INSIGHTS · 7 min read
Sri Lanka’s Reservoirs Gained 295 Million Cubic Metres in Thirteen Days. Three-Quarters of It Went to the Power Stations
The national reservoir figure has risen from 55% to 60% since mid-September, and every account of it reports a recovery. The Mahaweli Authority’s own status sheet prints two sub-totals, not one. Read them separately, and the recovery has an address: the reservoirs that generate electricity hold 40% of the country’s capacity and took 74% of the water. Fifteen days before a fixed sowing window for roughly 830,000 hectares, the tanks that serve it have moved two and a third points.
By the ESGNexus Editorial Team · September 2026 · Estimated reading time: 7 minutes
KEY TAKEAWAYS
|
Sri Lanka has spent three months reading a single number about its water. Through August and early September, that number was falling, and it was reported as a drought. Since mid-September, it has been rising and is being reported as a recovery. Both readings use the same figure, a national total that averages two systems with different jobs, different catchments, and, this month, sharply different fortunes.
The Mahaweli Authority of Sri Lanka’s Water Management Secretariat publishes a reservoir status sheet that does something no news account has done: it prints a separate subtotal for the reservoirs attached to power stations. The generation number that ESGNexus reported in August as unpublished is in fact published weekly on a government website, one line above the national total. This article is built on three dated readings of that sheet.
What the Sheet Actually Says
The gross capacities are identical in all three readings: 5,489.01 mcm nationally, of which the power-station block is 2,182.13 mcm, and the irrigation and multipurpose remainder is 3,306.88 mcm. Only the storage figures move.
| Reading (ESGNexus, dated) | National total | Power-station block |
| 16 September 2026 | 2,997.89 mcm · 55% | 1,334.14 mcm · 61% |
| 25 September 2026 | 3,214.08 mcm · 59% | 1,509.87 mcm · 69% |
| 29 September 2026 | 3,293.38 mcm · 60% | 1,553.19 mcm · 71% |
| Thirteen-day move | +295.49 mcm · +5 points | +219.05 mcm · +10 points |
Source: Mahaweli Authority of Sri Lanka, Water Management Secretariat, ‘Latest Status of Reservoirs’ — mahaweli.gov.lk. Three readings taken by ESGNexus on 16, 25 and 29 September 2026. The percentages are as printed on the sheet; the arithmetic below is ours.
Every percentage on that sheet reconciles against its own volumes. 3,293.38 divided by 5,489.01 is 59.9995%, printed as 60%. 1,553.19 divided by 2,182.13 is 71.18%, printed as 71%. The same division works on both earlier readings. This matters because it is the test of whether a document can be trusted with a story: the sheet’s totals agree with the sheet’s own components, to the second decimal place, on every reading.
Forty Per Cent of the Capacity Took Three-Quarters of the Water
Subtract the power-station block from the national total, and the remainder is the irrigation and multipurpose system. The two blocks sum to the whole exactly, with no residual, on every reading. That is what makes the next figure a measurement rather than an estimate.
| Block | Share of national gross capacity | Share of the 295.49 mcm gain |
| Power-station reservoirs | 39.75% (2,182.13 of 5,489.01 mcm) | 74.13% (+219.05 mcm; 61% to 71%) |
| Irrigation and multipurpose | 60.25% (3,306.88 of 5,489.01 mcm) | 25.87% (+76.44 mcm; 50.3% to 52.6%) |
Derived by ESGNexus from the three readings above. The gain figures sum to the national gain exactly: 219.05 plus 76.44 equals 295.49 mcm.
This is the distribution behind the recovery, and it is not a claim about anyone’s conduct. Nobody diverted the water. The rain of the last fortnight fell on the wet-zone highland catchments that feed Kotmale, Victoria, Randenigala and Rantambe, but not in the same quantity over the dry-zone tanks. Where the rain falls is a fact of weather, not policy. The fact about the country’s exposure is that the two systems have diverged, and that the number in every headline conceals this by averaging them.
For a Sri Lankan business, the practical consequence is that hydro-generation risk and irrigation risk have decoupled this month, even though they were tracked as a single metric. A company modelling its grid electricity cost or its Scope 2 trajectory should be reading the power-station line, which is at 71% and climbing steeply. A company modelling agricultural supply, rural demand, plantation yields or crop-insurance exposure should be reading the remainder, which is at 52.6%.
The Concentration Is Easing, Which Is the Part a Single Reading Would Have Missed
Two readings make a comparison and invite a thesis. Three make a trend, and this trend is turning. Between 16 and 25 September, the power-station block absorbed 81.3% of the water the country gained. Between 25 and 29 September it absorbed 54.6% of a four-day national gain of 79.30 mcm; the power block took 43.32 mcm, and the remainder took 35.98 mcm, which again sum exactly.
The rain is beginning to reach the rest of the system. Anyone who had published the 81% as a standing state of affairs would already be carrying a figure the source no longer supports — which is the argument for dating every reading of a file that is overwritten in place, and for saying out loud how many times you have read it.
The Two Tanks That Did Not Move
The sheet’s per-reservoir rows are less reliable than its sub-totals, for a reason set out below. Two rows in the Maha rice-growing districts returned identical values on both readings taken on 29 September, and they counter the national figure.
| Reservoir | 25 September | 29 September |
| Kaudulla | 22% | 21% — down |
| Minneriya | 39% | 39% — unchanged |
| Nachchaduwa | not read | 34% |
Source: as above. Kaudulla and Minneriya were each returned identically by two separate readings of the 29 September sheet.
While the national figure gained five points and the power-station block gained ten, one of these tanks lost a point, and the other remained unchanged.
A Fixed Planting Date, and an Allocation Plan for the Season That Ended in August
Storage tells you how much water there is. It does not tell you where it is going, and that is the question a cultivation season turns on. The Department of Agriculture’s advisory of 4 September 2026 set the Maha 2026/27 sowing window for 15 to 25 October, targeting roughly 830,000 hectares of paddy, and cited the expectation of above-average October and November rainfall as its rationale. That date is fifteen days away and fixed.
The document that outlines how the water is to be allocated for that season is a seasonal operating plan, and the Mahaweli Authority publishes it at a standing address on its website. The plan sitting at that address on 30 September 2026 is the plan for Yala 2026 — the season that ran to August and has finished. It contains no reference to Maha 2026/27; it carries no date of issue or approval anywhere — not on the cover, not in a header or footer, not in a signature block — and its monthly projection tables run from March 2026 to August 2026. It also states the point from which its numbers start:
| Simulations for this report were initiated using the estimated reservoir/tank conditions as of the beginning of March 2026. — Mahaweli Authority of Sri Lanka, Seasonal Operating Plan – Yala 2026, read 30 September 2026 |
Conditions at the beginning of March 2026 reflect the state before the dry season emptied the tanks over the middle of this year. The allocation document on the Authority’s website models a season that has ended, starting from a seven-month-old position taken before the drought that has dominated Sri Lankan water coverage since June.
The narrow and checkable claim is this: fifteen days before a fixed sowing window for roughly 830,000 hectares, the seasonal operating plan published where the Authority publishes seasonal operating plans is the one for the previous season.
That is not an allegation that no plan exists, that anything improper has occurred, or that any rule has been broken. Planning meetings for the 2026/27 Maha season have been held — the Ministry of Agriculture, Livestock, Land and Irrigation has posted notices of preliminary discussions at the Govijana Mandiraya and of a decisive discussion on the cultivation plan and irrigation water distribution at Gannoruwa. What is absent is the published allocation document, at the point in the calendar when a farmer, a miller, an agricultural lender or an insurer would want to read it. The predecessor plan is sitting there, which is what makes the absence legible rather than speculative.
What We Have Not Read
This article rests on a file with no public archive, and saying so precisely is what makes it checkable.
- The 16 and 25 September figures are ESGNexus’s own earlier readings of the same URL. The sheet is overwritten in place, and the Authority publishes no archive, so this is a series of three dated readings by one publication, not a departmental time series. It cannot be independently reconstructed from the source.
- The per-reservoir volume columns do not survive automated extraction. Kaudulla’s extracted volumes are 40.9% versus a printed 21%; Minneriya’s are 59.4% versus a printed 39%. Only the percentages and the two subtotals are used here, and the subtotals reconcile exactly, while the rows do not.
- Kalawewa is not reported in this article. Two readings of the 29 September sheet returned different values for that one row, and neither matched the division of its own extracted volumes. Three inconsistent figures for one reservoir on one morning are not a fact, and a reading of the row by eye would settle it.
- Senanayake Samudraya does not appear as a row on this sheet, in three readings over two days. Any figure attributed to it from this source should be treated as an extraction error.
- The Department of Meteorology’s current ENSO and IOD bulletin, reference SF-2026-09-ENSO-IOD, was issued on 10 September 2026 and gives a ggreater-than-90 chance of a very strong El Niño through the coming season. A quarterly successor is overdue and will land inside the sowing window. The seasonal expectations in this article are those of the September bulletin.
- The most recent seasonal operating plan for the Maha 2026/27 season, if approved, has not been read because it has not been published at the same address as the Yala 2026 plan.
What to Do Now
1. Stop using one reservoir percentage for two different risks. Grid electricity and Scope 2 assumptions belong in the power-station subtotal, which is at 71%. Agricultural supply, rural demand, and crop exposure belong on the remainder, at 52.6%. Averaging them produces a number that describes neither.
2. Name the population, the date and the publisher of every water figure you use. There are at least four reservoir series in circulation in Sri Lanka — the Mahaweli Authority’s national total, its power-station sub-total, the Irrigation Department’s major-reservoir figure and its all-scheme figure. Domestic coverage this month has been 34%, 45%, and 47.3%, alongside the 60% above, and they are not inconsistent. They count different reservoirs.
3. Put the 15 to 25 October window in your own operating calendar, not just your weather-watching. The date is fixed and is fifteen days out. Procurement, working-capital, and logistics decisions that assume a normal Maha should be tested against a scenario in which the dry-zone tanks enter the window near half-full.
4. Ask, in writing, for the Maha 2026/27 seasonal operating plan. Agricultural lenders, insurers, millers and plantation companies all have standing to request it from the Mahaweli Authority and the Department of Irrigation. It is the document that converts a storage figure into an allocation, and it is the one piece of this picture that is not currently public.
ESGNexus has asked the Mahaweli Authority of Sri Lanka whether a Maha 2026/27 seasonal operating plan has been approved and when it will be published, and will report the reply.
Sources & Further Reading
Mahaweli Authority of Sri Lanka, Water Management Secretariat, “Latest Status of Reservoirs”, sheets dated 2026/09/29 and earlier — mahaweli.gov.lk
Mahaweli Authority of Sri Lanka, Water Management Secretariat, “Seasonal Operating Plan – Yala 2026”, read 30 September 2026 — mahaweli.gov.lk
Department of Meteorology, Seasonal Forecasting Division, “Current condition and forecasts of ENSO and IOD for SON 2026”, reference SF-2026-09-ENSO-IOD, 10 September 2026 — meteo.gov.lk
Department of Agriculture, Maha 2026/27 cultivation advisory, 4 September 2026
Ministry of Agriculture, Livestock, Land and Irrigation, notices of the preliminary cultivation planning discussion at the Govijana Mandiraya and the decisive discussion on the cultivation plan and irrigation water distribution for the 2026 Maha and 2027 Yala seasons at Gannoruwa — agrimin.gov.lk
Newswire, “Rs 5.8bn World Bank-backed project to rehabilitate 296 irrigation tanks”, 1 September 2026 — newswire.lk
Related ESGNexus analysis: ‘Sri Lanka Has Two Reservoir Numbers. The One in the Headline Is Not the One That Generates Electricity’ (31 August 2026); ‘Sri Lanka Has Fixed Its Maha Planting Date’ (15 September 2026); ‘The Met Department Gave Sri Lanka a Seven-Month Forecast. The Coverage Kept Ten Days of It’ (21 September 2026).
| About ESGNexus ESGNexus is Sri Lanka’s independent platform for ESG, CSR, and sustainability intelligence. We track company-level ESG performance, regulatory developments, and sustainability data across Sri Lanka’s listed companies, large unlisted corporates, and state-owned enterprises. All editorial content is independently produced. Sponsored content is clearly labelled. |
Data disclaimer: Information in this article is sourced from publicly available documents. ESGNexus does not independently verify company disclosures. Errors and omissions excepted.