> "The height of a building gives a lot of information about the people > living in that building" - please explain it. For me it sounds amusing
The information about the height is code as "number of floors" in the source file (catastro). Even more, you can make a fair guess making "3m = 1 floor" (as is what we are doing when exporting). Now, you have the "total surface" of the building multiplying the number of floors times the build surface area. Next, you have several possibilities depending of what you are doing and the information you have: - In the worst case you can estimate the mean number of person per square meter with the population data from the National Statistics Agency. - If you have information about the total number of person in the building you can "fain grain" the information. For example, in our work, we have a fair estimation of the number of individual houses that there is in a particular build, so we can have the person/per build information (this information in given by a enterprise and is private but national agency have this information, we are not ). - Now you have a estimation of the "local density" of the area. High density areas normally consists of tall buildings in the city centre while low density consists of suburbs areas. And now you can start characterizing the zones if you like. And most important of all, we are talking about GIS here, if you want or need it in the other way, you can create the union of these geometries. But you can't do it from backwards, from one building get the separated parts. > From the aerial images it looks like 2, maybe 3 free-standing houses for > a single family. It is very difficult to determine that from the > data you imported because the houses consist of 3 to 5 separate > building ways all of them tagged in exactly the same manner. You > will need to do a complicated geometric analysis and add a lot of > guess work to determine the actual number of houses. This could be a problem, but please note that we have the parcel information too. With this data you can make the fair guess that you will have one building per parcel. Note that the case with non adjacent builds is trivial as you can just take the union of all the polygon as the base surface and the highest height or, as we are only interested in the surfaces, just take them as different builds. Obviously with this kind of guessing you will make lots of mistakes. For example in adjacent houses but normally if this case is mapped as a whole block and you don have any clue of the number of houses or the total surface. Even more, this can be guessed using the parcel information an/or the policy number and stuff like that. Please note that, in this case we are not interested in the "number of houses" as this is public information (at least in the Spanish Statistic Agency). Of course, this type of information complicate the analysis BUT you can still make the same calculation your are doing now AND you can do lots of new things. -- Cruz Enrique Borges Hernández Email: [email protected] DeustoTech Energy Telefono: 944139000 ext.2052 Avda. Universidades, 24 48007 Bilbao, Spain _______________________________________________ Imports mailing list [email protected] http://lists.openstreetmap.org/listinfo/imports
