Crooked Creek, AK
Crooked Creek, AK's wildfire risk, in USFS's own numbers
Crooked Creek's 63 buildings earn a 97th-percentile wildfire-risk score nationally under USFS's model — in USFS's highest wildfire-risk band nationally. (Source: USFS's Wildfire Risk to Communities methodology.)
USFS's separate burn-probability score — fire likelihood alone, before counting what's built there — puts Crooked Creek at the 97th percentile, close to its 97th-percentile risk score.
What "at risk" means for the buildings here
USFS classifies 95.2% of Crooked Creek's buildings as Direct exposure, higher than its 0% Indirect share and far above its 4.8% Minimal share — a profile where 60 structures sit close enough to vegetation that lot clearing matters most.
Where Crooked Creek ranks
Crooked Creek scores 97th nationally and 92nd within Alaska — close enough that its state context doesn't change the picture the national number already gives. Among the 31,521 US communities USFS scores, Crooked Creek ranks 900 for wildfire risk (1 is highest) and 30,530 by building count (1 is largest). Within Alaska alone, it ranks 28 of 322 places by risk. See the full county-by-county picture for Alaska on its state page.
Crooked Creek and the insurance market
Crooked Creek's 97th-percentile, very high rating is the kind of score behind a broader national pattern: insurers pulling back from the highest wildfire-risk markets, non-renewing or declining new policies outright, most visibly in California. Shopping around, not assuming automatic renewal, tends to matter here.
Hardening a home in Crooked Creek
Because Direct exposure dominates in Crooked Creek (95.2%), the highest-leverage fix is defensible space, not roofing or vents alone — covered in the home-hardening guide.
Where Crooked Creek's figures come from
Crooked Creek's 97th-percentile score and its burn-probability figure both come from the same USFS workbook, documented in the methodology guide. The exposure-zones guide covers what Crooked Creek's dominant direct exposure actually means, with real examples from across the dataset.