İstanbul’s dams are well over half full.

61.9%Monday 9 November 2026, 09:10Measured 11 hours ago. Source: İSKİ.

That is about 537 million cubic metres of water.

About 61 mm of rain fell on the dam catchments over 27 October – 3 November. A week ago the dams were 53.7% full.

We call 60–100% “safe”.

bottom 10%0%25%50%75%100%2007, 20142023, 202514 days69–81%30 days71–96%10% of capacity: water at the bottom0%25%50%75%100%20072014, 2023, 202514 days69–81%30 days71–96%
Height: share of total capacity. The red ▽ is today’s level, and the dotted lines tie each range to today’s surface. The triangles on the wall are the lowest points of past droughts; the filled one is the lowest day on record.
In 14 days 23 November
69–81% most likely 74%
In 30 days 9 December
71–96% most likely 83%

There’s an 80% chance the level stays inside each range.

Over the next month the water will most likely rise.

How well these forecasts hold up

The most likely level a month ahead has moved up 2.6 points since yesterday morning (80%, now 83%). Main reason: the updated rain forecast.

8 forecasts have a result, and 7 landed inside the range. With so few, we don’t give a hit rate. In most of the 120 short runs we made in recent days, the model put the level higher than it turned out. Our value for a month ahead may be high too.

Hindcast: We re-ran 2011–2023, each time with only the data known on that day. The 30-day forecast was off by 3.7 points on average, and 82% of outcomes landed inside the range (the target is 80%).

The full scorecard

We make the forecast again every morning.

The live archive started on 19 October 2026. It holds 22 mornings’ forecasts so far. 7 days ago the most likely level a month out was 75%. In today’s forecast it is 83%. The forecast has moved up 8 points.

0%25%50%75%100%measured52–81%71–96%18 Oct25 Oct1 Nov8 Nov0%25%50%75%100%measured52–81%71–96%18 Oct25 Oct1 Nov8 Nov
Each upright line is one morning’s forecast: the range we gave for 30 days later, with a short tick at the most likely level. The dark area is the measured level. Once a forecast’s date arrives, what happened is added as a filled triangle.

Every forecast and its result, on the Scorecard page

How rain shapes the water forecast

Catchment rainfall and reservoir fill, on the same dates. Some rain never reaches the lakes; the level depends on all the water flowing in and out.

About 57 mm of rain is expected over the catchments in the next 14 days (38–86 mm). In the same days of the previous 26 years the average was 30 mm; the expected amount is above that.

The first 14 days use the ECMWF weather forecast. After that, the model tries weather from the same season in past years. A 30-day water forecast is therefore not a 30-day weather forecast.

The soil holds some rain.
The rest runs into the lake. Dry ground absorbs more water. The same rainfall can bring more water to the dams when the ground is already wet; its effect can take time.
Rain is only part of the balance.
City use and losses lower the level; transfers from Melen and Yeşilçay raise it. Temperature affects use. Evaporation is modelled together with other losses.
Weather forecastPast years’ weatherCumulative catchment rain050100 mmForecast38–86 mmPast average30 mmReservoir fill0%50%100%69–81%71–96%61.9%15 Nov22 Nov29 Nov6 DecWeather forecastPast years’ weatherCumulative catchment rain050100 mmForecast38–86 mmPast average30 mmReservoir fill0%25%50%75%100%69–81%71–96%Today 61.9%15 Nov22 Nov29 Nov6 Dec
The area shaded with fine lines, inside the dashed outline, is the forecast range (10th–90th percentile); the thin line in it is the median. The inked line above is the average rain for the same dates in past years. Rain is in millimetres, fill in percent: the two panels share dates, not scales. The fill range is widened using past forecast errors. Water forecast issued: 9 November 2026. Weather forecast dated: 9 November 2026.
Daily chart values
DateRain range (mm)Past average (mm)Fill range
9 November0–0 (median 0)1–
10 November4–7 (median 5)363–63% (median 62.7%)
11 November5–8 (median 6)663–64% (median 63.2%)
12 November5–10 (median 6)863–65% (median 63.6%)
13 November5–11 (median 7)1163–65% (median 63.9%)
14 November7–29 (median 15)1263–67% (median 64.6%)
15 November23–43 (median 33)1465–69% (median 66.4%)
16 November24–45 (median 33)1666–71% (median 67.8%)
17 November24–45 (median 34)1766–72% (median 68.8%)
18 November25–45 (median 35)2066–73% (median 69.3%)
19 November30–53 (median 38)2367–74% (median 69.8%)
20 November32–59 (median 43)2668–75% (median 70.5%)
21 November33–67 (median 48)2868–76% (median 71.5%)
22 November38–86 (median 57)3068–78% (median 72.8%)
23 November––69–81% (median 73.8%)
24 November––69–82% (median 74.6%)
25 November––69–84% (median 75.3%)
26 November––69–85% (median 75.9%)
27 November––70–86% (median 76.5%)
28 November––70–87% (median 77.1%)
29 November––70–88% (median 78.0%)
30 November––70–89% (median 78.7%)
1 December––70–90% (median 79.3%)
2 December––70–91% (median 79.7%)
3 December––71–92% (median 80.1%)
4 December––71–93% (median 80.6%)
5 December––70–93% (median 81.0%)
6 December––70–94% (median 81.5%)
7 December––71–94% (median 81.8%)
8 December––71–95% (median 82.3%)
9 December––71–96% (median 82.6%)

Rainfall accuracy and water-forecast accuracy are different. The Scorecard checks predicted fill against measurements; it does not give a separate accuracy score for the weather forecast.

The calculation and its limits

The water forecast scorecard

The city uses about 3.2 million cubic metres of water a day.

That’s enough to fill the Basilica Cistern 40 times over, every day.

In the past week the water in the dams rose by about 71.2 million cubic metres. That is what the city uses in 22.5 days.

The dams hold 868 million cubic metres when full. With no rain and nothing from the pipelines, the fill level would drop by about 0.4 percentage points a day.

İstanbul is not on a big river, so it stores rain. That is what cisterns like the Basilica Cistern and the Ottoman dams in the Belgrade Forest were built for.

one day of usemagnified 15×0%25%50%75%100%2007, 20142023, 202561.9%9 Novemberone day of usemagnified 10×0%25%50%75%100%20072014, 2023, 202561.9%9 November

Daily use is an approximate value read off İSKİ’s yearly charts. The Basilica Cistern holds about 80,000 m³ (yerebatan.com).

On 5 November 2007 the dams were down to just 80 million cubic metres.

It is the lowest day on record. The dams were 9.2% full. The level fell until this day, then rose again with the autumn rain.

Today the dams hold about 6.7 times as much.

The same drought ran on into 2008.

Some of the water at the bottom of a dam never reaches the pipes. İSKİ doesn’t say how much. A rule of thumb often quoted in the press is 10% of capacity. If İSKİ’s percentages already leave this water out, that share should not be counted. Until İSKİ tells us, we show both.

bottom 10%0%25%50%75%100%2007, 20142023, 202561.9%9 November10% of capacity: water at the bottom0%25%50%75%100%20072014, 2023, 202561.9%9 November

On this date the dams are fuller than in most years.

The record has readings for 9 November in 25 years. In 21 of them the dams held less water than today, and in 4 they held more. The last year they were fuller on this date was 2015 (67.8%).

There is no reading for this date in 2024, so that year is left out of the comparison.

Pick a year. The water in the drawing moves to that year’s level on this date, and the red mark stays at today’s level.

  • 2024: no reading

Average 46.5%. Lowest 2007 (9.5%), highest 2009 (87.8%).

Water reaches the dams two ways and leaves them two ways.

Rain and the pipelines from the Melen and Yeşilçay rivers fill the lakes. The city’s taps and evaporation drain them. Rain is the least certain part of the sum.

0%25%50%75%100%RainEvaporationPipelinesThe city2007, 20142023, 202561.9%9 November0%25%50%75%100%Rain27 October – 3 November: about 61 mmEvaporationPipelines, 1.4 million m³ a dayThe city, 3.2 million m³ a day20072014, 2023, 202561.9%9 November
Rain
Dry soil keeps some of it. The rest runs into the lakes. 27 October – 3 November: about 61 mm.
Pipelines
About 1.4 million m³ a day. İSKİ decides how much.
The city
About 3.2 million m³ a day.
Evaporation
Water lost off the lake surface. More in summer.

Daily amounts are approximate, derived from the yearly totals İSKİ publishes.

Since 2000 the dams have fallen below 20% in 6 separate years.

Those years were 2007, 2008, 2014, 2021, 2023 and 2025. There are no readings between 2024 and 2026, so that part of the chart is empty.

0%25%50%75%100%2007 9.2%200520102015202020250%25%50%75%100%no readings2007 9.2%2014 14.5%2021 19.2%2023 16.2%2025 17.1%2002200420062008201020122014201620182020202220242026
How full İstanbul’s dams were, 23 October 2000 to 9 November 2026. The last reading of each week, plus every year’s lowest and highest day. 2000–2024: İBB Open Data. After that, İSKİ data via DSİ and dolulukorani.com.

Every year the dams fill in winter and drain in summer.

The lakes are usually at their lowest in November and their highest in April. That is why the year here starts on 1 October, as it does in the State Hydraulic Works’ (DSİ) accounts.

This year is 13.4 points above a typical year right now. In a typical year the dams are 48.5% full on this date.

0%25%50%75%100%typical yearrain, 27 October – 3 Novemberthis yearOctJanAprJul0%25%50%75%100%typical yearrain, 27 October – 3 Novemberthis yearOctNovDecJanFebMarAprMayJunJulAugSep
The grey area runs from the lowest to the highest fill seen in each week across 23 years, 2000–2023. The dashed line is a typical year, the dark line this year.

Beyond a month, it depends on the weather.

Nobody can see the rain more than two weeks ahead, so the range widens the further out it goes. We draw it so there’s an 80% chance the level stays inside.

In past years the dams usually rose over the month after this date.

From day 30 to day 90 we replay past years’ weather over today’s lakes: if the coming months turn out dry like 2024 or wet like 2018, how much water is left? We sort the results by rain into dry, average and wet. Nothing past day 90 is drawn.

0%25%50%75%100%MeasuredForecastScenario69–81%71–96%60 days90 daysWetAverageDryrain61.9%1 Nov1 Dec1 Jan1 Feb0%25%50%75%100%MeasuredForecastScenario69–81%71–96%60 days90 daysWet87–100%Average86–100%Dry80–100%rain, 27 October – 3 NovemberToday 61.9%1 Sept1 Oct1 Nov1 Dec1 Jan1 Feb
The solid dark area is measured water. The area shaded with fine lines, inside the dashed outline, is the 30-day forecast range, and the thin line in it the most likely value. After day 30 the lines are scenarios, and the range bars at their ends show each scenario’s range. Forecast issued on the morning of 9 November 2026, starting from the reading of 9 November 2026.
Dry
Day 90: 80–100%The 9 years with the least rain in this season: 2000, 2006, 2007, 2008, 2011, 2013, 2020, 2022, 2024.
Average
Day 90: 86–100%The 9 years with rain closest to average: 2002, 2003, 2004, 2015, 2017, 2019, 2021, 2023, 2025.
Wet
Day 90: 87–100%The 8 years with the most rain in this season: 2001, 2005, 2009, 2010, 2012, 2014, 2016, 2018.

Scenarios worked out in autumn show too narrow a range at 60 and 90 days. In the hindcast, 80% of outcomes should have landed inside. At day 60 it was 75%, at day 90 68%. What matters most is when the first big rain arrives.

The chart’s figures as a table
Days aheadForecastDryAverageWet
14 days, 23 November69–81% (74%)–––
30 days, 9 December71–96% (83%)–––
60 days, 8 Januaryscenarios only73–98% (86%)80–100% (90%)89–100% (97%)
90 days, 7 Februaryscenarios only80–100% (92%)86–100% (97%)87–100% (98%)

The scenarios in detail, and the transfer settings

We don’t give a date for when the water runs out.

Once rain and the pipelines are counted, no single date would be honest. That is also why the “days of water left” figures in drought reports rarely agree. We give only a range, with its assumptions beside it.

None of the three scenarios has the dams losing water over the next 90 days, so a “days of water left” count gives no number today.

Decisions such as water cuts, lower pressure or calls to save water are İSKİ’s, and İSKİ announces them. This page doesn’t know those decisions. It shows only the water in the dams and where it is heading.

The assumptions and the days per scenario are on the Scenarios page

Every morning we work the figures out again from the new reading.

Every figure on this page takes the same route. If a value was copied by hand from İSKİ’s page, the page says so.

İSKİ’s page is checked hourly each morning, after its usual publishing time, until the day’s reading is in. If the source has not published new data, the last measurement keeps its original date.

All seven steps, with the state of each today, are on the Methods page

Where these figures come from

Latest reading
Monday 9 November 2026, 09:10
Source
İSKİ
Forecast
morning of 9 November 2026, from the reading of 9 November 2026
Updated
Monday 9 November 2026, 19:48
Model version
2
Files
home.json, narrative.json, outlooks.json, story.json, dams.json, scorecard.json