>The coverage map was created using data submitted voluntarily by the four mobile carriers using certain standardized propagation model assumptions
Eh? Why link or even talk about this. Here's 3rd party coverage collected by members over at Cellmapper. It's not perfect or complete but it's real data without 'assumptions' being made: https://www.cellmapper.net
For one, their map isn't restricted to areas that have paved roads or are otherwise regularly occupied by people, unlike the cellmapper map.
Radio propagation maps are used when siting new cell towers, for example. I'm not a radio wave propagation modeller, but my understanding is that 'assumptions' include things like treating trees as columns with a height equal to tree height and a diameter equal to tree crown diameter, for example. Accuracy depends on the type and resolution of input data (eg photo-derived elevation models vs lidar point clouds).
See [1] for some examples. I have no affiliation with them, it's just a site I have bookmarked.
> For one, their map isn't restricted to areas that have paved roads or are otherwise regularly occupied by people, unlike the cellmapper map.
So the question becomes which is more useful for determining whether or not a given phone will work in a given location. My experience is that the cellmapper data was far more reliable.
My folks have a place in on the beach in a (summer) resort community up north. It's sparsely populated and has very spotty mobile phone service.
I've had Verizon and AT&T (post-paid) which worked everywhere, including on the beach (without picking up service across the lake), Google Fi which worked in some places, but not anywhere on the property the home existed, Sprint and T-Mobile which were nowhere within ten miles of the city.
And then there was Visible. This an MVNO owned by Verizon but is not Verizon, however, it's cheap, fast (where it works) and simple/pre-paid. I had assumed, "hey, Verizon worked great, this is cheaper than them, so this will work, too." I'm not normally terribly naive, but more wrong I could not have been and I should have expected there to be no service.
The Verizon map shows a large dead zone north of where the home is, and smaller ones nearby, but it appears I'd have adequate coverage for the location of the lake house with any provider. Verizon post-paid works brilliantly because post-paid service phones end up in the "Extended Network". Visible had absolutely no service, much less than indicated on the (non-extended) Verizon map and the FCC map.
The cellmapper map was the most accurate but suffered from the problems you mentioned -- it strongly indicated that I'd have no service with Verizon (Visible) in those locations.
For these maps to be useful, they have to tell me what quality of service exists for an average phone in a given area. Having poor service (anything on my current phone that's two bars or less[0]) is no service. I'm sure it's a surprise to nobody that the provider maps are overly optimistic, but the fact that a 20-mile by 20-mile area that's bright red on the Verizon map (which indicates non-Extended service) has zero bars -- absolutely no service whatsoever -- and the surrounding ten miles has non-functional service is ridiculous.
Do they have more accurate information about coverage internally? I hope so -- if they're basing network expansion on this marketing map, they'll see the population density and the current performance and figure it's not worth fixing anything. If a business is connecting devices, wirelessly, you'd expect them to have service quality/reliability information that's far more accurate than what they presented to me. Admittedly, it's a "hard problem", but the existing maps are unacceptably bad and seem far worse than they should be. The moral of the story, at least as far as MVNO-Verizon is concerned, if there's a dead zone anywhere near where you need service, expect that lack of coverage to be an order of magnitude larger than indicated.
[0] I'm fairly certain that "2 bars" means "the phone briefly connected to a tower it can't otherwise use" -- it's a tease, but ultimate it's no service.
> So the question becomes which is more useful for determining whether or not a given phone will work in a given location. My experience is that the cellmapper data was far more reliable.
Not the case in, e.g., Bedford NY or New Canaan CT, just checked.
These are examples of suburban areas within an hour commute of mid-town Manhattan, where cell coverage is spotty (combination of NIMBY and rolling terrain).
Same story for southern coast barrier islands off Long Island.
To your point though, the carrier map claims blanket coverage in North Greenwich (S of Merritt Parkway), while in reality along North Rd there are 7 and 8 figure homes that have no cell service from any of AT&T, T-Mobile, or Verizon.
Lately I’ve found a quick and dirty viz for coverage is looking at a given carrier’s 5G coverage maps. They need stronger signal for that to work, so it shows the “dark” areas more easily.
On the barrier islands mapping from the carriers, you can even see effects of dunes. That surprised me.
Yes, the OP map shows 100% perfect coverage in my neighborhood for AT&T and that’s definitely wrong. I just moved here and I’m contemplating getting a new carrier because it’s nearly unusable for data.
See my comment history for thoughts on AT&T. Unless you're in one of their strong markets (Georgia?) or using taxpayer subsidized and prioritized service through Firstnet, they are the worst of the major carriers. It's no surprise DISH partnered with them after running spectrum hoarding scam for over a decade.
There's a similar site that shows towers and coverage around the world! And everyone can contribute data too! Sadly you can only contribute on Android but not iOS.
It's based ISED's bimonthly CSV data dump of licensed radio transmitters. Heavy concentration of cellular antennas in some blocks might be because ISED's database lists even nano/picocells.
Eh? Why link or even talk about this. Here's 3rd party coverage collected by members over at Cellmapper. It's not perfect or complete but it's real data without 'assumptions' being made: https://www.cellmapper.net
At least there's a site for Canada that has exactly where every single tower is located: https://www.ertyu.org/steven_nikkel/cancellsites.html