California groundwater · Methods
How the figures in the California groundwater series were built. Last checked 1 October 2026.
The counts come from the Department of Water Resources' Household Water Supply Shortage Reporting System, published on the state open data portal as Dry Well Reporting (GIS Feature Service). Households report their own failures voluntarily, and each report carries a status (outage, interim solution, resolved or vulnerable) that is current as of the export, so every count is a floor.
Derived I analysed the public export downloaded in May 2026: 6,139 household water shortage reports, of which 5,891 are dry wells (those whose shortage type mentions a dry well or groundwater). By current status, 2,954 are outages (50%), 1,047 have an interim solution (18%), 1,880 are resolved (32%) and 10 are listed as vulnerable.
Derived The time charts place each report in the month its problem began. 1,573 dry-well reports (27%) have no start date and are left out of those charts, so the charts understate the total. The largest month on record is July 2021, with 283 reports.
Documented The "at least 2,600" figure for the 2012-2016 drought is from the Public Policy Institute of California: "at least 2,600 well-dependent households experienced water shortages" (PPIC). PPIC does not say how it was counted. The reporting system is a separate, smaller tally: it holds 2,332 dry-well reports created between 2012 and 2016.
Derived 1,193 of those 2,332 were created in August 2016, 1,154 of them on a single day (12 August), and 1,158 are in Tulare County. They look like a bulk entry of existing households: 94% have no start date and 94% are listed as resolved or interim. Because the time charts need a start date, these reports are counted in the totals but not plotted.
Derived By current status, the 2,332 reports created in 2012-2016 are 39% outage (916), 39% resolved (902) and 22% interim solution (511), with 3 vulnerable. The boxed years on the chart show only the 978 reports whose start date falls in 2012-2016, and 80% of those are outages (778, against 115 resolved and 83 interim). The chart looks outage-heavy partly because the 1,428 reports created in that period with no start date cannot be plotted, and 85% of them are resolved or interim.
| Stage | Meaning | Who pays | Unit |
|---|---|---|---|
| Unresolved | No state help yet, so the household covers its own water. | The household | Per month |
| Interim | Temporary water supplied by the state or its partners while a household waits for a permanent fix: bottled water, a tank filled by truck, or both. | The state | Per month |
| Resolved | A new well or a connection to a public water system. | Usually the state | One-time, per well |
The units differ between stages, so the three numbers are not additive.
No single source reports $600. It is the sum of documented costs for a household that has lost its well entirely:
Households typically pay nothing at this stage. DWR's Water Tank Program provides a 2,500-gallon tank, delivery and installation, a small pump and plumbing, and hauled water. DWR publishes no per-household price for the tank.
Derived Self-Help Enterprises' regional table (FY 2026-27 Fund Expenditure Plan, Appendix N, Table N-1) lists $1,740,312 across 2,681 bottled-water households, about $649 per household. $649 / 12 is about $54 a month. The table gives no period, so treat this as approximate.
Documented Appendix N (page 83) states an average annual cost of $13,000 per household for bottled or hauled water. $13,000 / 12 is about $1,083 a month.
Derived Table N-1 lists $15,027,287 for 529 hauled-water households, about $28,400 each, which does not reconcile with the $13,000 average, and the table does not state periods. The series uses $13,000 as the State's stated average.
Connection to a public water system (consolidation) is omitted from the headline number because it is priced per connection on whole-system projects, not per household bill:
The series does not estimate a total or annual public cost of groundwater depletion. Costs such as subsidence damage to infrastructure, property values and pumping energy are not quantified here.
The opening slides show how much Central Valley groundwater storage has been lost since 1960, and how long what is left would last if it kept falling at the recent rate. The answer is about 200 years, give or take a few decades depending on how the recent rate is measured.
Documented Storage from the early 1960s to 2014 is modelled storage change from the USGS Central Valley Hydrologic Model, taken from the dataset published with Liu et al. (2022). Later years, to 2023, come from DWR's GRACE series, built from NASA's GRACE and GRACE-FO satellites. The series runs monthly from October 2006 to December 2023, with a gap from May 2017 to May 2018 between the two missions. The two parts of the curve come from different methods, so the join between them is approximate.
Satellites measure changes in total water storage. DWR's series is labelled a groundwater storage anomaly, and I have not checked how DWR separates groundwater from snow, soil moisture and surface water.
Derived DWR's column header for the satellite series reads TAF, but the values are consistent with acre-feet, so I converted them as acre-feet.
It is a straight-line extrapolation of total stored water, not a forecast. The rate will not stay fixed: droughts have accelerated the loss, while the Sustainable Groundwater Management Act requires overdrafted basins to be brought into balance, which could slow it.
Derived The result is also sensitive to the window used for the trend, which is a choice. I used the whole satellite record because it has the most data and the best fit. The same 788 km³ gives:
So the range is roughly 200 to 245 years across the windows that fit reasonably. It is not a precise year.
Illustration The figure counts storage, not usable water. As the water table falls, what remains is deeper, costlier to pump and can be of poorer quality, so the water that is practical to use would run short sooner. The "less usable" slide draws that idea and carries no data.
Documented The cities-versus-agriculture chart redraws Figure 4B of Faunt et al. (2024), which shows urban and agricultural groundwater use simulated by the Central Valley Hydrologic Model, version 2 (CVHM2). Agriculture is roughly nine-tenths of the total across 1960 to 2020.
Derived The 2002 to 2022 bars are groundwater as a share of total water supply for each water year, computed from DWR's Water Plan Water Balance Data on the state open data portal. The highest year is about 70%, during the 2012-2016 drought.
Derived The shares are of irrigation water applied to crops in 2020, from DWR's Statewide Agricultural Water Use Data 2016-2020, which reports applied water in acre-feet by crop category. Almonds and pistachios together took 32.2%, other perennial tree crops 24.6% and all other crops 43.2%.
The export marking is indicative. Much of the almond and pistachio crop is sold abroad, but the slide does not measure an export share.
Documented The $61 billion total is California's agricultural production value, reported as $61.2 billion for 2024 (Valley Ag Voice). It is a statewide figure, not a Central Valley one.
"Untaxed" is shorthand. The US Constitution bars taxes on goods in the course of export (Article I, Sections 9 and 10), so exports cannot be taxed as exports. That does not mean farm income or land goes untaxed. "Mostly untaxed" for the rest of the sector is my characterisation and is not quantified here.