The Aquifer Under the Harvest

Executive Summary. Bangladesh's water abundance is real but misleading. World Bank data show renewable internal freshwater resources falling from 1,969.6 cubic metres per person in 1961 to 619.9 cubic metres in 2022, largely because a fixed national resource is shared among more people. Annual freshwater withdrawal is reported at 35.87 billion cubic metres, and agriculture accounts for 87.817% of measured withdrawal. Yet only 59.125% of the population used safely managed drinking water in 2024. These indicators do not prove national aquifer depletion, because internal renewable water, total withdrawal, groundwater levels, and safe service are different measures. Together they expose a governance problem: irrigation, drinking water, and aquifer condition are managed through separate accounts. Bangladesh needs one seasonal water balance that distinguishes surface from groundwater, measures recharge and pumping by zone, and protects drinking-water security while preserving farm productivity.

Per-person water is shrinking even when the resource is not

Renewable internal freshwater per person fell from 1,969.6 cubic metres in 1961 to 1,193.0 in 1980, 780.4 in 2000, 689.9 in 2010, and 619.9 in 2022. The series does not say that the rivers or aquifers lost the same proportion of water. Its denominator is population. A broadly fixed renewable resource divided among more people produces a falling per-person number even if hydrology is unchanged.

Source: World Bank WDI indicator ER.H2O.INTR.PC, selected observations.

That caveat does not make the decline harmless. A smaller per-person endowment means less room for unmanaged competition among farms, cities, industry, navigation, ecosystems, and dry-season drinking water. It also makes timing more important. Monsoon flows cannot automatically satisfy dry-season demand. Water available in one basin or depth cannot be assumed available in another. Salinity and arsenic can make a physically present source unusable for a particular purpose.

The national series is therefore a pressure indicator, not an aquifer diagnosis. It should trigger a better account: how much water enters each management zone, how much is stored, how much is pumped, how much returns, and how quality changes. Without that account, scarcity appears only after wells deepen, pumps fail, or drinking-water sources become unsafe.

The harvest is the largest measured claimant

Agriculture accounts for 87.817% of measured freshwater withdrawal in the latest WDI observation. All other uses together account for 12.183%. Total annual withdrawal is reported at 35.87 billion cubic metres.

Source: World Bank WDI indicators ER.H2O.FWAG.ZS and ER.H2O.FWTL.K3, latest reported observations.

This does not mean that farm water is waste. Irrigation converts dry-season land and labour into food, income, and price stability. A policy that simply commands farmers to pump less without changing crop incentives, energy pricing, conveyance, timing, or access to alternatives will transfer risk to rural households and consumers.

It also does not mean that all agricultural withdrawal comes from groundwater. The WDI share covers freshwater withdrawal by sector. The title of this essay points to the aquifer because dry-season irrigation makes groundwater governance especially consequential, but the national indicator must not be relabelled as groundwater extraction. Surface and groundwater need separate measurement.

The institutional problem is that the pump is local while the aquifer is shared. Each farmer decides against a private crop return and a private pumping cost. The decline in a connected water table is borne across neighbours and future seasons. Where electricity or fuel prices fail to reflect water scarcity, the resource has no effective signal until pumping becomes physically difficult.

That common-pool structure makes blunt pricing risky. A uniform charge can fall hardest on small farmers without distinguishing a balanced zone from a depleted one. Free power can push in the opposite direction by separating the pumping decision from its energy cost. The policy instrument should follow the measured condition. In a balanced zone, reliable access and maintenance may be the priority. In a persistently stressed zone, scheduling, crop support, collective pumping rules, and targeted energy reform may be justified. The national account should enable differentiated policy rather than manufacture one national scarcity price.

A fair response begins with information, not punishment. Metering every small pump may be costly and politically brittle. Feeder-level electricity data, pump registration, remote sensing of irrigated area, crop calendars, and sentinel observation wells can produce a usable seasonal estimate. The aim is to identify zones of persistent imbalance and offer alternatives before imposing restrictions.

Drinking-water progress remains too slow

Safely managed drinking-water coverage was 55.309% in 2000, 55.000% in 2010, 58.659% in 2020, and 59.125% in 2024. The series measures a service standard, not only the presence of a source. Water must be available when needed, accessible, and free from priority contamination.

Source: World Bank WDI indicator SH.H2O.SMDW.ZS, selected observations.

The slow movement matters because drinking water has the highest welfare value but not always the strongest pumping capacity. A deep irrigation pump and a household source can draw from connected systems while operating under different institutions. Coastal salinity, arsenic risk, falling seasonal levels, and contamination create different local constraints, but households experience the result as one service failure.

This is why a water balance must include quality. A cubic metre that is too saline for a crop, contaminated for drinking, or environmentally required downstream is not freely interchangeable. National totals can create false comfort by adding water that cannot serve the same place, season, or purpose.

The drinking-water series also disciplines the policy objective. Irrigation efficiency should not be pursued only to lower an abstract withdrawal ratio. It should release reliable water, energy, and institutional capacity for households and ecosystems while maintaining farm output. That requires measuring outcomes, not just distributing equipment.

Efficiency claims also require care. A field technique can reduce water applied at the plot while leaving basin consumption unchanged if farmers expand irrigated area or switch to a thirstier crop. Water not applied may previously have returned to the aquifer or canal. A credible pilot therefore tracks pumping, crop output, irrigated area, farmer income, and local water levels together. Equipment sales or demonstrations are implementation counts, not proof of saved water.

The strongest counterargument is that national indicators cannot see an aquifer

The strongest objection is that these WDI series are too aggregated and too slowly updated to support the title. The agricultural share repeats across many reported years, while the per-person series is mechanically driven by population. Neither reveals a water table, recharge rate, pumping depth, or seasonal cone of depression. A national essay could turn a real local problem into a falsely uniform crisis.

That objection is correct. The evidence here does not establish that every aquifer is declining or that a particular crop caused a particular fall. Bangladesh has multiple hydrogeological settings, large seasonal variation, and substantial transboundary flows. A management rule suitable for a stressed dry-season zone could be harmful in a water-abundant one.

But aggregation is also the finding. The national account tells policymakers that agriculture dominates withdrawal and per-person internal water is falling, while it cannot say where the balance breaks. The appropriate conclusion is not a national pumping ban. It is a zonal accounting system capable of distinguishing safe use from depletion.

The Bangladesh Water Development Board's assessment archive is the natural institutional starting point, but published assessment documents should connect to a maintained data product. A report can describe a monitoring campaign. A public time series can show whether the condition changed, which wells are comparable, and where observations are missing.

Public access is not cosmetic. Farmers and local water suppliers make better decisions when they can see whether a seasonal decline is local, recurrent, or exceptional. Researchers can identify sensor breaks and inconsistent wells before those errors become policy. A clear vintage also prevents an old assessment from being cited as a current condition. The account should preserve raw observations, quality flags, and revisions instead of publishing only a smoothed map.

What would change this conclusion

A specific test would change the argument: if a transparent seasonal account shows recharge matching or exceeding groundwater abstraction in the major irrigation zones, stable dry-season levels, and no deterioration in drinking-water reliability, then the claim of an ungoverned aquifer risk would be overstated. The test must be zonal and repeated. One wet year or one national average cannot settle it.

First, publish a seasonal surface-and-groundwater balance by management zone. Combine observation wells, river flows, rainfall, irrigated area, crop calendars, and pumping proxies. Owner: Bangladesh Water Development Board with agricultural and water-supply agencies. Success signal: each zone publishes comparable pre-monsoon and post-monsoon balances, observation coverage, and uncertainty.

Second, buy crop output with less pumping in stressed zones. Test irrigation scheduling, alternate wetting practices where agronomically appropriate, conveyance repair, and crop choices through measured pilots. Owner: agricultural extension and irrigation authorities. Success signal: verified water withdrawal per unit of crop output falls without reducing participating farmers' net income.

Third, reserve drinking-water security explicitly. Identify household sources connected to stressed or contaminated systems and fund alternatives before imposing farm restrictions. Owner: public-health engineering and local government. Success signal: safely managed service rises above the 59.125% latest observation while dry-season source failures decline in the targeted zones.

The aquifer under the harvest is not a metaphor for blaming farmers. It is a shared balance sheet that the state has not yet made visible enough to manage. The possibility is to protect both the crop and the source by measuring the link between them.

Sources

1. World Bank WDI. Renewable internal freshwater resources per capita, ER.H2O.INTR.PC. Retrieved 2026-08-23: https://data.worldbank.org/indicator/ER.H2O.INTR.PC?locations=BD
2. World Bank WDI. Annual freshwater withdrawals, agriculture, ER.H2O.FWAG.ZS. Retrieved 2026-08-23: https://data.worldbank.org/indicator/ER.H2O.FWAG.ZS?locations=BD
3. World Bank WDI. Annual freshwater withdrawals, total, ER.H2O.FWTL.K3. Retrieved 2026-08-23: https://data.worldbank.org/indicator/ER.H2O.FWTL.K3?locations=BD
4. World Bank WDI. People using safely managed drinking water services, SH.H2O.SMDW.ZS. Retrieved 2026-08-23: https://data.worldbank.org/indicator/SH.H2O.SMDW.ZS?locations=BD
5. Bangladesh Water Development Board. Groundwater assessment archive. Retrieved 2026-08-23: https://www.bwdb.gov.bd/archive/pdf/16594.pdf