Cook Inlet Gas Watch
A daily numeric record of Southcentral Alaska's gas position. Measured storage, modeled demand, and the supply nobody publishes. Read August 5th, 2026, 1 day on record.
What you are looking at
The reserve
Southcentral Alaska keeps gas in an underground field near Kenai, run by CINGSA. Utilities fill it in summer and draw on it in winter. The level is published once a day and never archived, so this page reads it and keeps the history.
The demand
How much gas the region burns is not published at all. It tracks how cold it is, so we model it from the Anchorage forecast and show the formula and its errors rather than asking anyone to take it on faith.
The gap
Subtract what came out of storage from what the region likely burned, and what is left came from somewhere nobody reports daily. That residual is the number no other source publishes.
This page will never tell you whether the lights stay on
It publishes what is measured, what is modeled, and what is missing. It does not publish a verdict. A compressor failure or a sanded well can produce curtailment on a day these numbers looked comfortable, and supply side deliverability is not public, so no adequacy conclusion can honestly be drawn from what is here. Anyone using this data to say the region is fine, or that it is not, is using it wrong.
Day by day
The record holds 1 day of readings so far, which is not yet a trend, so there is no chart here. The table below is the whole series. A daily plot appears here once there is more than one verified day to draw.
| Date | Storage Bcf | Percent of design | Modeled peak MMcf/d | Non CINGSA supply MMcf/d |
|---|---|---|---|---|
| 2026-08-05 | 6.46 | 49.7 | 103 |
The model, in full
Regional demand in MMcf per day is 75.0 plus 4.2 times heating degree days on a base of 65 degrees Fahrenheit. Version 1.2. It is calibrated to two published figures, a design day of 390 MMcf per day at 75 degree days and an average day of 190 MMcf per day. It is NOT fitted to observed sendout, because observed sendout is not public. Treat the coefficients as a working hypothesis, not a measurement.
Every figure below is recomputed at build time from 4,599 days of observed Anchorage degree days covering January 1st, 2014 to August 4th, 2026. Nothing on this page is a number somebody typed.
The model evaluated at the published design day temperature. This anchor is exact by construction, since the coefficients were fitted to it. The model returns 390.0 MMcf/d.
The coldest day in the committed Anchorage record, and what the model asks of the system on that day. Recomputed from the HDD history, never typed. The model returns 398.4 MMcf/d.
The 2025 to 2026 gas year integrated day by day from the committed record. Season bounds are July 1st through June 30th. The model returns 71.7 Bcf.
Mean modeled demand across every day in the committed record, against the published average day figure. The model returns 183.7 MMcf/d.
Across that record the model averages 183.7 MMcf per day against the published average day of 190, so it runs slightly light. The coldest day in the record is February 2nd, 2024 at 77 degree days, which models to 398.4 MMcf per day. In the whole record, 3 days model at or above the published design day. That is a fact about the weather, not a statement about whether the system coped.
How this gets better
Nothing here learns on its own. The formula is two numbers fitted to two published figures, and it moves only when a person moves it, as a data change with the reason recorded. Version 1.2, 2 revisions on record. No check has resolved yet, since the first needs a forecast made yesterday and a temperature observed today.
One gap no amount of collecting closes. Utility sendout is not published, so this can never be fitted directly to what the region actually burned.
Checked against what Alaska burned
The US Energy Information Administration publishes Alaska gas deliveries and underground storage monthly. Across 131 months of overlap this model runs 8 percent high against residential, commercial and electric power deliveries. That check is statewide and lags about two months, so it corrects the model rather than replacing anything here.
It also widens the picture. Through May 2026 Alaska held 30.0 Bcf of working gas across 5 storage fields, against 69.9 Bcf of capacity. CINGSA's design volume is 13.0 Bcf, so the field read here daily holds a minority of the state's stored gas.
What is not reported daily
These are the things no public feed reports daily, and their absence is why this page publishes numbers rather than conclusions. Two of them do have monthly statewide figures, which is where the check above comes from. What no source gives is a daily regional number, and that is the gap that matters here.
hilcorp storage inventory
daily field deliverability
enstar realtime sendout
How to cite it
The whole series is one JSON document at /gas-watch.json, licensed CC BY 4.0 (licence text). It carries the schema version, the model block with its full history, and one object per day with the provenance of every external fetch behind it. Storage is measured, demand is modeled, and each record says which is which. Version 1.0. The related tracked decision is Enstar's Cook Inlet gas storage plan.