Range85-100
Very likely
Strong setup. Set the alarm and plan around it.
The Science Behind the Score
How Fog-Index turns atmospheric data into a single fog probability — the algorithms, the honest report card, and where the model is headed next.
Quick summary
Fog-Index helps photographers decide if an upcoming morning is worth the alarm, where to shoot, and what kind of fog scene to expect. Planning Context translates weather data into practical choices for viewpoint and composition.
Sunrise Window
Focused on the Light that Matters
Running advanced predictive models against global weather data is computationally expensive. To keep Fog-Index accessible and sustainable, the processing power is concentrated where it counts most for photography.
The Sunrise Window
The model actively analyzes conditions during the morning Sunrise Window: Blue Hour at civil twilight (-6°) through the end of Golden Hour (+6°).
Exact local times vary by date and location.
Civil Twilight Start
Blue Hour begins at -6°.
Sunrise
Sun crosses the horizon at 0°.
Golden Hour End
Morning golden light wraps at +6°.
Fog Score Ranges
Fog Score ranges, at a glance
This is the core Fog-Index scale. Every forecast score lands in one of these four color tiers.
Range70-84
Likely
Good odds. Worth the drive with a backup angle.
Range60-69
Possible
Borderline. Keep plans flexible and scout-ready.
Range0-59
Unlikely
Long shot. Stay in bed if you want fog.
Planning cues come after this score to help with composition and positioning. They add context, not points.
Planning Context
From Fog Likely to Shot Ready
Fog-Index doesn't stop at “fog likely.” It helps you decide what the air will feel like, where to stand, where else the fog is, and what light to expect. When a morning scores high, Planning Context translates meteorology into shot strategy — fewer blind alarm clocks, more intentional choices.
Fog Thickness
A surface-level read at your saved spot, taken from visibility distance at sunrise. It answers the in-the-soup question — thin, layered, thick, or white-out — which is really the question of whether light can break through. Pair it with Sky Above.
ThinLight passes straight through; glow more than separation
LayeredLight breaks through in beams and layers — the sweet spot
ThickLight diffuses before it breaks through; close textures
White-outNo light penetration; minimal, intimate scenes
Fog Level
Where the fog sits relative to you, vertically. Fog-Index estimates the top of the layer from the atmosphere's vertical structure — pressure levels, humidity and cloud continuity, inversions — and compares it with your spot's elevation.
Overlook mode goes one step further. For a viewpoint that stands well above the nearby valley, Fog-Index also forecasts the valley floor beneath it and reports where the fog sits relative to you: below you, at your level, or above you, drawn on a cross-section of the real terrain.
Below youYou look down on the layer
Near your levelThe edge could land either side — have a backup
At your levelYou're standing in it — moody atmosphere
Sky Above
Estimates the light quality above the fog deck by reading only mid- and high-level cloud cover. Low cloud is intentionally excluded because it's often the fog layer itself.
OpenClear sky above — best light-break potential
FilteredThin mid/high cloud — soft glow likely
BlockedHeavy overcast — lower odds of dramatic light
Nearby Fog
Some of the best mornings happen a few miles away. Valley fog pools on the low ground and big weather models can miss it at your exact spot, so Fog-Index also checks the ground around you and tells you where the fog is.
- Up to eight areas, every forecast
- Each forecast also looks at up to eight areas about 7 miles out on the compass points, in the same request as your spot.
- Which way, how low, which town, what time
- You see the directions with fog, whether it sits on lower ground, the town that way, and when it starts. Your other saved spots in that direction show up with their scores, one tap away, and the fog map opens on that morning.
- A clue, not a promise
- It comes from a single model, the ICON forecast, so it never changes your Fog Score and never sends an alert on its own. It rides along in the alerts and digests you already get.
- Fog
- Below your threshold
- Nearby
- Fog S to W · toward Vancouver · lower ground · from 6 AM
How it works under the hood
- Fog Thickness is a surface-level read at your saved spot, taken from visibility distance at sunrise and checked against humidity, dew-point compression, weather code, and wind stability, to say whether the fog will be thin, layered, thick, or a white-out — and so whether light can break through it.
- Fog Level starts from a fog-top estimate derived from vertical atmospheric structure — pressure-level humidity, cloud-layer continuity, and inversion context — confidence-gated before surfacing.
- Sky Above reads mid-level (≤ 30% = open) and high-level (≤ 40% = open) cloud cover independently. Low cloud is excluded because it is often the fog deck itself.
- It then compares your saved location's elevation against that fog top with a safety margin: below you, at your level, or near your level when the two are too close to call or confidence is low. Overlooks add a second forecast at the valley floor, which can also put the fog above you.
- Nearby Fog reads ICON's fog codes for up to eight areas about 12 km out (two compass points can fall in the same forecast cell, which then counts once), in the same request as your spot. “Lower ground” means every foggy area sits at least 50 m below your spot, and town names come from GeoNames.
- Signal confirmation checks ICON separately through Open-Meteo, alongside Tomorrow.io and NWS. The primary weather inputs remain Open-Meteo best-match. A fog code from best-match or ICON contributes a single fog-code bonus, even when both confirm. Tomorrow.io and NWS add context without changing the numeric Fog Score.
Two Photographer Workflows
Start with a high-score morning. Then choose the workflow that matches what you want to make.
In the Soup
Above the Fog
In the Soup
For forests, lakeshores, creeks, and intimate scenes where atmosphere and depth are the goal.
- Read Fog Thickness first. Layered is the sweet spot for light breaking through; white-out favors close, minimal compositions.
- Use Sky Above to judge whether light can leak through or if the whole scene will stay flat.
- Keep Fog Level secondary here. It is mostly an exit plan if the layer looks too thick.
- When your spot looks clear, check Nearby Fog. It points to the direction, and the town, where the soup is.
- Arrive early and adapt: when the fog thins, go wider; when it thickens, simplify.
Above the Fog
For peaks, ridges, lookouts, and city overlooks where the plan is to climb above the layer.
- Switch on overlook mode so the card scores the valley floor beneath you and says whether the fog will be below you.
- Read Fog Level first. “Below you” is the morning you climbed for.
- Look at Nearby Fog for lower ground that fills with fog even when the valley floor straight below stays clear.
- Check Sky Above and the cloud split to judge sunrise and light-break potential.
- Use Fog Thickness as context for how the world may look if you stay inside the layer instead.
- Always keep one backup location at a different elevation band.
Overlook Mode
Will you be above the fog, or standing in it?
A Fog Score describes the air at your pin. On the best mornings at a viewpoint that air is perfectly clear, because the fog you drove for is sitting in the valley a few hundred feet below you. Overlook mode forecasts that valley floor as a second point and reports where the layer sits relative to where you will be standing.
Fog below you
A fog sea in the valley, clear air where you stand.
The morning people drive for. Cloud fills the canyon and the rim catches first light above it.
Fog top ~280 ft · you 682 ft
The ground is a measured elevation transect through your pin. The fog top is a modeled estimate, so Fog-Index draws it blurred when confidence is middling and not at all when it is low.
Two forecasts, one card
The valley floor is found from the terrain around your pin, then scored with the same model that scores the viewpoint. Your spot keeps its own score; the card simply leads with whichever of the two is the morning worth setting an alarm for.
Alerts that read the drop
An alert fires when either your spot or the valley floor clears your threshold, so a dry summit above a canyon full of cloud still reaches you. The subject line opens with where the fog sits rather than making you work it out.
It recognises a viewpoint
Saving a promontory, a cliff edge or a gorge rim switches the mode on by itself, with the drop to the valley floor shown before you save. Spots with nothing far enough below them are left exactly as they were, and it is one tap to turn off.
Fog Seasonality
A planning layer for timing trips, not just tomorrow morning
Forecast answers whether the next sunrise looks worth the alarm. Fog Seasonality answers when a place usually comes alive across the year. It turns the saved-locations dashboard into both a short-range forecast tool and a longer-range planning tool for photographers deciding when to book the drive, trip, or season.
Yosemite seasonality example
Historical fog patterns for Tunnel View. Explore the monthly and weekly record below.
Loading Yosemite’s historical fog patterns…
Monthly first
The default view stays broad and readable: stacked monthly bars showing how often a place lands in Likely or Possible territory across a rolling 10-year window.
Weekly drill-down
Inside the location drawer, weeks reveal onset and tail-off that month averages can hide. That matters for places where the first week and last week of the same month feel like different seasons.
Intentionally not the live model
Seasonality is built from ERA5-Seamless reanalysis archive data — 0.1° ERA5-Land ground-level detail blended with ERA5 — and a historical Fog-Index scorer variant. It keeps the sunrise-window philosophy and the same planning intent, but it drops live-only inputs so the historical layer stays stable, portable, and explainable.
Methodology note
Fog Seasonality is a historical average of fog tendency, not a promise for the coming season. A location can still miss in an unusually dry autumn or overperform in an unusually wet one. Use it to compare places and time trips, then use the live forecast to decide whether this week is the week.
Algorithm Report Card
The Algorithm Report Card
Every region has a unique atmospheric fingerprint. Fog-Index works worldwide today, and was originally tuned and tested in the Pacific Northwest. The Fog Potential Map currently covers the continental US, while broader regional tuning and support for more fog regimes continue to expand. The algorithm runs two dedicated scoring pathways in parallel — one for radiation fog, one for advection (coastal) fog — and automatically picks the best fit for conditions at your location.
I believe you should know exactly where the model excels and where it's still learning.
| Fog Type | Current Algorithm Fit | The Honest Truth (Notes) |
|---|---|---|
| Radiation Fog | 🟢🟢🟢🟢🟢Excellent | The primary target. The model is perfectly tuned to detect the high-saturation, calm-wind, clear-sky cooling that creates these classic sunrise layers. |
| Valley Fog | 🟢🟢🟢🟢🟡4.5 / 5 | A core strength. Valley terrain bonuses, inversion-layer detection, and winter stagnation persistence make the model highly effective at predicting deep, persistent valley events. |
| Advection Fog | 🟢🟢🟢🟢🟡4.5 / 5 | Strongly improved in v2.3.0. The model now uses sea-surface temperature (air-sea delta) plus an upwind 3-ray marine fan with onshore/offshore wind checks. Remaining gap: performance still depends on marine data availability and sparse coastal sampling. |
| Upslope Fog | 🟠🟠🟡⚪⚪2.5 / 5 | Improved in v2.5.0. A conservative upslope mode now activates for ridge fallback transport when moisture and wind support lift, and it bypasses ridge and inland fallback penalties. Remaining gap: still heuristic and not yet a full terrain-relative forced-ascent model. |
| Precipitation Fog | 🟠🟠🟠⚪⚪Developing | Improved in v2.3.0 with a frontal fog gate (recent rain + rapid moistening + gust guardrail). Still heuristic: it captures many rain-driven setups but not full frontal dynamics. |
| Steam Fog | 🔴⚪⚪⚪⚪Poor | Requires data that isn't available yet: the specific temperature difference between cold air and warm water. The model will likely miss these localized events. |
Global Coverage
Why One Model Can't (Yet) Predict the World
Atmospheric physics change drastically based on geography. A model perfectly tuned for Seattle's stable valleys will fail wildly in a tropical environment.
| System Type | PNW, USA | Costa Rica |
|---|---|---|
| System Type | Stable Stratiform & Marine (Current Focus) | Dynamic Convective (Future Focus) |
| Wind | Calm (valley fog) or moderate onshore (marine fog) | Strong wind (pushing upslope) |
| Temperature | Cold ground temperature | Warm ocean air |
| Cloud Type | Flat, uniform cloud layer | Tall, puffy cumulus clouds |
| Formation | Valley fog needs calm, stable air and clear skies (radiative cooling). Marine fog needs onshore wind pushing moist ocean air over cold water. | Comes from unstable air, rapid uplift, and intense convective activity — fundamentally different physics. |
Conclusion: Even with our parallel pathways for marine fog, tropical and convective fog systems remain outside the model's reach. Conditions like orographic lift and warm-ocean convection that cause fog in other regions still need dedicated tuning.
Fog Types
The Essential Fog Types
The Photogenic Six: Understanding Your Target
Clear-sky cooling
Radiation Fog (Ground Fog)
The classic "morning mist." Forms overnight on clear, calm ground. Hugs valleys and burns off quickly after sunrise.
Cold-air pooling
Valley Fog
A sea of clouds. Cold air drains into basins, creating thick, persistent layers that can last for days in winter inversions.
Marine transport
Advection Fog (Coastal Fog)
The coastal blanket. Warm, moist air rolls over cold water. Thick, persistent, and driven inland by wind.
Terrain lift
Upslope Fog
The mountain capper. Moist wind forced up terrain cools and clings to peaks and forested slopes.
Warm water, cold air
Steam Fog (Sea Smoke)
Sea smoke. Wispy vapor rising from warm water into biting cold air. Surreal, fleeting, and distinct.
Rain-driven saturation
Precipitation Fog
Frontal fog. Forms when warm rain falls through cold air, saturating it. Widespread, soft, and moody.
What's Next
The Horizon: What's Next
This model is forever being honed. With radiation, valley, and marine fog now covered, here's where we're headed next.
DONE
The Foundation
Radiation & Valley fog tuned for the US. The baseline algorithm that started it all.
DONE
The Marine Layer
A dedicated Advection Fog pathway now runs in parallel, scoring coastal marine fog with wind direction, onshore flow detection, and summer seasonality.
NEXT
Mobile Experience
Native iOS and Android apps for robust push notifications and location-based alerts on the go.
FUTURE
International Expansion
Expanding beyond the US to fog-rich regions worldwide — from the UK moors to the coastal valleys of Portugal and beyond.
LONG-TERM
Community Tuning
The ultimate vision: allowing trusted users to "verify" fog events, feeding real-world ground truth back into the model.
Ready to stop missing foggy mornings?
Join other fog-loving nature photographers and cinematographers who trust Fog-Index to alert them when conditions align.
Loading...