Big AoI for buildings after river surface is not cropped

[fnattino]

With #111 the river surface is no longer cropped, which means the AoI for buildings can become quite large.

For instance:

delineations <- delineate("Budapest", "Danube", corridor = TRUE, segments = TRUE, riverspace = TRUE)

takes a long time to download and process buildings, and it leads to the following riverspace:

I guess we should clip the river surface as the river centerline before calculating viewpoints

[fnattino]

Actually the size of the AoI for retrieving building is properly constructed, it is just the geometry used to generate viewpoints that need to be fixed.

[fnattino]

Not sure whether the problem is still related with building retrieval, the error that I currently get is the following

delineations <- delineate("Budapest", "Danube", riverspace = TRUE)

The default `network_buffer` of 3000 m is used for corridor delineation.
The default `buildings_buffer` of 100 m is used for riverspace delineation.
Saving data to cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_waterway__18.925_47.35_19.335_47.613.rds
Saving data to cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_natural_water_18.925_47.35_19.335_47.613.rds
Invalid geometries detected! Fixing them...
Saving data to cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_highway_motorway_trunk_primary_secondary_tertiary_motorway_link_trunk_link_primary_link_secondary_link_tertiary_link_18.905_47.323_19.143_47.64.rds
Saving data to cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_railway_rail_18.905_47.323_19.143_47.64.rds
Saving data to cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_building__18.944_47.349_19.105_47.614.rds
Saving data to cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/dem_Copernicus_DSM_COG_10_N47_00_E019_00_DEM_Copernicus_DSM_COG_10_N47_00_E018_00_DEM_18.872_47.3_19.177_47.663.rds
Linking to GEOS 3.13.0, GDAL 3.8.5, PROJ 9.5.1; sf_use_s2() is TRUE
Calculating viewpoints from both river edge and river centerline.
Error in scan(text = lst[[length(lst)]], quiet = TRUE) : 
  scan() expected 'a real', got '5251366.1175697632.'
In addition: Warning messages:
1: to_spatial_subdivision assumes attributes are constant over geometries 
2: package ‘sf’ was built under R version 4.4.1 
3: In cap_corridor(c(edge_1, edge_2), capping_method, network) :
  Corridor capping gives multiple polygons - selecting the largest one
4: In NextMethod() :
  number of items to replace is not a multiple of replacement length
Error in (function (msg)  : 
  TopologyException: side location conflict at 350348.42869537754 5251366.1175697632. This can occur if the input geometry is invalid.

[cforgaci]

@fnattino, what I see now is slightly different from the previous plot:

There is additionally a side branch of Danube. (It should not, but) could that be the cause of the problem?

[fnattino]

Not sure yet, I will have a look at this in more details!

[fnattino]

@cforgaci it turns out the problem is a bug in visor. Using the version from CityRiverSpaces/visor#17 solves the problem (see below). I think we can close this and make a new bugfix release of visor when that PR is merged?

> devtools::install_github("CityRiverSpaces/visor", ref = "fully-occluded-rays-fn")
...
> delineations <- rcrisp::delineate("Budapest", "Danube", riverspace = TRUE)
The default `network_buffer` of 3000 m is used for corridor delineation.
The default `buildings_buffer` of 100 m is used for riverspace delineation.
Invalid geometries detected! Fixing them...
Linking to GEOS 3.13.0, GDAL 3.8.5, PROJ 9.5.1; sf_use_s2() is TRUE
Calculating viewpoints from both river edge and river centerline.
Warning messages:
1: In read_data_from_cache(filepath) :
  Loading data from cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_waterway__18.925_47.35_19.335_47.613.rds
2: In read_data_from_cache(filepath) :
  Loading data from cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_natural_water_18.925_47.35_19.335_47.613.rds
3: In read_data_from_cache(filepath) :
  Loading data from cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_highway_motorway_trunk_primary_secondary_tertiary_motorway_link_trunk_link_primary_link_secondary_link_tertiary_link_18.905_47.323_19.143_47.64.rds
4: In read_data_from_cache(filepath) :
  Loading data from cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_railway_rail_18.905_47.323_19.143_47.64.rds
5: In read_data_from_cache(filepath) :
  Loading data from cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/osmdata_building__18.944_47.349_19.105_47.614.rds
6: In read_data_from_cache(filepath, unwrap = TRUE) :
  Loading data from cache directory: /Users/fnattino/Library/Caches/org.R-project.R/R/rcrisp/dem_Copernicus_DSM_COG_10_N47_00_E019_00_DEM_Copernicus_DSM_COG_10_N47_00_E018_00_DEM_18.872_47.3_19.177_47.663.rds
7: In cap_corridor(c(edge_1, edge_2), capping_method, network) :
  Corridor capping gives multiple polygons - selecting the largest one
> delineations$riverspace
Geometry set for 1 feature 
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: 338371.2 ymin: 5210070 xmax: 360334.6 ymax: 5298524
Projected CRS: WGS 84 / UTM zone 34N
MULTIPOLYGON (((339005.3 5220022, 338995.2 5220...

[cforgaci]

Thanks @fnattino! CityRiverSpaces/visor#17 is merged now. I will take care of making a patch release for visor.