Seattle Metro Crawlspace Climate and Housing Data

What the data shows

Restrained readings only. Each of these is a restatement of what the numbers say, not an inference beyond them.

1. The wet season, not the annual total, is what a crawl space has to survive.

   At every published station, between 69.3% and 76.1% of the annual precipitation normal falls between October and March. A drainage design sized against the annual average is undersized for those six months.

2. Rainfall varies more inside this metro than one "Seattle rainfall" figure suggests.

   Published annual normals run from 33.37 inches at Everett Snohomish AP to 63.28 inches at Snoqualmie Falls, which is 90% more. The three wettest stations (Snoqualmie Falls, Landsburg and Monroe) are the three furthest east; the other 5 all sit between 33.37 and 40.81 inches.

3. No single station describes a city, and for some cities nearby stations disagree widely.

   The published stations within fifteen miles of Issaquah and Sammamish differ by 28.59 inches of annual normal. For Shoreline and Edmonds the same measure is 4.45 inches. That is why this dataset publishes rainfall per station and gives each city a station pointer instead of a rainfall value.

4. A city’s housing age says something about the code its homes were built under.

   Median year built ranges from 1966 in Burien to 2002 in Issaquah. Where more of the stock predates 1980, more of it was built under earlier expectations for ground cover, ventilation and insulation. It says nothing about the condition of any individual crawl space.

5. A median can hide an old core.

   Seattle has the highest share of housing built before 1950 in the table, 30.1%, yet 5 cities have an older median year built than its 1976. The survey counts every housing unit, apartments included, so a city can hold a large pre-1950 stock and a recent median at once. Read the shares, not only the median.

6. Where the soil survey rates the ground, the wetter classes are a minority of it.

   Among the 16 cities where most of the mapped area carries a drainage class, the share in the wetter classes runs from 0.7% in Mercer Island to 23.4% in Auburn. A drainage class describes soil as mapped; what happens at one address still depends on that lot’s fill, grading and downspouts.

7. The soil survey says very little about Seattle, Burien and Shoreline.

   SSURGO assigns no drainage class to Urban land map units, and that leaves 87.1% of Seattle, 82.4% of Burien and 82.3% of Shoreline unrated. Their wetter-class shares (0.3%, 0.0% and 0.0%) are low because the ground is unrated, not because it was rated well drained. They are not comparable with the other 16 cities and should not be quoted as evidence of good drainage.

Table 1: precipitation stations

NOAA's 1991–2020 climate normals, one row per instrument. Annual precipitation across the published stations runs from 33.37 inches at Everett Snohomish AP to 63.28 inches at Snoqualmie Falls — a 90% difference between two stations about 33 miles apart. Between 69.3% and 76.1% of it falls between October and March at every station here.

NOAA completeness flags, republished unchanged: S standard (data for 24 or more years), R representative (10 or more), P provisional, E estimated. Five of these stations carryS (standard) with 25 to 29 years of record; three carryR (representative) with 20 to 23 years of record. If your use needs only the strongest stations, filter on this column — it is in the downloads for that reason.

Stations excluded, and the rule that excluded them

A station is published here only if NOAA's annual normal for it rests on at least 20 years of record, and it sits within 15 miles of at least one served city's Census internal point. The record rule drops stations sitting directly in cities this site serves, which is a real cost — but a short average presented as a thirty-year normal would be a worse one. NCEI's search over the area returned 56 stations: 8 are published, 12 are listed below, and 36 gauges with under ten years of record near a served city, or under 20 years and further away, are counted but not listed.

Fewer than 20 years of record

• SEATTLE BOEING FLD, WA USW00024234 — 19 years of record, NOAA flag R; 4.0 miles from Mercer Island
• BUCKLEY 1 NE, WA USC00450945 — 19 years of record, NOAA flag R; 13.4 miles from Auburn
• TACOMA NARROWS AP, WA USW00094274 — 17 years of record, NOAA flag R; 11.6 miles from Federal Way
• MCMILLIN RSVR, WA USC00455224 — 17 years of record, NOAA flag R; 11.8 miles from Auburn
• KENT, WA USC00454169 — 16 years of record, NOAA flag R; 2.5 miles from Kent
• EVERETT, WA USC00452675 — 13 years of record, NOAA flag E; 0.7 miles from Everett
• RENTON 3.6 SSE, WA US1WAKG0041 — 11 years of record, NOAA flag E; 3.4 miles from Renton
• MERCER ISLAND 1.5 NW, WA US1WAKG0005 — 10 years of record, NOAA flag E; 1.8 miles from Mercer Island

More than 15 miles from every served city

• BREMERTON, WA USC00450872 — 23 years of record, NOAA flag R; nearest served city Seattle, 15.8 miles
• CEDAR LAKE, WA USC00451233 — 26 years of record, NOAA flag S; nearest served city Issaquah, 16.4 miles
• PALMER 3 ESE, WA USC00456295 — 23 years of record, NOAA flag R; nearest served city Auburn, 16.8 miles
• STARTUP 1 E, WA USC00458034 — 21 years of record, NOAA flag R; nearest served city Redmond, 22.6 miles

Table 2: cities

Housing age from the American Community Survey, soil drainage from the USDA soil survey clipped to each city boundary, and a pointer to the nearest published weather station. Ordered by share of housing built before 1980, highest first. There is no rainfall column here on purpose — see the note above.

Built pre-1980, pre-1960 and pre-1950 are nested shares of all housing units, each summed from the Census decade buckets listed in the methodology. The pre-1960 column is the figure this site's city pages quote as "built before 1960".

Wetter soil classes means the share of the city's mapped area whose dominant USDA drainage class is somewhat poorly, poorly or very poorly drained. Soil unrated is the share with no drainage class at all — Urban land map units and water, which the soil survey does not rate. Both are percentages of the same total, so where the unrated share is large the rated figure is describing a minority of the city. Seattle's 87.1% is the highest in the table. In Seattle, Burien and Shoreline the unrated share is 50% or more, so their wetter-class figures are marked “not comparable”: they are near zero because the ground is unrated, not because it was rated well drained. These are also the three cities with the largest share of housing built before 1960 — the survey is least informative exactly where the housing is oldest.

Places excluded

None. Every one of the 19 places this site serves is an incorporated Washington city with a Census place boundary, so each has an ACS estimate and a boundary to clip the soil survey to.

Methodology

Every figure was pulled from a public interface on November 10, 2026, by script, and copied into the tables by script. Nothing was transcribed from a secondary site, and nothing was estimated, imputed or filled in. The queries are printed here so you can re-run them rather than take this page's word for the result.

Precipitation

NOAA NCEI data access service, datasets normals-annualseasonal-1991-2020 and normals-monthly-1991-2020. Stations were found by bounding-box search over the metro, then filtered to those whose annual normal carries at least 20 years of record and that lie within 15 miles of a served city. Annual normals and day-count normals are read out unchanged. The October–March total is derived by this site: NOAA publishes meteorological seasons, and the wet season here does not align with them, so the six monthly normals were summed. The arithmetic is ours; the inputs are NOAA's. Each city's nearest station, station count and spread are measured from the city's internal point in the Census Bureau's 2024 gazetteer, by great-circle distance.

https://www.ncei.noaa.gov/access/services/data/v1
  ?dataset=normals-annualseasonal-1991-2020
  &stations=USC00457773,USC00454486,USC00455525,USC00458278,USW00024233,USW00094290,USW00094248,USW00024222
  &dataTypes=ANN-PRCP-NORMAL,ANN-PRCP-AVGNDS-GE010HI,
             ANN-PRCP-AVGNDS-GE050HI,ANN-PRCP-AVGNDS-GE100HI
  &format=json&includeAttributes=true&includeStationLocation=true

Station search: https://www.ncei.noaa.gov/access/services/search/v1/data?dataset=normals-annualseasonal-1991-2020&bbox=48.1828,-122.7814,47.0348,-121.7132&limit=1&offset=0

Housing age

US Census Bureau, American Community Survey 2020–2024 5-year estimates, table B25034 (year structure built) and B25035 (median year built), one request per Census place GEOID. The three shares are sums of B25034 buckets divided by total units: pre-1950 is “1939 or earlier” plus “1940 to 1949”; pre-1960 adds “1950 to 1959”; pre-1980 adds “1960 to 1969” and “1970 to 1979”. For every place the ten buckets were summed and required to equal the table total. The published margin of error on each median is carried through to the table above and into the downloads.

https://data.census.gov/api/access/data/table
  ?id=ACSDT5Y2024.B25034
  &g=1600000US5363000

The same table from the Census API: https://api.census.gov/data/2024/acs/acs5?get=group(B25034)&ucgid=1600000US5363000

Soil drainage

USDA NRCS Soil Data Access. Each city's boundary was taken from the Census TIGERweb incorporated-places layer (ACS 2024 vintage), generalised, and used as the area of interest for a clipped map-unit intersection. The resulting polygons were joined to muaggatt.drclassdcd — NRCS's own dominant-condition drainage class for each map unit — and the areas summed by class. Percentages are of total mapped area including the unrated share. The survey areas involved, with the versions read on November 11, 2026: King County Area, Washington (WA633, version 22), Pierce County Area, Washington (WA653, version 22), Snohomish County Area, Washington (WA661, version 28), Mount Baker-Snoqualmie National Forest, Washington (WA762, version 10), City of Seattle, Washington (WA775, version 10) and City of Tacoma, Washington (WA776, version 9).

~DeclareGeometry(@aoi)~
select @aoi = geometry::STGeomFromText('MULTIPOLYGON((...))', 4326).MakeValid()
~DeclareIdGeomTable(@intersectedPolygonGeometries)~
~GetClippedMapunits(@aoi,polygon,geo,@intersectedPolygonGeometries)~
SELECT l.areasymbol, mu.mukey, mu.musym, mu.muname, m.drclassdcd,
       CAST(SUM(CASE WHEN x.gg.EnvelopeAngle() > 90 THEN x.gg.ReorientObject().STArea() ELSE x.gg.STArea() END) AS BIGINT) AS area_m2
FROM (SELECT g.id, GEOGRAPHY::STGeomFromWKB(g.geom.MakeValid().STAsBinary(), 4326) AS gg
      FROM @intersectedPolygonGeometries g) x
JOIN mapunit mu ON mu.mukey = x.id
JOIN legend l ON l.lkey = mu.lkey
JOIN muaggatt m ON m.mukey = x.id
GROUP BY l.areasymbol, mu.mukey, mu.musym, mu.muname, m.drclassdcd

Known imprecisions, stated rather than smoothed over. City boundaries were generalised to 0.0001 degree before the intersection; interior holes were subtracted. The mapped area counts water map units, so it can run above Census land area where a boundary takes in open water — visible in the soil_mapped_sq_km column, which is published so you can see the difference rather than having to trust that it is small. Everett's boundary is several separate polygons, and its soil row covers all of them.

Variables: what each one supports, and what it does not

The limitation column is the more important half. A variable used past what it can support is worse than no variable at all, because it arrives with a federal agency's name attached.

Annual precipitation normal Measured

Period: 1991–2020 climate normals · Geographic unit: One instrument at one point

Useful for: Establishes how much water the ground around a house has to shed, and shows how sharply that varies across a metro that is often described as if it had one climate.

Cannot support: A normal is a 30-year average for a single gauge. It is not a value for the surrounding city, and it does not describe any individual year. The wet season that floods a crawl space is a storm sequence, not an annual total.

October–March precipitation and its share of the annual total Derived

Period: 1991–2020 monthly normals, summed October through March · Geographic unit: One instrument at one point

Useful for: The most directly relevant climate figure for crawl space water management: between 69.3% and 76.1% of the year’s precipitation at every published station falls in six months. Drainage sized for the annual average is undersized for the months that matter.

Cannot support: Derived by this site by summing NOAA’s published monthly normals: the arithmetic is ours, the inputs are NOAA’s. NOAA publishes meteorological seasons (December to February, and so on); October to March was chosen because it is the wetter half of the year at these stations, and that choice is an editorial one.

Average days per year at or above 0.10, 0.50 and 1.00 inches Measured

Period: 1991–2020 climate normals · Geographic unit: One instrument at one point

Useful for: Separates frequency from intensity. At Seattle Tacoma Intl AP, the station with the longest record here, 93.4 days a year reach 0.10 inch and 5.6 reach a full inch: many days of light rain and comparatively few heavy ones. Different failure modes follow from each.

Cannot support: A daily total, not an hourly rate. A crawl space floods on intensity over hours, and these normals cannot show that. They are also not a trend: nothing here says whether heavy days are becoming more common.

Median year built, and share of housing built before 1980, before 1960 and before 1950 Measured

Period: ACS 2020–2024 5-year estimates · Geographic unit: Census place (incorporated city)

Useful for: Older housing stock means a larger share of homes were built under earlier codes, and under earlier expectations about ground cover, ventilation and insulation. It is the closest publicly available proxy for how much of a city’s housing predates modern crawl space practice.

Cannot support: Covers ALL housing units, apartments included, not single-family homes, and the ACS does not ask about foundation type, so nothing here identifies which units even have a crawl space. These are survey estimates with sampling error; the published margin of error on each median is given. Age is not condition: an old house with a maintained crawl space is in better shape than a new one with a blocked downspout.

Share of mapped area in wetter soil drainage classes Derived

Period: SSURGO as published, retrieved November 2026 · Geographic unit: Census place boundary, area-weighted over SSURGO map units

Useful for: Drainage class is the soil scientist’s summary of how readily water moves out of the ground. Among the 16 cities where most of the mapped area carries a class, the share in the wetter classes runs from 0.7% in Mercer Island to 23.4% in Auburn.

Cannot support: Describes natural soil in place. It says nothing about the fill under a specific house, the grading of that lot, or where its downspouts discharge, and those usually matter more at the scale of one crawl space. SSURGO also assigns no drainage class to Urban land map units or to water, so the unrated share is published per city. In Seattle, Burien and Shoreline it is 82.3% to 87.1% of the mapped area: their wetter-class shares are near zero because the ground is unrated, not because it was rated well drained, and they should not be compared with the other cities.

Spread among published stations within 15 miles Derived

Period: 1991–2020 climate normals · Geographic unit: Computed per city from the station table

Useful for: Quantifies how much the choice of station changes the answer. Where it is large, no single station describes that city, which is the reason this dataset publishes no city-level rainfall figure.

Cannot support: A range, not an uncertainty estimate. It reflects real geographic variation in rainfall as much as any error, and a large spread is a fact about the terrain rather than a defect in the measurements.

Variables considered and rejected

A dataset that shows only what it has looks complete. One that shows what it rejected, and why, lets you judge whether the selection was driven by rigor or by whatever happened to be easy to obtain. Seven variables were considered and not published.

Crawl space failure, moisture or mold rates

Considered because: It is the variable everybody actually wants, and its absence is the reason the rest of this page has to work so hard.

Rejected because: This site holds no such data, and none of the three federal sources used here measures it. A rate quoted without a measured sample behind it would be an invention.

What it would take: Instrumented measurement across a sample of real crawl spaces, stratified by construction era and by neighborhood, sustained over multiple wet seasons. That is a research program, not a page.

A composite crawl space risk score

Considered because: The three published variables could be combined into one number per city, and it would be the most linkable thing on this site.

Rejected because: A score implies a validated relationship between rainfall, soil, housing age and actual crawl space failure. This site has no outcome data to fit or test such a relationship, so the weights would be invented, and the invented part would be the part everybody quoted. Three defensible variables plus one invented weighting is not three-quarters defensible; it is a fabrication wearing the other three as credentials.

What it would take: Outcome data to fit and validate against, which is the previous row. Until that exists, a score cannot be built honestly, only asserted.

Foundation type: what share of homes have a crawl space

Considered because: It would turn housing-age context into something far more directly useful, and it is the first question an inspector or agent asks of this table.

Rejected because: The American Community Survey does not ask about foundation type, so the source behind the housing columns cannot supply it. County assessor records for King, Snohomish and Pierce counties were not examined for this dataset, so whether they carry a usable foundation field is an open question here rather than a finding.

What it would take: A bulk assessor extract carrying a foundation or basement field from each county these cities lie in, with a written mapping between the counties’ vocabularies. One county alone would not be publishable here, because the rows would stop being comparable.

FEMA flood-zone coverage per city

Considered because: It is authoritative, mapped, and available: the National Flood Hazard Layer would support a "share of city in a Special Flood Hazard Area" column.

Rejected because: It would be the most misread column on the page. A flood-zone share describes river and coastal flooding on mapped floodplain. Water also reaches crawl spaces as roof discharge, surface runoff and groundwater on lots far from any mapped floodplain, so the column would be read as a crawl space indicator it is not, and a homeowner well outside any Special Flood Hazard Area would take false comfort from it.

What it would take: A demonstrated relationship between flood-zone status and crawl space water intrusion. We are not aware of one, and publishing the column while disclaiming the inference would not stop the inference.

Depth to groundwater / winter water table

Considered because: Directly relevant: a high winter water table is one way water comes to stand under a house.

Rejected because: No source for a comparable city-level water-table figure was identified for this dataset. Well records are point observations of varying vintage and depth, made for drilling rather than for hydrology, and a seasonal high water table is precisely what a single reading does not capture.

What it would take: A published, maintained water-table surface for the metro. None was identified for this dataset.

Elevation and topography

Considered because: Hillside lots behave differently, and the site already writes about curtain drains on sloping ground.

Rejected because: A mean elevation for a city is not a fact about any house in it. What matters is the slope of one lot and what is uphill of it, which no city-level statistic can carry.

What it would take: Parcel-level slope, which is a per-property question: the sort a site visit answers and a table cannot.

Radon test results per city

Considered because: Radon bears on how a crawl space is sealed and vented, and the Washington State Department of Health publishes laboratory radon test results.

Rejected because: Washington publishes those results by county and by census tract only. There is no city or ZIP code geography to match the rows here, and the tract data is thin: in 2025, 26 of King County’s 495 tracts and 2 of Snohomish County’s 175 had ten or more tests. The results are tests people chose to submit, not a sample of homes. A county figure repeated on every city row would look like city data and would not be.

What it would take: Test results published at city or ZIP code level with enough tests in each area to mean something. Until then a county figure belongs in the text that discusses radon, cited to the Department of Health, not in this table.

Download

Both files are generated at build time from the same data the page reads, so a download cannot fall out of step with the tables. Each carries the sources, the retrieval date, the limitations and the two-table warning inside the file itself, so it still says what it is once it has been separated from this page.

Field dictionary

The same field names are used in both formats, and this table is the source they are built from.

Station table fields

City table fields

How to cite this, and how not to use it

Use this form, or any form carrying the publisher, the dataset name and the update date:

Seattle Crawlspace Pros. "Seattle Metro Crawlspace Climate and Housing Data." Compiled from NOAA, US Census Bureau and USDA NRCS data. Updated November 2026. SeattleCrawlspacePros.com.

The underlying data is US federal public-domain data and you owe nobody anything for it. What this site added is the compilation: the station selection rule, the city pairing, the soil intersection and the limitations. The underlying data is US federal public-domain data from NOAA, the US Census Bureau and the USDA. This compilation (the station selection, the city pairing, the soil intersection and the limitations) may be quoted and republished with attribution to Seattle Crawlspace Pros and a link to the dataset page. Please carry the measurement period with any figure.

Reasonable uses

• Explaining why water management matters more in some parts of this metro than others, with the soil and rainfall figures attached and their geography stated.
• Establishing what share of a city's housing predates a given code era — as housing-stock context, not as a claim about any address.
• Comparing rainfall between two named stations, with the distance from your subject property given.
• Sourcing a rainfall or housing-age figure in a report or listing, cited to NOAA or the Census Bureau directly. They are the authorities; this page is a convenience.

Uses this data cannot support

• Any claim about crawl space failure, mold or moisture rates. Nothing here measures those, in any city, at any confidence.
• A per-city rainfall figure. Precipitation is measured at points that disagree with each other by up to 28.59 inches within fifteen miles. Name the station.
• A risk ranking of cities. The columns are not commensurable and there is no validated model to combine them with.
• A statement about a specific property. Soil drainage describes natural soil in place; what matters at one address is that lot's fill, grading and downspouts.
• Good drainage in a city the soil survey does not rate. A near-zero wetter-class share in Seattle, Burien and Shoreline reflects unrated Urban land, not well-drained ground.
• An inference from housing age to condition. Older stock means more homes built under earlier codes. It does not mean those homes have problems.

Corrections and maintenance

• Next scheduled review: November 1, 2027. The review re-runs the published queries against whichever ACS 5-year release and SSURGO version are current on that day. NOAA's normals are recomputed once a decade, so the precipitation figures stand until the next set of normals is published.
• Found an error? If a figure here does not match what the source says, tell us — name the table, the row and the column. Substantive corrections are published permanently in the public corrections ledger with the wording as it stood.
• No risk score, on purpose. A site that offers crawl space inspections has an obvious incentive to publish a score that makes every city look urgent, which is exactly why this dataset publishes the inputs and leaves the judgement about any one house to an inspection.

Revisions

If you cited a figure from this dataset, this is where you check whether it has moved since. The same log is written into the header of the CSV and the revisions array of the JSON, so a file saved before a change can be checked against it.

None yet. No figure has been revised since this edition of the dataset was assembled on November 11, 2026.