Unusually Severe Weather: What Actually Counts
Three of the most widely used contract families in American construction hang weather entitlement on this phrase. Not one of them defines it. That gap is where claims are won and lost — and it is a data problem, not a legal one.
The same test, three names
Whichever paper your job runs on, the weather language asks one question: was this worse than what you should have expected here, at this time of year? The wording changes; the test does not.
| Document | The phrase it uses | Notice deadline |
|---|---|---|
| AIA A201-2017 | "abnormal for the period of time" (§15.1.6.2) | 21 days (§15.1.3) |
| FAR 52.249-10 | "unusually severe weather" | 10 days |
| FAR 52.249-14 | "unusually severe weather" | Per contract terms |
| ConsensusDocs 200 | "adverse weather conditions not reasonably anticipated" (§6.3.1) | 14 days to claim, +21 to document (§8.4) |
Note the notice clocks. They are not the same, and the federal one is the shortest by a wide margin. Ten days on a fixed-price federal job versus twenty-one under A201 is the difference between a claim you still have and a claim you lost while assembling evidence for it.
Why nobody defines it
The omission is deliberate. A numeric threshold written into a national form document would be wrong nearly everywhere it was applied — three inches of February rain is unremarkable in Seattle and a genuine event in Phoenix. So the drafters left the standard comparative and pushed the measurement onto the parties.
The practical consequence: the phrase has no meaning until you supply a baseline. An owner who says "that was a normal spring" and a sub who says "that was the worst spring in a decade" are not disagreeing about the contract. They are disagreeing about a number neither has produced.
The baseline is the whole argument
The customary reference is a 10-year monthly average of daily observations from the nearest official station — long enough to smooth out a freak year, short enough to reflect the current climate. Federal construction contracts often go further and simply hand you the table (see federal contract weather delays), which is why federal weather claims tend to be more arithmetic than argument.
Two mistakes to avoid when building one:
- Do not use the airport 40 miles away if a closer station reports. Opposing counsel will find the closer station, and if it tells a different story, your exhibit becomes the thing being cross-examined.
- Do not use a monthly rainfall total as the baseline. Contracts extend time for lost days. Rainfall is the input; unworkable days are the unit of entitlement. See below for why the two diverge badly.
Where we draw the line, and why
Ilystics publishes an abnormal threshold for every month in 45 US metros. It is the 10-year mean of lost workdays for that month plus one standard deviation, capped at the number of workdays in a month. Above that line, the month sits in roughly the worst sixth of years on record for that market — the point at which "unusual" stops being rhetoric.
Be straight about what this is. Mean-plus-one-standard-deviation is not a legal standard. No court or board has adopted it, and no contract we know of specifies it. It is a defensible, disclosed convention for turning a comparative phrase into a number you can put in front of a GC — and its virtue is that the underlying distribution is published alongside it, so anyone can check the work or argue for a different line.
Worked example: Richmond, VA
Ten years of NOAA GHCND observations, scoped to soil-dependent site work — earthwork, grading, excavation, foundations and underground utilities:
| Month | Avg. precip | Typical lost workdays | Abnormal above |
|---|---|---|---|
| January | 3.67 in | 15 | 17 |
| April | 2.86 in | 10 | 13 |
| July | 5.44 in | 12 | 17 |
| November | 2.54 in | 9 | 12 |
Read the January and July rows together, because they contain the entire argument for measuring days instead of inches. July gets about 48% more rain than January and loses three fewer workdays. Summer rain in Richmond arrives as short convective storms onto warm, fast-drying ground. January rain is lighter, slower, and lands on cold saturated clay that stays untrafficable for days after the sky clears.
A claim built on "we got 5.4 inches in July, well above the annual monthly average" invites the response that July is always the wettest month here. A claim built on "we lost 19 workdays against a 12-day norm and a 17-day abnormal threshold" is a different conversation. Same weather, same site, opposite outcome.
Severity is necessary, not sufficient
Proving the month was unusual gets you past the first gate only. Every one of these clauses also requires that the weather actually delayed work:
- The affected work has to be on the critical path. A rained-out activity with float did not make the project longer, and no clause gives you time for it.
- Drying days count — if you prove them. Nothing in this language limits the impact to rainfall itself. If saturated ground made the site untrafficable Thursday, Thursday is claimable, but it takes soil evidence rather than adjectives. See jobsite soil drying time.
- Time, not money. Under all three document families, unusually severe weather is the classic excusable-but-non-compensable delay. You escape liquidated damages. You do not recover idle crew and equipment costs unless something else in the contract says so.
What to do, in order
- Find your notice deadline before you need it. Ten, fourteen or twenty-one days — it is in the contract you already signed, and it starts running without asking you.
- Check the supplementary conditions for a table. If the contract fixes anticipated weather days per month, that table is your baseline and the historical argument is over before it starts.
- Pull the baseline for your exact site, not your region.
- Log daily, including the days after the rain. The drying days are the ones nobody documents and everybody loses.
- Send notice early and supplement later. A thin timely notice beats a thorough late one.
Get the baseline this test is measured against
Ilystics generates an Anticipated Weather Exhibit for your exact jobsite: 10-year NOAA baseline, monthly abnormal thresholds, and USDA soil drying analysis. Free, no account.
Generate a Free ExhibitAlready in a dispute? Request a $249 Claim Report for the actual-vs-baseline analysis.
Ilystics provides weather and soil analysis, not legal advice. Section and clause references are to AIA Document A201-2017, FAR 52.249-10 and 52.249-14, and ConsensusDocs 200, and are provided for general orientation only — your contract, its supplementary conditions, and your jurisdiction control. The abnormal threshold described above is an Ilystics analytical convention, not a legal standard. Consult your attorney or claims consultant on entitlement.