7-day fog outlook · Chicago / Lake Michigan

7-day fog outlook — Chicago / Lake Michigan

run 2026-10-19T21:30-05:00 · home point 41.8872, -87.6272

Set-up score = the same factor blend the day-1 report uses, before calibration. It ranks mornings; it is not a probability. Day 1 is calibrated in the main report. ★ = best-ranked morning beyond tomorrow.

morning	set-up	band	type	light	T−Td min	wind	mid+high cloud	sunrise	models
Tue 20 Oct	0.69	decent set-up	deep fog bank	sun will reach the fog top	0.0 °C	1.5 kt	1 %	07:08 · 103.4°	5 (HRRR primary), spread 0.88
Wed 21 Oct ★	0.59	marginal	shallow ground fog	sun will reach the fog top	0.3 °C	1.7 kt	0 %	07:10 · 103.9°	5 (HRRR primary), spread 0.61
Thu 22 Oct	0.49	marginal	shallow ground fog	sun will reach the fog top	1.9 °C	3.7 kt	0 %	07:11 · 104.3°	4 (ECMWF primary), spread 0.47
Fri 23 Oct	0.58	marginal	shallow ground fog	sun will reach the fog top	1.3 °C	4.0 kt	0 %	07:12 · 104.8°	4 (ECMWF primary), spread 0.66
Sat 24 Oct	0.44	poor	shallow ground fog	overcast above the fog	2.0 °C	4.9 kt	68 %	07:13 · 105.3°	4 (ECMWF primary), spread 0.48
Sun 25 Oct	0.19	poor	shallow ground fog	overcast above the fog	5.0 °C	5.0 kt	92 %	07:14 · 105.8°	4 (ECMWF primary), spread 0.48
Mon 26 Oct	0.20	poor	shallow ground fog	overcast above the fog	2.8 °C	6.4 kt	100 %	07:16 · 106.2°	4 (ECMWF primary), spread 0.29

What a set-up score has meant (day-1 backtest, raw score before calibration, airport mornings)

strong set-up (≥ 0.75): fog on 25 % of 44 mornings
decent set-up (≥ 0.60): fog on 17 % of 82 mornings
marginal (≥ 0.45): fog on 12 % of 154 mornings
poor (≥ 0.00): fog on 8 % of 285 mornings

Ranking skill by lead (AUC from the lead backtest; 0.5 = none, 1 = perfect): 1 day ahead 0.746, 2 days ahead 0.726, 3 days ahead 0.738, 5 days ahead 0.627

Tue 20 Oct — set-up 0.69, decent set-up

Confidencesee the main report (calibrated, HRRR included)Typedeep fog bankLightsun will reach the fog topSunrise07:08 at 103.4° ESEHours saturated3Per modelgfs_hrrr 0.983, ecmwf_ifs025 0.614, gfs_global 0.108, icon_seamless 0.657, ncep_nbm_conus 0.673

Wed 21 Oct — set-up 0.59, marginal

ConfidenceHRRR reaches this morning; models usually agree on the airmass; backtest AUC at this lead 0.726 (measured at 2 days)Typeshallow ground fogLightsun will reach the fog topSunrise07:10 at 103.9° ESEHours saturated0Per modelgfs_hrrr 0.719, ecmwf_ifs025 0.736, gfs_global 0.123, icon_seamless 0.5, ncep_nbm_conus 0.597

Thu 22 Oct — set-up 0.49, marginal

Confidenceset-up only: dewpoint, wind and sky from the global models; backtest AUC at this lead 0.738 (measured at 3 days)Typeshallow ground fogLightsun will reach the fog topSunrise07:11 at 104.3° ESEHours saturated0Per modelgfs_hrrr None, ecmwf_ifs025 0.572, gfs_global 0.121, icon_seamless 0.573, ncep_nbm_conus 0.594

Fri 23 Oct — set-up 0.58, marginal

Confidenceset-up only: dewpoint, wind and sky from the global models; backtest AUC at this lead 0.738 (measured at 3 days)Typeshallow ground fogLightsun will reach the fog topSunrise07:12 at 104.8° ESEHours saturated0Per modelgfs_hrrr None, ecmwf_ifs025 0.736, gfs_global 0.121, icon_seamless 0.776, ncep_nbm_conus 0.604

Sat 24 Oct — set-up 0.44, poor

Confidencepattern only; expect the number to move; backtest AUC at this lead 0.627 (measured at 5 days)Typeshallow ground fogLightovercast above the fogSunrise07:13 at 105.3° ESEHours saturated0Per modelgfs_hrrr None, ecmwf_ifs025 0.558, gfs_global 0.092, icon_seamless 0.573, ncep_nbm_conus 0.482

Sun 25 Oct — set-up 0.19, poor

Confidencepattern only; expect the number to move; backtest AUC at this lead 0.627 (measured at 5 days)Typeshallow ground fogLightovercast above the fogSunrise07:14 at 105.8° ESEHours saturated0Per modelgfs_hrrr None, ecmwf_ifs025 0.091, gfs_global 0.064, icon_seamless 0.544, ncep_nbm_conus 0.12

Mon 26 Oct — set-up 0.20, poor

Confidencepattern only; expect the number to move; backtest AUC at this lead 0.627 (measured at 5 days)Typeshallow ground fogLightovercast above the fogSunrise07:16 at 106.2° ESEHours saturated0Per modelgfs_hrrr None, ecmwf_ifs025 0.288, gfs_global 0.086, icon_seamless 0.374, ncep_nbm_conus 0.107

Honesty notes

Days 2–7 use ECMWF, GFS, ICON and NBM (HRRR where its 48 h reach allows). The set-up score is the raw factor blend; it ranks mornings and is not a calibrated probability.
Skill falls with lead. Use days 5–7 only to spot a pattern; re-check nightly and treat the day-1 report as the decision.
No vertical profile, river gauge or NBM bulletin feeds the outlook; those are day-1 checks only.