| Series | First | Latest | Min | Max |
|---|---|---|---|---|
| Overall tightness | -13.45 | -8.56 | -49.22 | 52.42 |
| East | -9.27 | -3.82 | -43.37 | 54.94 |
| Midwest | -9.15 | -5.56 | -62.51 | 68.64 |
| Mountain | -10.67 | -14.42 | -88.64 | 46.40 |
| Pacific | -25.74 | -15.30 | -54.24 | 57.16 |
| South Central | -12.43 | -3.72 | -65.02 | 55.77 |
tightnessPct_i = (fiveYearAvg_i − current_i) / fiveYearAvg_i × 100; composite = equal-weight mean of tightnessPct_i (positive = tight)Weekly working gas in underground storage in each of the five storage regions (East, Midwest, Mountain, Pacific, South Central), expressed as seasonal deviations. Each region's tightness is (five-year same-week average minus current) divided by the five-year average, times 100. Only complete prior years enter the baseline. A positive reading means stocks sit below the seasonal norm (tight). The composite is the equal-weight mean of the regional tightness readings that exist that week.
As of 31 July 2026, the latest reading is -8.56 % deviation. That is up 0.8 pp over the past year and below its long-run median of -1.95 % deviation.
A positive composite means regions hold less natural-gas storage than their seasonal norms on average (tightness). A negative composite means the regions are above their norms (slack). The per-region lines show whether tightness is broad or concentrated. One deeply positive region inside a flat composite is a regional dislocation, not a national one. Regional dislocations move regional basis and pipe economics before they move national prices.
Regions hold very different volumes. Raw barrels cannot be compared across regions. A draw in a small region and a draw on the Gulf Coast can both read as the same barrel change with opposite market meaning. Scaling each region's shortfall or surplus by its own seasonal average puts every region on one percentage scale before averaging.
The composite is the mean of the regional tightness scores, not a volume-weighted mean. The smallest region gets the same voice as the largest. The purpose is breadth of tightness (how many regions sit below their norms), not the total barrel shortfall.
The five-year baseline rolls forward. The current year never enters it. A percentage deviation around a tiny average can look extreme on paper. Regional definitions follow storage-reporting conventions, not pipeline geography. Treat the chart as context for regional supply conditions, not investment advice.
Gas storage is regional by construction. The South Central (primarily salt caverns) and the East fill and drain on different schedules and prices differ by region. A winter that drains the East and Midwest faster than the South Central shows up here as a regional divergence that the Lower-48 total smooths away; regional gas desks read exactly that pattern.
As of 31 July 2026, the latest reading is -8.56 % deviation. That is up 0.8 pp over the past year and below its long-run median of -1.95 % deviation.
This benchmark is built on weekly data. The page is refreshed when the source publishes new observations; the freshness block below the chart shows the exact data-through date and when PIER20 last fetched the file.
The chart is built from Overall tightness, East, Midwest, Mountain, Pacific and South Central, sourced from the U.S. Energy Information Administration (EIA).
Zero means the average of the regional tightness scores is zero. The regions are collectively at their five-year seasonal norms. It does not mean every region is at its norm. Offsets can hide inside the average. Read the component lines for coverage.
The national page is the Lower-48 total versus its band. The headline number; This page keeps the five regions separate on a common percentage scale, so a regional imbalance (a cold winter in the East, a salt-cavern build in the South Central) is visible even when the total looks balanced.
The composite averages only the regions with valid scores that week. A region with no baseline or a missing observation drops out. The composite line starts once the first region has a five-year window. The component lines show coverage.
One tested idea, through the evidence stack. No migration of any kind.