Closed cjacobscrisioni closed 7 months ago
Adjusted network-based calculations, swapping start- and endpoints. See d450536..feedfc1. Very good performance now.
Webmercator: As all other calculation are done in WGS84 (LatLong) (with the exception of connect to road vector data, which is done in WebMercator), I think this operation should also be done in WGS84; for this, using latitude specific aspect ratio's has been filed as feature request at https://github.com/ObjectVision/GeoDMS/issues/673
Webmercator: As all other calculation are done in WGS84 (LatLong) (with the exception of connect to road vector data, which is done in WebMercator), I think this operation should also be done in WGS84; for this, using latitude specific aspect ratio's has been filed as feature request at ObjectVision/GeoDMS#673
@MaartenHilferink can you already share the latitude specific aspect ratio's / formula? So I can include them in the network-based approximation.
Concerning borders. @MaartenHilferink is working on something to include districts (https://github.com/ObjectVision/GeoDMS/issues/673)
@cjacobscrisioni
I'll add examples in /Preprocessing/GridDists/AccessGrids_grid/gggg/xxxx that refer to
/SourceData/RegionalUnits/Countries/per_grid
to determine zonal boundaries and use 25f as the configured value for the zonal boundary impedance
See https://github.com/ObjectVision/GeoDMS/releases/tag/v14.15.4 and latest commit of 2BURP
Tested all implemented versions, all working perfectly now. Still need to decide whether to stick with the network-based variant or one of the costs versions (will depend on calibration tests as well). Latitude-corrected versions show sizeable differences even in Denmark, indicating the relevance of doing so. Thanks!
Grid dist calculations