Mapbox Satellite imagery turned into a copy of Bing imagery

I generally check multiple imagery sources when editing the map. For a couple of years I’ve been using Mapbox Satellite imagery as one of my sources, but over the last week or two it has changed.

Mapbox Satellite always had horrible image quality in many rural areas, but excellent quality in urban and suburban area where I usually edit. I usually edit near Tulsa, Oklahoma. Mapbox Satellite had some of the clearest imagery of any source in much of the Tulsa metropolitan area. Similar to Esri Clarity Beta, it was always wintertime imagery instead of summertime (like Bing), so I could trace streams and streets with heavy tree cover because the pictures were taken when there were no leaves on the trees.

A week or two ago, that clear imagery started being available only at certain zoom levels, neither zoomed out too far nor zoomed in too far. Otherwise it would be replaced with another imagery that was less helpful. And now, just within the last two days, what I get from Mapbox Satellite is just a copy of Bing, with the color and contrast enhanced a bit. Cars and even a garbage truck picking up trash that are visible on Bing imagery are there in the exact same place on Mapbox Satellite. Mapbox Satellite is no longer useful to me if it’s just going to be a copy of Bing imagery.

They are not really a copy of Bing Aerial, they just use the same image provider = MAXAR.
So you have MapBox, Bing and ESRI using the same images from the same provider at the same time, the difference is the quality and calibration of the images. Google also uses MAXAR images but at least they are more recent for urban areas.


You can’t make any general statements about the quality of imagery of the Bing, Mapbox or Esri layer. Their capture date, resolution, offset etc. varies between regions. In some regions, they incorporate aerial imagery from authorities (often open data datasets).


Well, in the areas I was editing, I always had images from three different time periods. Esri (Clarity) Beta had the oldest images (from a decade or more in the past), Mapbox Satellite somewhere in the middle, and Bing the newest. Sometimes in some areas they would vary from that. Sometimes the regular Esri (not Clarity) has newer imagery than any of them, but sometimes it’s just a copy of the Esri (Clarity). I always got a wide variety of different time periods of images from these different providers. Now, with whatever they changed, Mapbox only shows the exact same thing Bing shows. Not only the same time period (which was never the case before) but the exact same time, basically the same image.

You should keep in mind that ESRI Clarity is a mosaic of images without cloud cover from any date .The use that should be made of ESRI Clarity It is related to not finding coverage from other providers for the same area.
It basically amounts to saying that if all providers have cloud coverage over a mapped area, it is better to use ESRI Clarity with images from several years ago than to have nothing at all.
Generally the best and most recent images are of urban areas, urban areas that evolve dynamically, with the evolution of rural areas being a little less for which The images from several years ago to map roads are still valid, the roads evolve slowly, you can find sections of roads that have not changed in 100 years but not around them that may have news regarding buildings, crops, forests,… What I mean is that for urban areas use any provider except ESRI Clarity and for rural areas use any provider unless the area has cloud coverage for which ESRI Clarity is there, even if they are images from several years ago.

You can find information about the dates of the images on the MapBox blog, I think I remember an update of the MAXAR 2022 base (European and North American urban areas) and some selected areas with the MAXAR 2023 base, these satellite image updates can also be detected when there are changes to the MapBox Satellite API version.

Well, in Oklahoma things change slowly. Especially in the rural area, but even in urban areas you will not have things changing as fast as they do in Shanghai or Bangkok. The newest imagery is not always necessary to get things mapped right.

ALL of my imagery providers lack cloud cover in the areas I usually work in. What I like about Esri Clarity is that it also lacks tree cover, because the images were taken in wintertime when the leaves are not on the trees. That makes it possible to get creeks mapped in the right location. And sometimes to get building shapes right. And to more clearly see the edges and centerlines of roads and sidewalks where there are trees covering them.

Until last week, I could get that same benefit from Mapbox Satellite, in basically the entire Tulsa metropolitan area. It was newer and clearer than Esri Clarity for seeing buildings and roads, but the color contrast was low so it was still easier to see creeks in Esri Clarity. Its only drawback was that its offset was not very consistent from one area to another, but that was easy to work around by aligning it with one of the other imagery providers.

But now that Mapbox is just a copy of Bing in this area, it’s really not useful anymore, and Esri Clarity is the ONLY source I have for seeing accurate positions of creeks and streams, or buildings and roads that have tree cover. And that sucks, because its so old that buildings sometimes didn’t exist back then or have changed.

I guess it’s something I have no control over, and I should just be glad that Mapbox gave me such good imagery for the last few years. It just sucks that it’s no longer there now.


If the problem is the clarity of the images, this can be fixed by creating your own style in Mapbox Studio, configuring the brightness, contrast, and other little things of the satellite images. You can create a b&w style where some elements are more visible or stand out better than other geometric shapes.

Esri Clarity Beta is a misnomer; it’s actually just a very old version of the Esri World Imagery layer, and not necessarily any clearer. At this point, there’s usually little reason to use the Esri Clarity Beta layer.

If both Bing and Mapbox have the same imagery in your area, Esri World Imagery might have the same as well. But unlike the other providers, Esri has archived every release of their imagery layer in the past decade as Esri World Imagery Wayback. This diary post shows how to load Wayback imagery in any OSM editor. OpenHistoricalMap’s fork of iD recently added rudimentary built-in support for Wayback layers, which you can use today as part of the usual list of background layers. Rapid is adding more usable support for an upcoming release.

When using Wayback imagery, note that the dates correspond to when Esri released the imagery, which might be months to years after the imagery was captured. On the Wayback site, select a layer and click on the map to see the capture date. In OHM iD, this information is in the Background panel (CtrlShiftB or ⇧⌘B). In Rapid, the date selector will be based on the capture date.

Aside from aerial imagery, give the USGS 3DEP layer a try. This elevation model is available pretty much anywhere in the U.S. and great for tracing waterways based on their banks. In some regions, it can actually be better than leaf-off imagery, since late winter snow or early spring flooding can obscure the actual banks in visible imagery.

1 Like

What you’re saying may be generally true, but in my experience editing around the Tulsa area, Esri Clarity is usually different imagery from Esri. Sometimes the same. But usually different and MUCH clearer. The regular Esri is so fuzzy as to be almost unusable in a lot of areas around here.

Per your suggestion, I’ll give 3DEP a try for waterways. It may be easier to trace in some areas. There are some places where the trees are such that the streams really can’t be reliably seen on any imagery.

The Esri World Imagery Wayback releases from about 2014 should be identical to Esri Clarity Beta. It’s possible that Mapbox’s previous aerial imagery in Tulsa came from a provider that Esri has not used in the past, but it doesn’t hurt to flip through Esri Wayback over the years in case there is high-resolution aerial imagery in there.

The situation remains as described. For completeness, I am documenting what I find from the different imagery sources now, and what I found on the Esri Wayback site. I did not find the previous Mapbox Satellite imagery on the Wayback site, but I did verify the dates of the Esri imagery I get.

In the Tulsa city limits and other urban areas that border Tulsa, I generally get this quality of images.

From Bing, I get an image that’s usually recent within the last couple of years, and clear enough to map from, but just a little blurry. The cracks in the center of a concrete road can’t be seen or are not well defined. Here is an example. The roads are asphalt here, but the driveways are concrete. The cracks and expansion joints in the driveways can be seen, but are not well defined.

Generally within the city, I get the same thing from both Esri and Esri Clarity. When zoomed out, I get the 2014 imagery that you were talking about. It’s less clear than Bing, but seems to be more consistently aligned from one place to another (but all of the sources are getting better on alignment). It also is better for tracing streams or roads with a lot of tree cover, since it’s all winter imagery with no leaves on the trees. When I go to higher zoom levels, instead of the 2014 imagery I get some imagery from 2016 that is quite a bit clearer than Bing. (Outside of the city, in the suburbs I generally only get the 2014 imagery.) With the 2016 imagery, on a concrete road, the cracks are usually well defined so that I can easily map the centerline of a road if they put an expansion joint in the center of the concrete. Here is what it looks like. You can see all the cracks in the driveways, with better definition than the Bing Imagery. You can tell that the thing on the east side of the intersection is very likely a fire hydrant, and see that there’s probably no stop sign at this intersection.

What I previously was getting from Mapbox: if you can imagine a significant increase in resolution over the 2016 Esri Clarity imagery, that’s what I was getting. I could always see the cracks/centerlines of roads, not just usually. For a fire hydrant, I could usually see a fire-hydrant shaped shadow and be absolutely sure that’s what it was and not just a phone pedestal. Depending on the angle of the sun, I could almost always tell the difference between a stop sign and a yield sign based on the shape of the shadow.

What I currently get from Mapbox Satellite: it’s just a copy of the Bing imagery with the color and clarity enhanced a bit. Not really useful. If it was at least from a different time period it could be useful for comparison, but it’s really not useful since it’s just a copy of the same imagery from Bing. Here’s the screen shot.

My guess is the previous imagery was probably coming from INCOG (the Indian Nations Council of Governments), a Native American government organization based in Tulsa, and maybe they decided not to provide the imagery to Mapbox anymore. But, that’s just a guess.

You need to take Bing out of the equation, MapBox does not use Bing Aerial images.

Satellite and aerial imagery=
0–8: NASA MODIS satellite imagery
9–12: Maxar Vivid and NASA/USGS Landsat 5 & 7 satellite imagery
13–19: Maxar Vivid satellite imagery, Nearmap aerial imagery over US cities, and open aerial imagery.

Well, it’s clear that they do, if you look at my screenshot images. It’s the same picture. You can see it yourself if you go anywhere in the Tulsa city limits or other urban areas bordering on Tulsa city limits, and look at the Bing imagery and look at the Mapbox imagery. It’s the same, and it has been since right before I originally posted this.

We’re all talking about the same thing. Microsoft hasn’t contracted their own flyovers for over a decade now. Mosaic providers like Microsoft, Esri, and Mapbox all license satellite imagery from Maxar and supplement it with aerial orthoimagery from a combination of government and commercial vendors. There are only so many commercial vendors, so often the imagery ends up being the same, modulo each mosaic provider’s custom stitching and color correction.

Companies like Mapbox have to balance the upstream sources’ quality (resolution, alignment) against freshness. It’s great that they allow us to trace from their mosaics, but we aren’t their primary audience. The product is mainly used as a backdrop for their base maps in consumer applications. That results in an equation that looks sort of like: prefer the prettiest imagery, update it with uglier imagery if we can no longer get away with its age, and do it without breaking the bank.


I understand your frustration, I more than others, living in a country that is not an object of primary interest to the different free providers of satellite images. It makes me envious to see those beautiful and updated satellite images in European cities and cities in the United States.
Now the main problem that we all have is that the majority of satellite images used in OSM come from a single provider (MAXAR), and apparently the most used global base is that of 2022, A simple comparison against the TomTom satellite images (Maxar 2022 base) gives me to understand that now everyone uses that common base for the entire planet except for updates of interest such as from MapBox and ESRI.
A fun fact I just discovered, ESRI Aerial uses old images for zoom 20, which could cause errors for mappers who want to map in detail. I would believe that it should be the opposite, as you zoom in and get closer to the ground these images should be the most recent.
I suppose OSM editor developers already know this.
In some apps it is possible to restrict the zoom levels in the online sources file, and from the established maximum zoom, make digital zoom.

ESRI Aerial z19 and z20 respectively,
ESRI Aerial z20
ESRI Aerial z19

And the MapBox Aerial image over the same place shows=

And BING Aerial shows =

What I mean is that recent satellite images are always better, mapping over old images could cause a bias in the mapping, combining images from one and the other without realizing as the images show, a simple movement of 50 MTS in any direction can end in a CS that shows a polygon whose half is a parking lot and the other half is a park.
Satellite images are treated like any photograph to which you can add editing effects.
In OSM iD you can manipulate:
Brightness, contrast, saturation and sharpness.

(Edited to correct city below - it’s been 22 years since I’ve been on the OU Tulsa campus, so it has changed a lot, and I thought at first that the imagery I was seeing was their main campus in Norman.)

The situation at your coordinates in NormanTulsa, OK looks the same as what I usually find in Tulsa and its suburbs. Since I’ve been mapping for quite a while, I’m very familiar with the differences and different ages of the images. Much of what I map has been on the ground long enough that the older images don’t really matter, and I work from them because they’re clearer, but I always check against the most recent ones to be sure things haven’t changed.

In rural areas nowhere close to a big city, sometimes there is no clear imagery source, although it’s probably still better than what you see in Colombia. I have sometimes tried to map things in other countries, and sometimes not been able to see well enough to do what I wanted to. So I understand that my frustration is a first-world problem. ( I typed that earlier, but I think I changed it before posting the comment.) Almost everything that I would want to map in my home area is visible, just sometimes a little obscured so I’m not exactly sure where a road centerline is, stuff like that.

This reminds me of several years ago when I was mapping around Joplin, Missouri. At the time, Mapbox had just released 6- and 3-inch imagery for much of the U.S., but I was staring at nothing but clouds in any of the available layers. At the time, Mapbox had a form you could fill out requesting better imagery for a specific locality. They would sometimes patch in a different selection of imagery or even consider it in an eventual purchase of higher-quality imagery. I don’t know if they still have a program like that though.

Another time, I was mapping roads in southeastern Indiana, but the only reasonably up-to-date layer was the state’s orthoimagery. Normally that leaf-off layer was very useful for mapping in the woods. But leaf-off imagery is taken during the winter, and the area had just seen a heavy snowfall. At first I thought the plain white background was a bug in iD. :see_no_evil:

To avoid any misunderstanding, all of the imagery. mosaics mentioned in this discussion are commercial products. They are made available for cost free use in OSM editors by their owners because they see that as good way to support improving OSM (and in the end all of them use OSM so its a nice win-win situation).


They are commercial products that I use without paying, respecting the monthly consumption limits, that is, free, my small personal projects do not generate a high volume of consumption that may be subject to billing.
Through a free MapBox account you access 90% of its products. These products offer a high volume of monthly consumption for personal projects.
Bing Maps offers all of its APIs similarly, with no monthly consumption volume for Bing Aerial and other map styles (quadkey), Apikey only intervenes when requesting the metadata for the url of the aerial images or other styles, but its other services (Bing Maps Rest Services) if they require the use of an apikey to work. The Azure product branch is its strong point in terms of billing.

You are absolutely right, sometimes I tend to see everything from the same perspective and mixed my small personal projects with OpenStreetMap without taking into account its actual size.

By the way, this blog post also announces that Mapbox would be bringing in Vexcel imagery in U.S. cities. Vexcel’s website advertises a plan to collect 3-inch imagery for a number of cities this year, including Tulsa. :crossed_fingers:

1 Like