I think at a minimum there should be the ability to tag roofs where roof level = 0 (and probably it's < 3 storeys or whatever, also possibly filtered to countries which have enough rainfall to not all be flat).
_Originally posted by @peternewman in https://github.com/westnordost/StreetComplete/issues/1889#issuecomment-650794445_
See #1889 for some of the background and previous discussion.
So, we need a list of countries where it is uncommon to have non-flat roofs at all.
In Poland flat roofs happen (especially for garages, supermarkets, industrial sites, offices), but non-flat roofs without entire building level in roof are typical - especially for houses.
I don't expect any country in Europe to have almost exclusively flat roofs.
So it looks like:
https://data.worldbank.org/indicator/AG.LND.PRCP.MM?name_desc=false&view=map
https://databank.worldbank.org/reports.aspx?source=2&type=metadata&series=AG.LND.PRCP.MM
Poland is one of the lower European countries at 600mm/year.
The data is CC-BY ( https://datacatalog.worldbank.org/public-licenses#cc-by ), so do you for example just want the rainfall by country (maybe averaged over the last five years) and then a threshold can be set (and potentially tweaked in future), or do you just want straight boolean data in SC?
Rainfall is a very rough measure anyway because the real data we want is not the average rainfall, but whether flat roofs are most common. We just assume that has to do with rainfall.
So better round up those countries where rainfall is quite low and then have a look at satellite pictures / streetview to confirm it.
Rainfall is a very rough measure anyway because the real data we want is not the average rainfall, but whether flat roofs are most common. We just assume that has to do with rainfall.
Yeah fair enough.
So better round up those countries where rainfall is quite low and then have a look at satellite pictures / streetview to confirm it.
Okay, here's a sheet for collaborative data-gathering:
https://docs.google.com/spreadsheets/d/1Oc4i2KsF-FyJ-Bxpu1dcbGK1Wrae7mHeg_CM2rpNBc8/edit?usp=sharing
I'd suggest some sort of binary search would make sense, unless you feel we need to check every country below a rough level @westnordost ?
I'd suggest some sort of binary search would make sense, unless you feel we need to check every country below a rough level @westnordost ?
So we've made some good collective progress @westnordost . How much more do we need to do overall to progress this from a data gathering perspective? Are you happy with a binary search at the bottom end, perhaps with a few spot checks?
I see there are a number of countries with no rainfall data we presumably either just skip these for now or will need to check them individually (which may be diminishing returns).
I can look at adding a lookup to the country codes you need from what it currently has too, which I guess will be required assuming it's going into the same style data files as the other stuff?
Since rainfall is only an indication and there are certainly also cultural aspects, it would be better to actually check each/more via satellite imagery.
Most helpful comment
Yeah fair enough.
Okay, here's a sheet for collaborative data-gathering:
https://docs.google.com/spreadsheets/d/1Oc4i2KsF-FyJ-Bxpu1dcbGK1Wrae7mHeg_CM2rpNBc8/edit?usp=sharing
I'd suggest some sort of binary search would make sense, unless you feel we need to check every country below a rough level @westnordost ?