Frequently asked questions

Accurate enough to choose a street; not accurate enough to choose a side of a tree. Tree cover comes from a 30 m grid across the continental US and a 10 m global land-cover layer beyond it, so it describes a stretch of road rather than a specific spot — finer again where a city has published its own canopy survey. Tree heights are about a metre across but come from a model with a published mean absolute error near 2.8 m, and we take the tallest canopy in a neighbourhood rather than any one tree's height.

The sun position is the precise part. Altitude and bearing are solved for your date, hour and latitude, so the direction and relative length of shadows through the day are reliable even where the canopy input is coarse.

Almost certainly buildings. Where we have imported real building footprints, turning on “Building shade” (next to Sun-aware) adds their shadows, so a shaded street canyon reads correctly. Everywhere else, buildings still cast no shadow in the model at all — that's the dominant error in any other downtown, and wider coverage is a licensing problem more than a data one. The coverage map shows where building shade is switched on.

Yes, including the shade. Tree heights have always been global, and tree cover now falls back to the ESA WorldCover 10 m land-cover product wherever the US canopy dataset does not reach — so a route outside the continental US gets a real shade figure and a real map tint instead of an empty strip. Routing, elevation, distance, crossings, sidewalks, markers and GPX export work anywhere.

The caveat worth knowing: WorldCover calls each 10 m cell tree cover or not, where the US dataset gives a percentage per cell. So where WorldCover is all there is, the answer is blunter about partial cover — still good for “is this street tree-lined”. Some places are far better than that now: England is covered by two national tree layers, mapped woodland by woodland and crown by crown. The coverage map lists every canopy source in use.

Scanning the day for the time your particular route is most shaded, ignoring any hour when the sun is less than 15 degrees above the horizon. Without that floor the answer is always “one minute after sunrise”, which is true and completely useless.

Not with Sun-aware off: then the green tint shows tree cover — “where are the trees in this area” — which doesn't depend on the hour. Switch on Sun-aware and the tint changes to shade at the time you pick, cast in your browser from each tree's height and the sun's position, and it moves with the slider along with the route line and the shade strip. The same date and time also drive the wind forecast, when Wind is on.

Dozens of them now, across five countries, and the two layers don't cover quite the same places — which is exactly why the list lives on the coverage map rather than here. That page is generated from the live source list, so it is right on the day you read it; any list typed into this answer would be wrong within a fortnight.

The rule is the same everywhere. OpenStreetMap is the baseline worldwide. Inside the real boundary of a place that publishes its own field-verified inventory, that inventory is treated as ground truth and takes priority. Sidewalks are the thinner of the two on OpenStreetMap almost everywhere, so a local source makes far more difference there than it does for crossings.

We add to this most weeks. If somewhere you run isn't covered yet, tell us where — knowing where people actually run is what decides which city gets researched next.

Green means there's a traffic signal there. It's the generally safer place to get across a busy road, and it's worth being able to see that without hovering anything. Hovering any crossing gives its specific type — marked, unmarked, signalised, or with a pedestrian island — and where the data came from.

No — and that distinction matters. Nothing drawn means nobody has recorded it either way. A red-orange line is the only thing that means “somebody checked and there is no sidewalk here”, and it's only drawn from sources that genuinely survey for absence.

The map names its own sources. In the Layers panel, each layer shows which source is supplying it for the area you're looking at; click the name for the detail — what kind of evidence it is, how old it is, and a way to tell us if something looks wrong. It follows the map, so it changes as you pan into a place with its own data. Wind names its forecast the same way, and the elevation profile names where its heights came from. Every source is also listed, with its licence, on How it works.

No, and that's deliberate. They're map layers you look at while deciding, not something that quietly moves your line somewhere you didn't put it. A route you have to check isn't a route you drew.

Mostly from MET Norway, the Norwegian Meteorological Institute; outside the Nordic countries its forecast is built on the European Centre for Medium-Range Weather Forecasts' global model, one of the most accurate there is. In the cities the app knows best, the next two days come from sharper models read straight from the weather services' own files, with gusts: NOAA's 3 km HRRR in the US, and Germany's ICON-EU across Europe, the UK included. Where a sharper model and MET Norway disagree about an hour, the app says so. The forecast is hourly for about the next two and a half days, then every six hours to about ten days; further ahead than that the map shows no wind rather than a guess. How it works explains the whole calculation.

It is the pace your usual effort turns into on that stretch. Air pushes back harder the faster it meets you, so into a headwind the same effort carries you more slowly, and with the wind behind you a little faster — though never by as much as the headwind costs. Set the pace you run in still air under Wind in the Layers panel, and each stretch, and the route as a whole, is described in those terms.

Sometimes, and the app only says so when it's true. On open ground, a loop in a steady wind costs the same either way: every metre into the wind one way is a metre with it behind you the other. It matters when part of the loop is sheltered and part exposed (better to meet the headwind where it's sheltered), when the wind changes during your run, and it changes when you face the headwind: early while you're fresh, or on the way home. The strip under the profile says which, and offers a Reverse button when reversing really helps. It never reverses anything for you.

Yes. From every point on your route the app looks upwind for trees and buildings, using the same measured heights the shadows use, and works out how much of the wind gets past them from field research on windbreaks. Every stretch then gets the wind that actually reaches it: a tree-lined street far less than an exposed causeway. A band along the bottom of the wind strip shows where your route is sheltered and where it is exposed, and a stretch's card says what shelters it. Buildings count where the app has building data. How it works explains the calculation, and what it can't know.

Yes. Export as GPX and load it as a course on a Garmin, Coros or Suunto the way you would any other file. GPX import works too, if you'd rather look at the shade and climbs on a route you already have.

To save routes, yes — they have to belong to someone. The marketing pages, the coverage map and the public gallery don't need one.

New accounts are approved by hand while the app is still being built, so there is a wait between signing up and getting in. You'll get an email the moment yours is approved. Confirming your email address in the meantime is worth doing — it's a separate step, and it doesn't have to wait for approval.

Not a native one. The site works in a phone browser, but honestly, route planning is a big-screen job — you'll want the map. The watch is the thing you take out with you.

Because there are good apps for all three already, and none of them tell you whether the road will be in the sun. This does one job.

Nothing while it's in beta. See pricing for what we're thinking about after that — short version: the planner stays usable for free, and nothing is going to start charging without telling you first.

Only you, unless you choose otherwise. Saved routes are treated as sensitive data because they usually start at your front door, so every route is private when you save it and stays that way until you change it yourself.

There are four settings per route. Private is the default and means exactly that. Unlisted gives you a share link that works for anyone holding it and is listed nowhere; revoke it and the link stops working immediately. Public also lists the route in the gallery. Public with the start hidden does the same, but trims the first and last stretch of the line before anyone else sees it — the trimming happens on our side, so the full line is never sent to a viewer's browser at all.

Nothing reaches the gallery that you didn't set to public yourself. The privacy page spells out exactly what's stored.