Three different kinds of certainty, three different planning weights. 3. **Write the shared peak as a shared obligation, not a courtesy.** The customers who load your hottest day are the customers with the clearest reason to fund winter storage, meadow restoration, and instream flow — permanently, not project by project. That third move is where *ensurance* fits. A certificate on a river agent is how several parties who depend on the same water hold a single, reportable position in its condition, with proceeds funding the work that keeps the condition up. The Crooked's parent is [rivers-lakes.ensurance](/rivers-lakes.ensurance?from=guide), and the recharge work joins a live theme at [aquifer-recharge.syndicate](/aquifer-recharge.syndicate?from=guide). For a utility, the solution with a name is [source watershed investment](/solutions/utilities?from=guide&topic=crooked-river) — a watershed partnership you can report. For a city, it is [municipal watershed protection](/solutions/governments?from=guide&topic=crooked-river), and the unit a council can actually read is per peak summer day that banked winter water covers. ## frequently asked questions ### do data centers use 10 percent of prineville's water? In March 2026, the city's public-works lead said the three data centers in town use about 10 percent of the water used in Prineville. In May 2024, the Bend Bulletin tied that share specifically to peak days — about the ten hottest. Both statements are on the record, and they are not interchangeable. Cite the one that matches your question, and label the year. ### what does aquifer storage actually do for a hot day? It shifts the withdrawal off the summer peak. Water is pumped and injected in winter, when Prineville uses less than a million gallons a day, and recovered in summer, when the system can run above four. The aquifer is the reservoir; the calendar is the product. ### is the 150 million gallons a year real? It is anticipated, not current. Phase II was still in the "not yet generating benefits" appendix of the August 2025 report. The operating figure attributed to Meta's cost share on Phase I is 30.3 million gallons a year, counted since 2021. ### what happens to mckay creek on those same days? About six miles stay dry under irrigation withdrawals. The exchange that would change that — stored reservoir water to the irrigators, creek rights left instream for steelhead — is anticipated at roughly 35 million gallons a year and is not yet counted. ## who already pays The useful thing about Prineville is that the ledger is not hypothetical. Apple paid for the first well. The city operates the system and holds the permit. Meta's cost share is counted in gallons every year and published. The Ochoco National Forest holds the meadow. Ochoco Irrigation District — about 898 irrigators across roughly 20,062 acres — is the party on the creek exchange still ahead. That is a functioning coalition with one gap: every contribution so far has been a project, and projects end. The next post in this series is about that ledger, and about the second layer — a hold on the same river that other people can carry beside it. [talk to someone who can help →](/contact?from=guide&topic=crooked-river) ## sources [Volumetric Water Benefits: 2024 Report, LimnoTech for Meta, August 2025](https://sustainability.atmeta.com/wp-content/uploads/2025/08/REVISED_FINAL_Meta_2024_Volumetric_Water_Benefit_Report_08.21.2025.pdf) — Phase I recharge attributed at 30.3 million gallons a year since 2021, Ingram Meadow at 1.5 million, 31.8 million combined for 2024, Phase II at an anticipated 150 million, McKay Creek at an anticipated 35 million, and the 1 to 4 million gallons a day peak the system was built to cover. [City of Prineville release on the Apple grant](https://www.meadowlakesgc.com/sites/default/files/fileattachments/city_administration/page/12381/press_release_-_prineville_apple_002.pdf) — the $8.7 million grant for the first injection well, and the city's winter-versus-summer demand figures. [KTVZ, 5 March 2026](https://ktvz.com/news/problem-solvers/2026/03/05/problem-solvers-tech-investment-in-state-of-the-art-water-system-could-reshape-prinevilles-economy/) — the city public-works lead on the three data centers using about 10 percent of the water used in Prineville, and on rates and system development charges. [Bend Bulletin, 9 May 2024](https://bendbulletin.com/2024/05/09/prineville-doubles-down-on-water-conservation-with-tech-giant-backed-project/) — the peak-day reading of that share, about the ten hottest days, and the 5,100 acre-feet of stored reservoir water backing winter pumping. [Apple, Water Stewardship in Data Centres (Alliance for Water Stewardship, 2025)](https://a4ws.org/wp-content/uploads/2025/01/AWS_Water-Stewardship-in-Data-Centres_2025.pdf) — low global baseline stress scoring for Prineville, and summer peak availability as the constraint local input identified. [U.S. Census QuickFacts, Prineville city, Oregon](https://www.census.gov/quickfacts/prinevillecityoregon) — 10,736 at the 2020 census; 11,994 estimated 1 July 2025. ## the series 1. [the crooked already has a meadow and a wellfield](/guide/what-the-crooked-river-depends-on?from=guide) — what the river depends on 2. [who depends on the crooked river](/guide/who-depends-on-the-crooked-river?from=guide) — a city, two campuses, and 898 irrigators 3. [six miles of mckay creek still go dry](/guide/mckay-creek-still-goes-dry?from=guide) — the dewatered tributary and the exchange that would refill it 4. **the ten hottest days in prineville** — you are here 5. [apple and the city already paid for the well](/guide/who-already-pays-for-the-crooked-river?from=guide) — who already pays, and the second layer