Managed Aquifer Recharge
WASR 4500/6500
WASR 4500/6500 · Special Topic · Interactive Companion

Managed Aquifer Recharge

The biggest reservoir on Earth has no dam, no spillway, and no evaporation. It is the rock and sand beneath your feet, and for most of history we only ever took water out of it. Managed aquifer recharge (MAR) is the deliberate act of putting water back — river floods, storm runoff, treated wastewater, even surplus drinking water — so it can be recovered in a dry season, hold back the sea, prop up a sinking coast, or feed a stream through summer. This companion builds the idea from a farmer’s check dam to a city’s injection wells: why store water underground, how to get it there, what happens to it once it is down, and whether it pays and lasts. Click a part of the landscape below (or a chapter card) to explore, experiment, and test yourself.

Built from two references in the course folder: the ITRC Managed Aquifer Recharge Guidance (December 2023) and the UNESCO/IAH showcase Managing Aquifer Recharge: A Showcase for Resilience and Sustainability (Zheng, Ross, Villholth & Dillon, eds., 2021; 28 case studies from 21 countries). Evaristo Critical Zone Hydrology Lab, University of Georgia. Scores are self-assessment only — nothing is recorded, and progress resets if you reload.

Learning objectives — what you should be able to do
  1. Define managed aquifer recharge, distinguish it from disposal and from incidental recharge, and name the four key elements of a MAR project (intended use, source water, receiving aquifer, recharge technology).
  2. Match recharge technologies to hydrogeologic settings — unconfined vs. confined aquifers, low-permeability layers, land and source-water constraints — and explain ASR, ASTR, riverbank filtration, and in-channel structures.
  3. Quantify the basics: storage losses, groundwater mounding, mixing and recovery efficiency, pathogen log-removal credits, and clogging-driven decline in infiltration.
  4. Anticipate the water-quality risks of putting one water into another — arsenic mobilization, clay dispersion, salts that soil cannot remove — and the monitoring that catches them.
  5. Evaluate a scheme with levelised cost, benefit–cost ratio, energy intensity, and the nine environmental and social sustainability indicators, and explain why a high BCR and a low sustainability score can describe the same project.
Chapter 1 · What MAR is, and why anyone bothers

Why Put Water Underground?

Rain falls when it wants to and crops drink when they must, and the gap between the two is the oldest problem in water. Dams bridge it above ground, in the sun. Aquifers bridge it below ground, in the dark — and the dark has advantages.

1 · The definition, and the four questions

The MAR process model

The definition is short and every word carries weight: MAR is the purposeful recharge of water to aquifers for subsequent recovery or for environmental benefit (Dillon et al. 2009, quoted by both references). Purposeful rules out the accidental — the leaking canal, the over-watered lawn. Recovery or environmental benefit rules out disposal: as the UNESCO book puts it, MAR “is not a method for waste disposal.” Groundwater recharge is a natural phenomenon; MAR either enhances or restores it where it already happens (a floodplain), or introduces it where it does not (a desert basin). The ITRC guidance organizes every project around four elements that must fit together, arranged in a wheel. Click each element to see the questions it asks.

The order matters. A project starts by naming its objective, then finds a source of water, then asks whether the subsurface can take it and give it back — and only then chooses a technology. Whether a MAR project succeeds or fails, the ITRC warns, “is largely dependent on the thorough understanding of project-specific factors” (ITRC p. 6). Water rights sit outside the wheel but can stop it turning.
2 · Seven jobs

What MAR is used for

Historically the reason was simple: put water back in overdrawn basins. But an aquifer that is being refilled can do several jobs at once. The ITRC lists seven intended uses — water supply resilience, improving groundwater quality, mitigating saltwater intrusion, using stormwater, using floodwater, reducing land subsidence, and protecting riparian ecosystems or minimum streamflow — and illustrates each with a real project. Read each project below and tag the job it is doing; several could plausibly claim two, so pick the one the reference leads with.

3 · Above ground or below?

Two ways to store a year of water

Here is the case for the dark. Windhoek, Namibia, gets 360 mm of rain a year and loses 2,170 mm to evaporation; about half of the water in its surface reservoirs evaporates, while losses from its underground store are estimated at under 3 % over ten years (UNESCO Case 13). San Antonio’s ASR project makes the same point in one line: water stored in an aquifer “is not subject to evapotranspiration losses” (ITRC p. 97). Aquifers also take no farmland, drown no valleys, and grow no algae. Their drawbacks are just as real — you cannot see the water, you cannot always get all of it back, and it can pick up chemistry from the rock on the way. Store a year of water both ways and watch the sun take its share.

A simplification to keep in view: the reservoir here is a tank with vertical sides and no inflow, so the yearly loss is simply evaporation depth divided by mean depth. Real reservoirs shrink in area as they drop (which slows the loss) and also leak (which speeds it). The aquifer’s loss rate is the Windhoek estimate, and it hides a different cost: some stored water mixes with what was already there and may never be recovered. Chapter 2 takes that up.
4 · How big is this, really?

Guess the number

Before the numbers, a guess. Groundwater is the quiet giant of water supply, and MAR is still a small, fast-growing correction to how we treat it. Slide to your estimate, then reveal. The answers come from the UNESCO executive summary and synthesis chapters and the ITRC state survey.

5 · Drawing the line

Is it MAR?

Every definition has edges. The ITRC explicitly excludes engineered injection that is not meant to recharge an aquifer for use: disposal wells (UIC Class I and II), open-loop geothermal systems, solution mining (Class III), septic drainfields, and CO2 sequestration (Class VI). The UNESCO book excludes “incidental causes of increased recharge, such as land clearing, soil conservation, terracing and contour bunding, where the main purpose is to enhance agricultural production, erosion protection or flood mitigation.” And yet the line moves: Italy’s riverbank-filtration wells are “by far the most common MAR scheme in Italy, even though not formally recognized as such” (Case 19), and Nebraska turned the winter seepage from leaky irrigation canals into a permitted MAR project by choosing to divert the water for that purpose (Case 17). Sort the twelve situations into MAR, not MAR, or incidental recharge.

Toolkit

Units you will meet

The American guidance speaks in acre-feet, gallons per minute, and cubic feet per second; the international showcase speaks in cubic metres and megalitres. Both are in this companion, in the units of their source. Type any quantity to see it every other way — and in swimming pools and households, which is how most people actually picture water.

Chapter 1 quiz

Test yourself: why put water underground?

Scored, with explanations after grading. Retake as many times as you like — questions reshuffle.

Chapter 2 · Recharge technologies

Ways to Recharge

Water gets into an aquifer one of two ways: it soaks down through the soil, or it is pushed in through a pipe. Everything in this chapter is a variation on those two verbs — and the choice between them is made by the ground, not by preference.

1 · The catalogue

Seven techniques, one cross-section

The ITRC groups technologies into surface recharge (infiltration basins; retention and diversion structures) and subsurface recharge (injection wells, dry wells, infiltration galleries). The UNESCO book uses four families — in-channel modification, bank filtration, water spreading, and recharge wells. Surface methods are generally preferred where they are possible: clogging is easier to control and the vadose zone treats the water on the way down. Subsurface methods win where there is no room, where a clay layer blocks percolation, or where the target aquifer is confined and cannot be reached from the surface at all. Choose a technique to see how it works, where it fits, and what it costs to keep running.

The one rule that decides most projects: “MAR technologies that involve the percolation of source water through the vadose zone, such as infiltration basins, can only be used with unconfined aquifers.” A confined aquifer is sealed above by a low-permeability unit, so “MAR projects that utilize confined aquifers must inject the source water directly into the aquifer” (ITRC p. 26). Dry wells and galleries are a middle path: gravity-fed, but able to reach past a shallow clay lens.
2 · Read the ground

Pick the method from the site

Six real sites, each drawn as a stratigraphic column with its constraints. For each, choose the recharge method the project actually used, then check. The lesson repeats six times because it is the whole lesson: the technique is chosen by geology, water quality, and land — a 6-metre clay blanket sends you to wells, a shallow water table under a city sends you to dry wells, a confined sandstone 150 m down sends you to ASR.

3 · The bubble and its seasons

Aquifer storage and recovery

An ASR well injects when water is cheap and plentiful (winter, wet years, even overnight) and pumps the same well when it is scarce (summer peaks, droughts, emergencies). Conceptually the stored water forms a “bubble” around the well that can be captured with an acceptable percentage recovery, which is why ASR is developed where groundwater gradients are flat (ITRC p. 59). If the aquifer is brackish, the edge of the bubble mixes with native water by advection and dispersion, and the fraction you can recover before the blend fails a standard is the recovery efficiency. Two tricks raise it: inject into a thin confined layer to limit mixing, and build a buffer zone of injected water that is never recovered (ITRC p. 35; Hilton Head’s buffer was 0.91 million m3 and cost about US$144,000). An ASTR system separates injection and recovery wells so the water travels — more uniform residence times, more predictable treatment. Run the cycle, then drag the chemistry.

Real recovery efficiencies: Batiaghata, Bangladesh, recovers about 31 % of what it injects into a brackish sand, and modelling suggests 50 % is the ceiling with a single central well — “100 % can never be expected as there is dispersive mixing” (UNESCO Case 1). Hilton Head, with its buffer zone, injects about 1 million m3 a year and recovers 0.9–1.1 (Case 18). Dinteloord uses partially penetrating wells — inject low, recover high — to fight the buoyancy of fresh water floating on brackish, and models more than 95 % (Case 28). Des Moines’ first cycle reached the drinking-water limit at 100 % recovery by volume with only 30 % of the pumped water being the injected water; each cycle improves (ITRC p. 135).
4 · What goes down must spread out

The groundwater mound

Pour water onto an unconfined aquifer faster than it can flow away and the water table rises beneath the basin into a mound. Its height depends on the recharge rate, the specific yield, and the hydraulic conductivity: low conductivity and low specific yield mean water piles up because it “is being added to the aquifer faster than it can flow away and only a small volume of pore space is available” (ITRC p. 27). Too much mounding floods basements, lifts structures, or intercepts the basin floor and stops infiltration altogether — which is why Delaware requires at least 2 ft between basin floor and mounded water table, and Nevada at least a 10-ft vadose zone (ITRC p. 51). The ITRC names the standard screening tool: Hantush’s (1967) analytical solution. This lab runs it. Set the basin and the aquifer, then watch the mound grow through a recharge season.

h2hi2 = (w ν t / 2K) · Σ S*(α, β),   ν = K / Sy,   S*(α,β) = ∫01 erf(α/√τ) erf(β/√τ) dτHantush (1967) for a rectangular basin of half-length a and half-width l: the four S* terms use (a ± x)/√(4νt) and (l ± y)/√(4νt); is the mean saturated thickness during recharge, found by iteration (the approach of the USGS spreadsheet of Carleton 2010, which the ITRC cites)
Two limits to sanity-check any mound: a huge basin over a tight aquifer behaves like a bucket — the rise approaches w t/Sy, every drop staying where it landed; a small basin over a gravel aquifer barely mounds at all because the water leaves sideways as fast as it arrives. Idaho’s basalt program, with conductivities up to 24,000 ft/day, reports that “mounding has not been an issue” (ITRC p. 108); India’s UTFI pond saw a peak mound of 0.8 m or less (UNESCO Case 15).
5 · The enemy

The clogging clock

“Clogging of well screens and aquifer pore spaces is one of the main challenges facing sustainable operation of MAR and has the potential to affect any MAR project regardless of the recharge technology” (ITRC p. 39). In a basin it forms a thin skin at the soil–water interface — suspended solids, microbes, algae, dust, salts — that has been found to cut hydraulic conductivity by as much as five orders of magnitude. The cure is drying and scraping: Dresden’s machine washes the top 5–10 cm of a basin in 6–7 hours and sustains 7–10 m/day of infiltration (UNESCO Case 8); Orange County’s basins range from 3 to 120 ft3/s, and its best performer takes recycled water with essentially no suspended solids (ITRC p. 51). In a well, the signs are an injection rate that falls at constant head or a head that rises at constant rate; the rule of thumb is to rehabilitate when capacity has dropped about 25 %, and to backflush every few weeks to months (ITRC pp. 39–40). Run a 180-day season with your source water and maintenance plan and see how much actually gets into the ground.

Cheapest clogging control is upstream: settle the solids before they reach the basin (retention ponds, riverbed filtration — Orange County’s buried pipes and dry wells under the riverbed took infiltration from 0.1–0.7 to 2–5 ft/day), keep iron below 0.3 mg/L to starve iron bacteria, and keep a disinfectant residual in a well during storage periods longer than about two weeks (ITRC pp. 39, 72). Dry wells are the cautionary tale: “very difficult, if not impossible, to clean out or rehabilitate once clogged with sediment” (ITRC p. 68), so pretreatment there is not optional.
Chapter 2 quiz

Test yourself: ways to recharge

Scored, with explanations after grading. Retake as many times as you like — questions reshuffle.

Chapter 3 · Water quality: treatment, reactions, mixing

The Aquifer as a Treatment Plant

Put one water into another and something happens. Usually it is good: sand filters, microbes eat, time kills pathogens. Sometimes it is bad: oxygen meets pyrite and arsenic walks out of the rock. MAR “has been shown to degrade or improve water quality” (ITRC p. 32) — the job is to know which, before the pump starts.

1 · Time as a disinfectant

Log removal and residence time

Pathogens are counted in logs: one log is a factor of ten, four logs is 99.99 % gone. Regulators budget them across barriers. For Hampton Roads’ SWIFT project — advanced-treated wastewater injected into Virginia’s Potomac aquifer — the target is 12 logs for viruses and 10 for Cryptosporidium and Giardia, and the plant alone does not get there: flocculation, biofiltration, and UV deliver 6 to 8, and the aquifer supplies the rest through soil aquifer treatment. California’s rules make the aquifer’s contribution explicit: 1 log of virus reduction credit for every month the water stays underground, up to 6, plus a minimum “response retention time” of two months so that a treatment failure can be caught before the water reaches a well, and a six-month travel time to the nearest drinking-water well for recycled water spread in basins (ITRC pp. 33, 51, 87). Riverbank filtration shows the same power without a plant: Haridwar’s Ganga wells remove ≥ 4 logs of pathogens and ≥ 2.5 logs of turbidity (UNESCO Case 20). Build the budget: switch barriers on and off and slide the residence time.

Barrier, not miracle. Haridwar’s bank filtrate still carried up to 1,600 MPN/100 mL of total coliforms against an Indian standard of zero, “thereby necessitating post-treatment by disinfection” (Case 20). Log removal is a fraction, and a fraction of a very large number can still be a large number.
2 · The filter and its blind spots

What the ground removes — and what it doesn’t

Soil aquifer treatment is filtration, adsorption, and biodegradation as water percolates through the vadose zone. It is astonishingly good at some things and helpless against others. At San Luis Río Colorado, Mexico, the treatment plant cuts fecal coliforms from 2,400,000 to 261 per 100 mL, and the lagoons and the 25 m of sand beyond them take those 261 to zero — while chloride rose from 511 to 541 mg/L, above the 250 mg/L recharge limit, because of evaporation in the plant and salts in the strata the water crosses (UNESCO Case 3, Table 1). The ITRC’s own summary of a basin’s disadvantages says it plainly: “Soil aquifer treatment does not remove all organic contaminants” and “does not remove salts” (p. 50). And an aquifer can add things: arsenic from its minerals, nitrate flushed out of a dry vadose zone. Sort the twelve observations.

3 · Chemistry at the recharge front

The arsenic switch and other incompatibilities

Where injected water meets native groundwater and the rock, four kinds of reaction wait: redox, mineral dissolution and precipitation, ion exchange, and sorption/desorption (ITRC Fig. 3-8). The famous one is Florida’s. Oxygen-rich surface water injected into the reducing Floridan aquifer oxidized traces of arsenopyrite in the limestone and released arsenic into the stored water; when the drinking-water standard fell from 50 to 10 ppb in 2006, a whole class of ASR systems was suddenly out of compliance without changing anything they did. The fix at DeLand cost about US$42,000: dose sodium hydrosulfide at 3–6 ppm to strip the oxygen before injection, and recovered arsenic fell below 1 ppb (ITRC pp. 88–89). The opposite trap is water that is too clean: reverse-osmosis product with almost no dissolved ions can disperse clays and clog the aquifer, which is why Hampton Roads chose ozone and biofiltration over RO (ITRC p. 84). First read the mobility table for seven troublemakers; then diagnose five real cases.

Why the injected water is not the culprit: the arsenic was always in the rock. What the project imported was a change in redox state, and redox is what the minerals answer to. That is also why the ITRC insists on characterizing the aquifer matrix — cores, mineralogy, sequential extractions — and not just the two waters (p. 41).
4 · Arithmetic you will actually use

The mixing problem

Recovered water is a blend. If a fraction f of what you pump is native groundwater and the rest is what you injected, then for any conservative constituent

Crecovered = f Cnative + (1 − f) Cinjecteda two-member mixing line — solve it for f and you have the native fraction at which a standard is breached

Des Moines injects lime-softened drinking water (TDS about 192 mg/L, pH 9.5–9.7) into the “Jordan” aquifer, whose native water carries 1,225 mg/L of TDS, 518 mg/L of sulfate, and radium above its MCL. The utility watches the TDS of recovered water and halts recovery as it approaches the secondary standard of 500 mg/L (ITRC pp. 133–135). The ITRC also gives a textbook case: fresh water at 50 mg/L chloride into a brackish aquifer at 400 mg/L, blended to meet 250 (p. 20). Solve along — four steps — and the calculator unlocks.

5 · Eyes on the water

What to monitor, and why

Monitoring is not optional in MAR — it is usually a permit condition, and it starts before the project with a baseline. The ITRC’s Appendix B explains what each parameter is for. Some are obvious (pathogens), some are the early-warning lights of the chemistry above (dissolved oxygen, pH, redox potential), and some protect the machine rather than the drinker (turbidity and iron for clogging). Match each parameter to the reason it is measured.

The LA barrier ratio: the three seawater barriers in Los Angeles County run 314 injection wells — and 777 observation wells (ITRC p. 94). Monitoring is not a rounding error in a MAR budget; it is often the larger share of the holes in the ground.
Chapter 3 quiz

Test yourself: the aquifer as a treatment plant

Scored, with explanations after grading. Retake as many times as you like — questions reshuffle.

Chapter 4 · Economics, governance, sustainability

Money, Rules, and Staying Power

A recharge scheme is a bet that a cubic metre stored today is worth more than it cost, to someone, later. Whether the bet pays depends on arithmetic you can do on one page — and whether it keeps paying depends on rules, neighbours, and nitrate that take decades to show up.

1 · The one-page arithmetic

Levelised cost

To compare a check dam in Rajasthan with an injection scheme in Perth, the UNESCO economists (Chapter 4, Ross) standardized everything to one number: the levelised cost — the constant annual revenue needed to recover all capital and operating costs over the project life, divided by the annual volume of water. Capital costs are inflated to 2016 values with a GDP deflator, converted to US dollars, and turned into an annual charge with a capital recovery factor of 0.0650 (5 % discount rate, 30-year life); operating costs are added; the total is divided by cubic metres recharged or recovered.

CRF = i(1 + i)n / [(1 + i)n − 1]    levelised cost = (capital × CRF + annual O&M) / annual volumeat i = 5 % and n = 30 years, CRF = 0.0650: every dollar of capital costs 6.5 cents a year, forever-ish

The book’s Attachment 1 works the Turku, Finland, scheme line by line: €190 million of capital in 2013, €5.6 million a year to run, 22.8 million m3 recharged and 22.3 million recovered. Solve along — eight lines — and then the calculator unlocks with Turku and Hilton Head preloaded, so you can see what a discount rate does to a benefit–cost ratio.

2 · Compared to what?

The cost ladder and the benefit ladder

Across 20 schemes the average levelised cost was US$0.75/m3 for recycled water (six schemes), US$0.16/m3 for natural water through wells or basins (eleven), and US$0.10/m3 for riverbank filtration (three) (UNESCO Ch. 4, Table 2). Benefits are harder: water is rarely sold at its true value, so the book values a scheme by the avoided cost of the next-cheapest alternative, or by the net value of the crops it grows. That makes every benefit–cost ratio an answer to the question “compared to what?” — and the volume-weighted BCRs still exceeded 2 for both natural and recycled water. Hover or tap a bar to read what each number was compared with.

Three schemes are not on the ladder because they are insurance, not supply: North London, Hilton Head, and Windhoek bank water for droughts and emergencies, and their cost is quoted per unit of daily capacity — about US$730, 490, and 860 per m3/day of recoverable supply (Ch. 4, Table 1; the same chapter’s Table 3 prints 980 for Hilton Head, and the executive summary gives the range 730–980). London’s figure is a fifth to a tenth of the cost per unit capacity of effluent reuse, desalination, or inter-catchment transfer (Case 12).
3 · Watts per cubic metre

The energy ladder

One of the nine sustainability indicators is energy: kilowatt-hours per cubic metre of recovered water, including treatment, pumping, and the cost of fighting clogging. It spans two orders of magnitude across the 28 cases. Gravity schemes on permeable ground barely register; schemes that recycle wastewater through membranes and pump it uphill do not. Windhoek’s 3.9 kWh/m3 is the highest in the book — and still beats its alternatives, an Okavango transfer at 4.9 and desalination at 11.3 (Case 13). Compare the schemes and read the comparators.

4 · The farmer’s corner

Check dams, flooded fields, and recharge for pay

Most of the world’s recharge structures are small, rural, and unmonitored. India alone is estimated to have millions of streambed structures; the four in Rajasthan’s Dharta watershed “are the only ones known to be monitored by farmers to accurately assess their performance” (UNESCO Case 4) — daily gauge readings by trained villagers, 250 wells checked weekly. Their numbers overturn two intuitions: a check dam recharges more than it can hold (1.66 times, because it fills and empties repeatedly), and a bigger dam is not a better one (the oversized dam returned a BCR of 1.7; the well-matched one 8.3). Elsewhere, recharge has become something a farmer is paid to do: Kumamoto pays rice growers to pond their fields; Arizona issues storage credits recoverable at 95 %; Pajaro Valley’s Recharge Net Metering credits pumping fees; Florida trades 10 million gallons a day of injection for 6.15 of withdrawal. Explore the three tabs.

A warning from Montana: under Western water law, a farm that irrigates less efficiently is also a farm that recharges the aquifer — and that seepage sustains the river in late summer. When a Deer Lodge hayfield proposed retiring pivot acreage to leave water in the Clark Fork, the only alternative the state could permit was the one that put the lost return flow back into the ground through a drainfield; the plain “save the water” options failed because they depleted the river from October to March (ITRC pp. 127–131). Efficiency can be illegal.
5 · Will it last?

The sustainability scorecard

Two editors of the UNESCO book scored all 28 cases on nine indicators — six environmental (groundwater level, recovered-to-recharged ratio, recovered-water quality, source-water quality, ecological flow, energy intensity) and three social (regulatory framework, risk-based permitting, stakeholder consultation) — on a scale from −5 (debilitating) to +5 (restorative). A mean above +1 is “good,” 0 to +1 “acceptable,” below 0 “needing improvement.” The exercise produced the book’s sharpest lesson: Jordan’s Wala scheme has a benefit–cost ratio near 7 and the second-lowest sustainability score, 0.2 — silt filling the reservoir, floods carrying bacteria through karst to the wells, and a catchment protection plan that lapsed. Score a case yourself, then compare with the experts. Expect to disagree with them; they disagreed with each other.

The trend the scores revealed: high-income countries averaged 1.9, upper-middle 1.3, lower-middle 0.7 — and the gap was widest on groundwater quality, regulation, and permitting, not on hydrogeology. “Effective MAR projects exist in places without supporting governance and scientific arrangements,” the editors conclude, “but investments in MAR would be more secure if such measures were in place” (UNESCO p. 16).
6 · The world tour

Forty projects, six orders of magnitude

The 28 UNESCO cases range from 640 m3 a year at a Bangladeshi village to 342 million at the Arizona Water Bank, with a median of 3.3 million; the 12 ITRC cases add American ASR wells, seawater barriers, dry-well pilots, and two modeling studies. Filter the deck by technique, source water, or continent, and click a dot on the size strip to find its card. Every card ends with the lesson its authors drew.

Chapter 4 quiz

Test yourself: money, rules, and staying power

Scored, with explanations after grading. Retake as many times as you like — questions reshuffle.