28 Water, rivers and the delta

Chapter 28 of 60 in the Bangladesh 2036 research base. Contents of the series.

The water decade to FY36 is decided underground and in the delivery budget before it is decided across the border

Bangladesh is a delta whose water account has three ledgers, and the claim this chapter defends is that they now rank in this order: the groundwater ledger, the delivery ledger, and only then the transboundary ledger that absorbs the politics. The groundwater ledger is the one the country lives on: 90.68 percent of households drink from tube wells and 5.33 percent from piped supply [BBS Census 2022], the survey record behind chapter 16's inheritance puts mean groundwater arsenic at 90.02 parts per billion with 39.09 percent of sampled points above the national 50 parts per billion standard [Harvard Dataverse 2020, DOI 10.7910/DVN/JPYWDD], and renewable freshwater per head has fallen to 620 cubic metres, under the 1,000 cubic metre scarcity line [WB WDI 2026]. The delivery ledger is the one the budget does not fund: the Delta Plan 2100 remains a framework without published investment targets [GED Plans 2018], the FY26 development programme gives the water ministry 8,489.86 crore BDT, 3.69 percent of the 230,000.00 crore BDT programme the National Economic Council approved on 18 May 2025 [Planning Commission ADP 2025], and the projects that would move Dhaka off the aquifer slip, Saidabad treatment plant phase 3 replanned twice to June 2029 at 16,014.83 crore BDT, 3.5 times its original cost, with 7.65 percent spent [DWASA 2026]. The transboundary ledger is the one with a deadline: the Ganges water sharing treaty signed in December 1996 completes its 30 year term in December 2026 [MoWR 1996], and no successor, and no Teesta agreement, is recorded in the sources this chapter draws on. Chapter 11 established that the fast hazards are beaten and the slow hazards are unfunded; this chapter goes one institution and one instrument deeper, into the utilities, the dredgers, the projects and the treaty that decide whether the slow water hazards are governed, and it leaves the coastal economy and disaster finance account to chapter 29.

The record: nine households in ten drink from a tube well while the flagship surface water projects replan

The drinking water base is the aquifer. The Census 2022 district tables put 90.68 percent of households on tube wells against 5.33 percent on piped supply, with Dhaka district the inversion at 61.5 percent piped and 37.75 percent tube wells [BBS Census 2022, district drinking water tables]. In the salinity belt the aquifer fails and the census shows the fallback: Bagerhat households report 21.99 percent drinking surface water and 16.74 percent rainwater, against 53.08 percent tube wells, and Satkhira 7.66 percent surface [BBS Census 2022, district drinking water tables]. In Dhaka's informal settlements the municipal connection is the dependency: 68.85 percent of slum dwellings depend on supply water from the utilities [Figshare 2024, DOI 10.6084/m9.figshare.31641787]. The quality side of the same aquifer is measured: the stored survey records mean groundwater arsenic of 90.02 parts per billion, 59.94 percent of samples above the World Health Organization guideline of 10 parts per billion and 39.09 percent above the national standard of 50 [Harvard Dataverse 2020, DOI 10.7910/DVN/JPYWDD], the reading chapter 16 carries into its boro account and chapter 31 inherits into the health ledger.

Dhaka's utility shows what the aquifer dependency means at scale. The Dhaka Water Supply and Sewerage Authority produced 2,560 million litres per day in FY20 from 896 deep tube wells and five treatment plants, serving 392,400 connections, with about 78 percent of supply from groundwater and 22 percent from surface water treated from the Shitalakshya and Buriganga [DWASA 2020]. The authority's own report records the water table falling 2 to 3 metres per year under continuous abstraction, revenue income rising from about 300 crore BDT in FY08 to 1,506 crore BDT in FY20, its own 12 year turnaround record, and a target of lifting the surface water share to 70 percent [DWASA 2020]. The audited accounts to FY25 show the utility is financially operational, operating profit of 617.85 crore BDT in FY25 against 437.79 crore BDT in FY24 [DWASA 2026], but the annual report record available to this chapter ends at FY20, so the current groundwater share could not be confirmed, with the authority's post FY20 reports the resolving source. The secondary utilities show the same structure in smaller cities: Khulna WASA carried a system loss of 15.5 percent and an operating loss of 84.46 crore BDT in FY23, while Rajshahi WASA ran a 0.94 percent system loss and a 13.55 crore BDT operating loss [MoF SOE 2023].

Floods remain the base condition of the delta, and the recent record is high. The Water Development Board series shows the country at or above a quarter of its territory flooded in four consecutive years, 40.0 percent in 2020, 33.0 in 2021, 27.1 in 2022 and 25.7 in 2023 [BWDB 2023]. The satellite record extends it: FAO's flood mapping put the August 2025 monsoon peak at 33,686.33 square kilometres inundated, 22,342.83 square kilometres of it cropland, with 23.76 million people exposed, against an August 2024 peak of 24,708.74 square kilometres, and the April 2026 pre monsoon window at 6,580.85 square kilometres and 3.39 million exposed [FAO 2026]. EM-DAT logged 3 flood events in 2024 and 2 in 2025, the most since the 3 events of 2004 [EMDAT 2026]. The long tail of a single flood is measured in schooling: the 1988 flood cohorts completed 4.79 years of schooling against 5.19 for unaffected cohorts, a deficit of 0.4 years [Zenodo 2024, DOI 10.5281/zenodo.17240928]. Riverbank erosion is the hazard without an official series, a gap chapter 11 records; the survey record fills it partly: among 2,180 riverside respondents, 89.5 percent report that erosion in their reach has worsened or changed, 90.78 percent report flood patterns altered and 33.76 percent report floods more frequent [Zenodo 2020, DOI 10.5281/zenodo.18255755]. National erosion displacement statistics remain unconfirmed, with the Water Development Board the resolving source.

The state's river machinery is an engineering programme with an annual rhythm and a thin fleet. The Water Development Board's FY20 report records about 38.00 kilometres of channel and about 5.00 million cubic metres dredged in FY19, rising to about 10.08 million cubic metres in FY20, under a procurement project for 21 dredgers of which 8 were in service by FY20 [BWDB 2020]. Its flagship coastal project, the World Bank financed Coastal Embankment Improvement Project phase 1, carries a cost of 3,280.00 crore BDT across six coastal districts and stood at 41.82 percent financial progress in the FY20 report [BWDB 2020]. The irrigation claim on the aquifer grows with boro: irrigated area reached 19,750 thousand acres in 2022 from 19,217 thousand acres in 2019 [BBS Agri 2022], the demand side of the withdrawal record chapter 16 owns.

The money is the delivery ledger. The FY26 development programme allocates 8,489.86 crore BDT to the Ministry of Water Resources across 78 projects, 3.69 percent of the 230,000.00 crore BDT programme the National Economic Council approved on 18 May 2025, and 10,641.40 crore BDT across 98 projects to the environment, climate change and water resources sector as a whole, 4.63 percent, both shares as the programme prints them [Planning Commission ADP 2025]. Chapter 11 carries the combined climate relevant share; the water specific fact is the gap between the annual allocation and the plan: the Delta Plan 2100's investment and financing targets are not held in the sources this chapter draws on and could not be confirmed, with the plan document the resolving source [GED Plans 2018].

Mechanism: the aquifer is unpriced, the rivers are transboundary, and the substitute moves at project speed

The groundwater mechanism is an unpriced commons with a quality mirror. Drinking water for nine households in ten and irrigation for 19.75 million acres draw on the same alluvial aquifer, at a volumetric price of approximately zero in both uses, so withdrawal rises with population and with boro until scarcity or quality forces the adjustment the market will not make. The quality mirror is arsenic: the same abstraction that lowers water tables redistributes flow paths, and the survey's 39.09 percent exceedance of the national standard is the measured shadow of the withdrawal record [Harvard Dataverse 2020, DOI 10.7910/DVN/JPYWDD]. On the coast the mechanism completes itself as chapter 11 documents: salinity pushes households off tube wells onto surface and rainwater, the Bagerhat and Satkhira census shares above, so the poorest belt is the one that has already lost the utility the rest of the country still draws on. Dhaka is the abstraction concentrated: 896 deep tube wells under one city, the water table falling 2 to 3 metres per year, and the utility's own answer, surface water treatment, is a capital programme, not a tariff change [DWASA 2020].

The river mechanism is confinement. Embankments cut the sediment exchange that built the floodplains, confined rivers rise between their levees, erosion protection on one bank transfers the attack to the next, and the response is an annual dredging and bank protection programme, about 10.08 million cubic metres in FY20 from a fleet of 8 bought dredgers [BWDB 2020], against river systems whose sediment loads the programme does not publish, which leaves the maintenance versus accumulation arithmetic not established, with the Water Development Board the resolving source. The flood peak record, 33,686.33 square kilometres in August 2025 [FAO 2026], is the annual exam the embankment line sits, and the schooling deficit of the 1988 cohorts [Zenodo 2024, DOI 10.5281/zenodo.17240928] measures what a failed exam costs for decades.

The transboundary mechanism is that the dry season hydrology of the Ganges, Teesta and Meghna systems is set upstream. The 1996 Ganges treaty fixed a sharing arrangement at Farakka for a 30 year term that completes in December 2026 [MoWR 1996]; the specific sharing formula and the status of a successor could not be confirmed from the sources this chapter draws on, the Ministry of Water Resources and the Joint Rivers Commission being the resolving sources. The Teesta arrangement initialled in 2011 was never signed, a status that is likewise not established here. What is not ambiguous is the exposure: the northwest's dry season irrigation, the barind's groundwater stress and the Gorai's freshwater needs all sit downstream of that negotiation, which is why the treaty ledger ranks third in this chapter's claim but first in its deadline.

The delivery mechanism explains the ordering of the claim. Water projects in this decade are rescheduled, not cancelled: Saidabad phase 3, approved in 2015 to supply an additional 450 million litres per day from the Meghna, was revised in 2021 and again in March 2025, its cost rising from 4,597.36 crore BDT to 16,014.83 crore BDT, its completion target moving to June 2029, and its cumulative financial progress standing at 7.65 percent [DWASA 2026]. The Extended Dhaka Water Supply Resilience Project, 4,040.00 crore BDT for 80 kilometres of transmission, 1,575 kilometres of distribution and 55 district metered areas, started in January 2024 with a June 2029 target and had spent 4.95 crore BDT, 0.12 percent, by April 2026 [DWASA 2026]. The Dhaka Sanitation Improvement Project, 5,187.89 crore BDT revised, targets the Pagla treatment plant upgrade serving about 1.5 million people with an April 2028 completion [DWASA 2026]. The mechanism is the one chapter 04 documents for the development programme generally, execution lags and replanning, with one water specific consequence: the substitute for the aquifer arrives at project speed, roughly half of current Dhaka production if the three treatment plants all deliver, while aquifer demand arrives at the population speed chapter 14 prices, so the groundwater share holds near 78 percent and the water table keeps falling while the projects replan.

The decade ahead: three delivery clocks run to 2029 and the treaty clock runs first

The treaty clock is the one with a date. The Ganges treaty completes its 30 year term in December 2026 [MoWR 1996], so the first years of the outlook decade carry the renewal negotiation with India, under a Dhaka government formed after the 2024 transition, and the Teesta track runs behind it [MoWR 1996, successor status not established]. No named forecaster publishes a successor scenario in the sources this chapter draws on; the World Bank's Country Climate and Development Report is the standard projection of basin stress and climate migration, and its specific figures for Bangladesh could not be confirmed from the sources this chapter draws on, exactly as chapter 11 records [WB CCDR 2022]. The institutional track runs in parallel: the draft Water Resources Planning, Management and Conservation Act would create a Department of Water Resources Planning and Management on top of the Water Act 2013 framework [MoWR 2025] [WARPO 2013], and the draft's passage status could not be confirmed, with the ministry the resolving source.

The delivery clocks all target 2028 to 2029. Saidabad phase 3 targets June 2029 for 450 million litres per day [DWASA 2026]; Gandharbpur, planned at 500 million litres per day from the Meghna, and the Padma plant at Jasholdia, planned at 450 million litres per day, carry schedules from the FY20 report that are not confirmed pending current project documents [DWASA 2020]. On the delivery arithmetic this chapter's scenario assumes: if the three plants deliver as planned, about 1,400 million litres per day of surface supply against 2,560 million litres per day of current production moves Dhaka's surface water share from 22 percent toward and past half, which is the operational meaning of the authority's 70 percent target [DWASA 2020]. The sanitation and resilience projects target April 2028 and June 2029 respectively [DWASA 2026]. Each slip is measurable in the groundwater share and in the Water Development Board's abstraction records, and each delivery is equally measurable; the FY36 surface water share the scenario tracks is the single number that decides whether the Dhaka aquifer story of chapter 14 stabilises or compounds.

The delta plan clock has no date, which is the finding. The Delta Plan 2100 names the institutions and the investment needs but its financing targets are unpublished in the sources this chapter draws on [GED Plans 2018], so the operational plan for the decade is the annual programme: a water ministry line of 10,532.1 crore BDT in FY26 [Planning Commission ADP 2025], a dredging fleet of 8 [BWDB 2020], and a flagship coastal project at less than half delivered [BWDB 2020]. The scenario this chapter assumes follows chapter 15's stall line for the water budget, nominal flat and real declining, unless the plan acquires a financed programme, and the flood peak record of 2025 [FAO 2026] is the stress the annual programme would have to beat.

Three risks print in wells, budgets and treaty texts, and the upside converts delivery into the surface water switch

Risks. First, a treaty lapse: December 2026 passes without a signed successor and dry season Ganges sharing falls to ad hoc arrangement, with the southwest's dry season surface flows the first casualty; the revealing indicators are the Joint Rivers Commission negotiation record and dry season flows at the Hardinge Bridge gauge, a series not held in the sources this chapter draws on and not established, with the commission the resolving source. Second, a delivery failure in Dhaka: the treatment plants replan again and the groundwater share holds near 78 percent while connections grow, and the revealing indicators are the project milestone reports and the authority's deep tube well count against its 896 of FY20 [DWASA 2020] [DWASA 2026]. Third, a quality and scarcity regime: arsenic exceedance above 39.09 percent rises as wells deepen and freshwater per head, 620 cubic metres in 2022 [WB WDI 2026], falls toward the 550 cubic metre line chapter 11 sets, and the revealing indicators are the next arsenic survey and the per head series.

Upside. First, the surface water switch: the June 2029 targets hold, roughly 1,400 million litres per day of new treated supply against 2,560 million litres per day of production [DWASA 2020] [DWASA 2026], the groundwater share halves, and the 2 to 3 metre annual fall flattens, with the revealing indicator the authority's supply mix report. Second, a basin cooperation dividend: a successor Ganges treaty plus a signed Teesta arrangement converts a negotiation year into dry season flow security for the northwest irrigation economy chapter 16 models, and the revealing indicator is the signed texts and the dry season gauge record. Third, the delta plan acquires a budget: the draft water act's new department [MoWR 2025] publishes the first Delta Plan financing report with a budget line, converting the 4.6 percent annual share into basin scale programming, and the revealing indicator is that report itself [GED Plans 2018].

What to watch: five indicators whose thresholds mark the water regime

1. Dhaka surface water share of supply. Current value 22 percent in FY20, the last annual report available to this chapter, with later readings not confirmed pending the authority's reports [DWASA 2020]. Threshold: above 40 percent by FY31, the horizon the delivery schedule targets, marks the switch regime; a reading below 30 percent confirms the aquifer regime the stall scenario assumes compounding into the 2030s.
2. Groundwater arsenic exceedance of the national standard. Current value 39.09 percent of sampled points above 50 parts per billion [Harvard Dataverse 2020, DOI 10.7910/DVN/JPYWDD]. Threshold: a reading above 45 percent in the next survey confirms the deepening abstraction regime; a reading below 30 percent means mitigation and source switching worked.
3. Monsoon flood peak extent. Current value 33,686.33 square kilometres at the August 2025 peak, with 23.76 million people exposed [FAO 2026]. Threshold: two consecutive monsoon peaks above 35,000 square kilometres mark the embankment line failing its exam; a run of peaks below 20,000 square kilometres marks the river programme holding.
4. Water ministry share of the development programme. Current value 8,489.86 crore BDT in FY26, 3.69 percent of the 230,000.00 crore BDT programme the National Economic Council approved [Planning Commission ADP 2025]. Threshold: a sustained share above 6 percent marks the adaptation financing turn for water specifically; a share below 3 percent confirms the stall scenario's real erosion assumption.
5. Ganges treaty successor status. Current value: the 1996 treaty completes its 30 year term in December 2026 with no successor recorded in the sources this chapter draws on [MoWR 1996, status not established]. Threshold: a signed renewal with a dry season sharing schedule before expiry marks continuity; lapse into ad hoc sharing marks the negotiation failure that the stall scenario assumes.

Sources used

[BBS Census 2022] Bangladesh Bureau of Statistics, Population and Housing Census 2022, district drinking water source tables via the lake/water_sanitation parquet and bdpolicy.db series census2022_tubewell_water_access_pct, census2022_piped_water_access_pct. [Harvard Dataverse 2020] Harvard Dataverse groundwater arsenic survey for Bangladesh, DOI 10.7910/DVN/JPYWDD, mean parts per billion and World Health Organization and national guideline exceedance shares, via bdpolicy.db series tox_groundwater_arsenic_. [WB WDI 2026] World Bank World Development Indicators via the climate/wb_environment_bd_deep parquet, series: ER.H2O.INTR.PC renewable internal freshwater per capita. [GED Plans 2018] General Economics Division, Bangladesh Delta Plan 2100, framework quoted, investment and financing targets not held in the sources this chapter draws on. [Planning Commission ADP 2025] Annual Development Programme 2025-2026, Programming Division, Bangladesh Planning Commission, June 2025, ADP at a glance sector table and ministry wise allocation table, every share taken on the 230,000.00 crore programme approved by the National Economic Council on 18 May 2025, https://adp.plancomm.gov.bd/book/2025-2026-ADP-BOOK.pdf, accessed 6 September 2026; project status reports via the ocr_text/dhaka_wasa project files. [DWASA 2020] Dhaka Water Supply and Sewerage Authority annual report 2019-20 via ocr_text/dhaka_wasa, water production, deep tube wells, groundwater share, connections, revenue and treatment plant plans. [DWASA 2026] Dhaka Water Supply and Sewerage Authority audited financial statements FY2024-25 and project implementation status reports of April 2026 for Saidabad phase 3, EDWSRP and DSIP via ocr_text/dhaka_wasa, unit lakh taka converted to crore. [MoF SOE 2023] Finance Division state-owned enterprise evaluation tables via bdpolicy.db, Kwasa and Rwasa system loss, customers, production and operating results, FY2021-22 to FY2022-23. [BWDB 2023] Bangladesh Water Development Board historical flood affected area record via the lake/climate parquet and bdpolicy.db series bwdb_flood_affected_area_sq_km, bwdb_flood_affected_pct. [FAO 2026] FAO flood impact mapping for Bangladesh via bdpolicy.db series fao_flood_peak_area_monthly_sq_km, fao_flood_peak_cropland_monthly_sq_km, fao_flood_peak_pop_exposed_monthly. [EMDAT 2026] EM-DAT Emergency Events Database, UCLouvain, flood event counts via bdpolicy.db series emdat_flood_events_total. [Zenodo 2024] Zenodo 1988 flood schooling microdata, DOI 10.5281/zenodo.17240928, affected and unaffected cohort schooling years via bdpolicy.db series edu_1988_flood_. [Zenodo 2020] Zenodo riverbank erosion and flood perception survey, 2,180 riverside respondents, DOI 10.5281/zenodo.18255755, via bdpolicy.db series erosion_perception_worsened_pct, flood_pattern_changed_pct, flood_more_frequent_pct. [BBS Agri 2022] Bangladesh Bureau of Statistics agriculture statistics, irrigated area, via bdpolicy.db series bbs_irrigated_area_k_acres. [BWDB 2020] Bangladesh Water Development Board annual report FY2019-20 via ocr_text/bwdb_deep, dredging volumes, dredger procurement and the Coastal Embankment Improvement Project phase 1 cost and progress. [Figshare 2024] Figshare, Dhaka slum dwellings dependent on municipal supply water, DOI 10.6084/m9.figshare.31641787, via bdpolicy.db series dhaka_slums_supply_water_dependence_pct. [MoWR 1996] Ministry of Water Resources, the Ganges water sharing treaty of December 1996, 30 year term; treaty terms and successor status not held in the sources this chapter draws on. [MoWR 2025] Ministry of Water Resources, draft Water Resources Planning, Management and Conservation Act text via ocr_text/mowr, draft status. [WARPO 2013] Water Resources Planning Organization, Bangladesh Water Act 2013 via ocr_text/warpo. [WB CCDR 2022] World Bank Bangladesh Country Climate and Development Report, resolving source for basin stress and migration projections, figures not held in the sources this chapter draws on.

---

Verified line by line against primary sources: 67 claims checked, 5 corrected.

Previous: 27 Renewables and the energy transition

Next: 29 Coastal zone and disaster risk

All sixty chapters and the ten narrative chapters.