How much working natural gas will be in storage in the US Lower 48 for the week ending 4 December 2026, in billion cubic feet?

TL;DR

The median forecast for Lower 48 working natural gas storage for the week ending December 4, 2026 is 3,684 Bcf. The 10th percentile is 3,552 Bcf, and the 90th percentile is 3,822 Bcf. Warm-weather guidance lifts the outlook above the mid-October starting inventory, while late-November cold and changes in production or exports account for the broad uncertainty range.

Context

The forecast is dated October 22, 2026. The latest reported inventory is 3,644 Bcf for October 16, following a 77 Bcf weekly injection. That places stocks 35 Bcf above the five-year comparison average of 3,609 Bcf, but 152 Bcf below the report’s year-earlier comparison of 3,796 Bcf (EIA weekly report). The starting position is neither an unusually large surplus nor a clear national shortage.

The target is the inventory observation for December 4, not inventory on the scheduled December 10 publication date. From October 16, the relevant physical balance covers seven weeks. Resolution uses EIA’s first-published Total Lower 48 figure; later revisions do not count. The reported total also matters more than a sum of regional figures, because independently rounded components need not add exactly to it (EIA notes and definitions).

Evidence

The historical backbone is simple: forecast the change from the known starting inventory, rather than extrapolate October injections through the winter transition. The supplied history contains 877 consecutive Friday observations from January 1, 2010 through October 16, 2026. Its successive inventory differences match the accompanying weekly changes. It is useful long-run evidence, but it is a compiled historical vintage rather than a complete archive of first releases.

For each completed year, the corresponding seven-week window starts on the Friday nearest October 16. The full history of those changes is below. These are calculations from the supplied storage history, not EIA forecasts.

Year	Seven-week storage change, Bcf
2010	+34
2011	+206
2012	−37
2013	−208
2014	−34
2015	+67
2016	+117
2017	+50
2018	−181
2019	−88
2020	−78
2021	+44
2022	+120
2023	+93
2024	−38
2025	−63

The mean change is +0.25 Bcf and the median is zero. The sample standard deviation is 111.67 Bcf. The central lesson is that injections early in the window and withdrawals later in it largely cancel. The latest ten windows also average close to zero. There is no strong historical case for a large seasonal increase or decrease before considering current conditions.

Calendar alignment changes the apparent uncertainty much more than it changes that central result. A historical Friday can fall several days before or after October 16. Shifting both endpoints later replaces relatively injection-friendly October days with withdrawal-heavy December days. A calendar correction that does not require interpolation supports approximately 83 Bcf of baseline predictive standard deviation. Much of the raw dispersion therefore reflects date placement, not just different market balances. This does not justify a narrow final forecast: weather, operations, and first-publication uncertainty remain.

Recent injections provide less predictive value than the calendar correction. The latest four-week build is 293 Bcf, and the latest eight-week build is 460 Bcf. Tests across the small annual sample do not show a reliable enough relationship to extrapolate that pace aggressively. The forecast therefore allows only a −10 Bcf balance effect. Recent weakness is a reason for restraint, not a reason to override the seasonal baseline.

Weather is the strongest reason to expect storage above that baseline. The observed gas-weighted weather data show 43 heating degree days during October 17–20, against 42 normal. The early part of the period was therefore approximately normal, rather than an already-realized large warm anomaly (NOAA observed degree days).

The stronger signal lies ahead. The forecast issued October 22 gives 61 cumulative gas-weighted heating degree days for October 22–28, against an 85-degree-day normal. Fewer heating degree days imply less heating demand and, other things equal, more gas left in storage (NOAA seven-day forecast).

The broader outlook supports the same direction. The October 22 discussion favors eastern warmth for October 28–November 1 and warmth across the eastern half and much of the Plains for October 30–November 5 (CPC short-range discussion). The October 16 outlook extends a warm tilt across northern and eastern areas through October 31–November 13 (CPC weeks 3–4 outlook). The October 15 monthly outlook favors warmth across much of the northern two-thirds during November, while warning about short-term fluctuations (CPC November discussion).

These products overlap and share information. They support one coherent warm-weather signal, not several independent confirmations. The forecast allows a judgmental +50 Bcf weather effect relative to the historical storage baseline. That is an interpretation of the guidance, not a gas-volume forecast issued by NOAA. It also accounts for the fact that the historical baseline already contains warmer autumn weather than the older daily normal.

Supply and export evidence partly offset the warmth. EIA’s October outlook projects fourth-quarter dry production of 113.8 Bcf/day, compared with 109.7 Bcf/day in the preceding year’s fourth quarter. It also projects higher LNG exports, rising from 17.4 to 18.2 Bcf/day, alongside other changes in consumption and trade. Production growth is not all available for storage, and LNG growth is not the whole balance (October STEO).

Current operational evidence is mixed too. EIA’s supplement reports recovering production alongside higher deliveries to LNG terminals (weekly supplement). Regional stocks add a caution: South Central storage is 40 Bcf below its five-year average, including a 44 Bcf deficit in salt storage, despite the national surplus. That signals local tightness, but historical tests do not support turning it mechanically into a large national reduction (EIA storage report).

Project timing also matters. The final Corpus Christi Stage 3 train reached substantial completion on August 28, before the starting inventory observation (Cheniere announcement). Plaquemines’ October filing targets October 31 commercial operation for Phase 1 but reports commissioning exports beginning in December 2024. Commercial commencement does not mean all demand appears for the first time that day (DOE progress report). Likewise, announced pipeline capacity is not the same as incremental national production.

EIA’s monthly inventory outlook provides a consistency check, not an additional adjustment. Its inventory figures include Alaska and describe month-end stocks, while this question concerns a Lower 48 Friday. Alaska must be removed, and early-December withdrawals need not equal a uniform share of the full month’s withdrawals. After those corrections, the monthly outlook supports the broad forecast level without supplying a precise weekly answer (October STEO).

What's non-obvious

The most easily missed issue is the weather reference period. A warm forecast against an older normal is not entirely new information when the storage baseline comes from more recent autumns. The historical weather analysis found about 283 heating degree days for October 22–November 11, compared with 303 in the older normal. Ignoring that difference would overstate the incremental storage benefit of current warmth. The forecast keeps a positive weather adjustment, but measures it against the weather embedded in the historical balance.

The other trap is to confuse a winter narrative with this specific endpoint. October injections do not continue unchanged into December. New LNG facilities do not create their full demand on a commercial-operation date. More pipeline capacity does not guarantee more production. The decisive quantity is the cumulative net balance through December 4, not any one headline about winter, supply growth, or exports.

Uncertainties

Late-November weather is the largest unresolved variable. The strongest guidance covers the earlier part of the window. The longer-lead warm tilt does not establish an uninterrupted warm path through December 4. A persistent Midwest or eastern cold outbreak would increase heating demand and deepen withdrawals. Continued warmth would move the outcome in the opposite direction.

Operational balances are the next gap. Production recovery, LNG utilization, and pipeline restrictions can each change storage by meaningful amounts over several weeks. Announced capacity and modeled flow estimates do not establish sustained realized supply or demand. Verified daily flows and evidence that recent balance changes persist would help distinguish a temporary fluctuation from a durable shift.

The historical sample contains only 16 annual windows. It is too small to establish precise relationships among inventories, recent injections, regional stocks, and subsequent withdrawals. Calendar correction improves comparability, but does not remove structural changes in the gas market or uncertainty in translating weather into demand.

Finally, the forecast resolves to a first publication, while historical estimates can be revised. There is no established directional revision effect to apply here (EIA revision policy). These limitations support retaining the saved forecast’s broad range rather than treating the warm outlook as a reason for high precision.