Virgin Islands · updated daily

Water Watch

Hundreds of cisterns across the territory report their water level continuously. Added up, they measure something no weather station does: how much rainwater households are actually catching, and how much is left in the tanks.

Capture vs last month

-3%

20.1 against 20.7

Since significant rain

7 days

last on October 17, 2026

Below a quarter full

16.1%

of 635 cisterns

How much is being caught

Gallons entering cisterns for every 1,000 gallons of tank we monitor. It is measured per unit of capacity rather than as a raw total because the number of cisterns we monitor has grown steeply — a raw total would climb even while every individual household caught less.

All water entering cisternsExcluding identified deliveries
020406027 Jul11 Aug26 Aug10 Sep25 Sep10 Oct24 Oct19.16.7

The lower line removes gains that look like a delivered tanker — a large volume arriving in an hour or two. The gap between the lines is roughly how much of the territory's water is arriving by truck rather than off a roof.

Table of these values
DayAll water entering cisternsExcluding identified deliveries
2026-07-2721.19.3
2026-07-2820.510.3
2026-07-2927.911.8
2026-07-3025.29.0
2026-07-3138.511.3
2026-08-0129.515.9
2026-08-0211.67.0
2026-08-0322.214.1
2026-08-0422.313.8
2026-08-0521.913.7
2026-08-0616.511.1
2026-08-0716.59.3
2026-08-0817.310.7
2026-08-0917.310.4
2026-08-1020.110.1
2026-08-1117.913.6
2026-08-1220.413.4
2026-08-1310.25.9
2026-08-1412.17.4
2026-08-1517.79.3
2026-08-1612.67.1
2026-08-171810.0
2026-08-1814.66.4
2026-08-1915.67.4
2026-08-2018.48.0
2026-08-2119.88.3
2026-08-2219.39.7
2026-08-2314.38.2
2026-08-2420.89.8
2026-08-25239.2
2026-08-2617.310.2
2026-08-2719.111.1
2026-08-2818.29.2
2026-08-2917.47.7
2026-08-3019.414.4
2026-08-3117.99.9
2026-09-0129.215.9
2026-09-0237.623.6
2026-09-0323.69.3
2026-09-0420.49.1
2026-09-0517.810.1
2026-09-0618.211.6
2026-09-0732.823.0
2026-09-0827.816.8
2026-09-0932.218.6
2026-09-1014.08.7
2026-09-1115.36.4
2026-09-1213.46.2
2026-09-1322.014.9
2026-09-1429.421.5
2026-09-1517.98.9
2026-09-1613.85.6
2026-09-1715.27.9
2026-09-1818.28.9
2026-09-1915.45.7
2026-09-2014.66.3
2026-09-2115.87.7
2026-09-2227.212.0
2026-09-2325.311.5
2026-09-2418.17.7
2026-09-2526.113.3
2026-09-2618.99.0
2026-09-2711.37.5
2026-09-2815.19.5
2026-09-2918.611.7
2026-09-3016.77.5
2026-10-0114.18.4
2026-10-0213.57.8
2026-10-0332.822.7
2026-10-0412.06.1
2026-10-0515.88.7
2026-10-0635.220.0
2026-10-0741.431.0
2026-10-0816.95.8
2026-10-0924.214.3
2026-10-1022.816.2
2026-10-1120.611.3
2026-10-1210.94.7
2026-10-1318.813.1
2026-10-1418.17.8
2026-10-1517.610.0
2026-10-1611.85.3
2026-10-1721.611.3
2026-10-1823.111.1
2026-10-1920.08.4
2026-10-2018.17.5
2026-10-2116.69.6
2026-10-2215.26.3
2026-10-2321.77.9
2026-10-2419.16.7

Rain we have caught

On October 17, 2026 enough cisterns filled at once for us to call it rain rather than deliveries. This is what it put in the tanks.

Caught across the fleet

39,460 gal

across 818 monitored cisterns

Over

1 hour

hours where the fleet filled together

Per cistern

48 gal

averaged over every monitored tank

Counted only during the hours the rain was actually falling across the territory, so it excludes the trucks and pumps that top up tanks around the clock. It is what the monitored fleet caught, not what fell — a tank with a bigger roof catches more of the same shower, and only a fraction of the territory’s cisterns are monitored.

Every rain we have measured

14 separate spells of rain have been heavy enough to register across the fleet, averaging 183 gallons into each monitored cistern.

Spells of rain measured across the monitored cistern fleet
DatePer cisternTotal caughtHoursCisterns
October 17, 2026latest48 gal39,460 gal1818
October 13, 202639 gal32,614 gal1838
October 9, 2026 – October 11, 2026413 gal351,486 gal8849
October 9, 2026137 gal116,836 gal3853
October 10, 2026182 gal154,939 gal4852
October 11, 202694 gal79,711 gal1849
October 6, 2026 – October 7, 2026407 gal347,748 gal5856
October 6, 202685 gal72,093 gal1845
October 7, 2026322 gal275,655 gal4856
September 29, 2026116 gal96,959 gal2836
September 26, 202644 gal36,256 gal1822
September 14, 2026 – September 15, 202698 gal79,852 gal2818
September 14, 202649 gal39,943 gal1813
September 15, 202649 gal39,909 gal1818
September 9, 2026181 gal150,039 gal3828
September 7, 2026254 gal208,221 gal4819
August 31, 2026 – September 2, 2026362 gal293,200 gal5815
August 31, 202634 gal27,362 gal1803
September 1, 2026169 gal136,575 gal2809
September 2, 2026159 gal129,263 gal2815
August 27, 2026 – August 28, 202678 gal60,573 gal2788
August 27, 202635 gal27,013 gal1780
August 28, 202643 gal33,560 gal1788
August 11, 2026 – August 12, 2026132 gal95,667 gal2734
August 11, 202665 gal46,176 gal1709
August 12, 202667 gal49,491 gal1734
August 1, 202678 gal54,170 gal1693
July 21, 2026305 gal153,899 gal2505

Hours counts the hours the fleet was filling together, not how long it rained. Rain on consecutive days is treated as one spell and totalled, since a storm crossing midnight is one weather system; that is a rule of thumb rather than meteorology, so a multi-day spell lists its days beneath it. The per-cistern figure divides by every monitored tank, including those the rain missed, so an individual cistern under the shower caught more.

What the last rain put back

The question a rain spell actually raises is whether it changed the situation. Comparing the fleet the day before it began with its best day in the 48 hours after it ended:

Median cistern

49% full

up from 49% before the rain

Below a quarter full

15%

was 14% of monitored cisterns

Below a tenth full

2%

was 2%

The 48-hour window is our choice, not a measurement: catchment keeps draining into tanks after the rain stops, and household use starts pulling the median back down soon after that. Across the fleet the spell delivered less than a day of ordinary household use per monitored cistern. A rise here means tanks refilled, not that the drought ended — one spell does not undo a season, and the capture chart above is the longer story.

What a household gets through

A typical occupied household uses about 140 gallons a day, measured across 500 properties. Properties that drew no water at all over the month — empty houses and seasonal lets — are left out, since they are not households in use. Counted per property rather than per tank: many homes here keep two or three cisterns and switch the house feed between them, so a per-tank figure would describe plumbing rather than a household.

Measured over a rolling month rather than a single day: cistern draw is bursty, and a gauge on a large tank cannot resolve one day’s use at all, so day-by-day figures read far too low. A median, not an average. The spread is wide — guest properties, irrigation and the occasional leak pull the average to several times this — so the middle of the range describes an ordinary household far better than the mean does.

How full the cisterns are

Levels have held up far better than rainfall has. That gap is being paid for with trucked water, so steady tanks are not the reassurance they look like.

Median cisternShare below a quarter full
025507510027 Jul11 Aug26 Aug10 Sep25 Sep10 Oct24 Oct48.8%16.1%
Table of these values
DayMedian cisternShare below a quarter full
2026-07-2742.3%19.7%
2026-07-2842.4%19.3%
2026-07-2942.4%18.2%
2026-07-3042.3%18%
2026-07-3140.9%21.1%
2026-08-0143%18.6%
2026-08-0243.2%19.4%
2026-08-0342.3%20.3%
2026-08-0442%21%
2026-08-0543.1%20.4%
2026-08-0642.4%21.3%
2026-08-0742.4%20.8%
2026-08-0842.1%21.8%
2026-08-0940.9%22.8%
2026-08-1041.2%23.5%
2026-08-1141.3%23.4%
2026-08-1241.4%21.9%
2026-08-1341.7%22.6%
2026-08-1441.5%22.8%
2026-08-1540.9%22.8%
2026-08-1640.8%24.3%
2026-08-1740.4%24.4%
2026-08-1840.5%24.2%
2026-08-1941.2%25.1%
2026-08-2040%26.1%
2026-08-2139.9%26%
2026-08-2240.2%25.4%
2026-08-2339.4%26.3%
2026-08-2439.7%25.8%
2026-08-2539.8%24.9%
2026-08-2639.7%24.7%
2026-08-2740.6%24.6%
2026-08-2840.2%24.5%
2026-08-2940.6%23.8%
2026-08-3040.7%23.8%
2026-08-3140.2%23.8%
2026-09-0141.1%22.1%
2026-09-0242.6%21.1%
2026-09-0342.6%20.7%
2026-09-0441.9%21.5%
2026-09-0541.2%22.6%
2026-09-0641.3%22.3%
2026-09-0741.9%21.1%
2026-09-0843.6%19.1%
2026-09-0944.3%18%
2026-09-1044.2%18.2%
2026-09-1143.5%18.2%
2026-09-1243.1%18.3%
2026-09-1343.2%18.2%
2026-09-1443%17.6%
2026-09-1543.7%17.6%
2026-09-1643.5%17.6%
2026-09-1743.2%18.2%
2026-09-1843.4%18.7%
2026-09-1943.7%18.7%
2026-09-2043.5%19.6%
2026-09-2143.3%19.7%
2026-09-2242.8%20.4%
2026-09-2342.2%20.4%
2026-09-2442.6%19.9%
2026-09-2543.2%19.7%
2026-09-2643.7%19.7%
2026-09-2742.9%20.7%
2026-09-2841.9%21.3%
2026-09-2942.6%21.5%
2026-09-3042.2%20.7%
2026-10-0142.1%20.6%
2026-10-0242.9%20.6%
2026-10-0343.4%19.8%
2026-10-0443.6%20.2%
2026-10-0543.2%20.2%
2026-10-0643.2%19.5%
2026-10-0745.4%17.4%
2026-10-0846.4%16.4%
2026-10-0947.1%15.7%
2026-10-1049.2%14.8%
2026-10-1149.7%14.6%
2026-10-1250%14.9%
2026-10-1350.3%14%
2026-10-1450.1%14.5%
2026-10-1551.1%13.9%
2026-10-1649.2%14%
2026-10-1749.3%14.5%
2026-10-1849.2%15.6%
2026-10-1948.2%15.7%
2026-10-2048.5%15%
2026-10-2148.1%15.5%
2026-10-2247%15.4%
2026-10-2349%14.8%
2026-10-2448.8%16.1%

What actually helps, if you have a cistern

Most of the water a household loses in a drought is not water it chose to use. It leaves through a fitting nobody has looked at in years, or it never reaches the tank in the first place. These are the checks worth doing this week, roughly in the order they pay off.

Read the tank when nobody is using water.
Note the level last thing at night and again before anyone is up. If it is lower by morning, the cistern is losing water — and it is losing it all day too, where ordinary household draw hides it. This single check is what separates “we use a lot” from “we have a leak”, and it costs nothing but two glances.
The most expensive leak makes no noise.
A toilet flapper that no longer seats passes roughly 30 gallons a day in complete silence — around 21% of everything a typical household here gets through. Put a few drops of food colouring in the tank, leave it fifteen minutes, then look in the bowl. Colour in the bowl means the flapper is passing, and it has probably been passing for months.
Listen to the pump.
A pressure pump that starts up while every tap in the house is closed is replacing water escaping somewhere past it. Unlike a dripping tap, that somewhere is usually under a slab, behind a wall, or in a buried line — which is exactly why it runs so long before anyone finds it.
Fix the catchment before the rain, not after.
Only 14 spells of rain in our whole record have been heavy enough to register across the territory. When rain is scarce, a blocked gutter or a first-flush diverter stuck shut does not cost you a little water — it costs you a whole event. Clear the gutters and downpipes, check the screens, and confirm the diverter actually drains and resets between showers.
Know your own number.
About 140 gallons a day is the middle of the monitored fleet. A household well above that, with the same number of people, is usually looking at irrigation, a pool or a cistern top-up, or a leak — in that order of size, and in the reverse order of how visible they are.

What is at stake is easy to underrate. The last significant rain put about 48 gallons into the average monitored cistern, less than a day of ordinary household use. A silent 30-gallon-a-day leak gets through the same volume in 2 days, and it does that whether or not it rains.

A leak is only cheap to fix while it is still small, and the ones that matter are the ones nobody can see. Continuous monitoring is how they get caught — it watches the level around the clock and raises an alert when a tank falls in a way normal use does not explain. How leak detection works.

How to read these numbers

This measures water captured, not rainfall.
A cistern records what its roof delivered into it. Roof area, gutter condition and first-flush diverters all vary between properties, so this is a measure of harvested water — closely tied to rainfall, but not the same quantity as inches on a gauge.
The decline is at least this steep, not exactly this steep.
The headline compares all water entering cisterns, trucked deliveries included. Deliveries rise during a drought, which holds that number up. We separate the ones we can identify by their signature — a large volume arriving within an hour or two — but the separation is a rule of thumb, not a receipt.
Our record is short, so we make no historical claim.
This series is a little over three months long, and the monitored fleet grew several-fold inside that window. We cannot compare this drought to last year or to any past dry season, and nothing here should be read as “the worst since” anything.
Monitored cisterns are not a random sample.
They belong to households and businesses that chose to install monitoring. That is not a cross-section of the territory, and the fleet is concentrated wherever our customers are rather than spread evenly across it.

Figures are aggregated across the whole monitored fleet and run through October 24, 2026. No individual property, customer or location is identified here, and nothing on this page can be traced back to one household.

Journalists, utilities and emergency management are welcome to use these figures with attribution to Your Ideas VI. If you need a cut of the data we have not published, ask us — we would rather help than have the numbers quoted loosely.