# recreation model - user inputs

coline
Member

Hi NatCap team,

I am trying to run the recreation model over a 4.500km² area located in the French Alps at a 1km² resolution, and I have a few questions regarding the user-provided inputs. If someone feels like helping me, it would be greatly appreciated :)

I tried to provide the Invest prompt with flawless shapefiles (I've tried to input shapefiles with all polygons merged as one, or all polygons as single part polygons, with absolutely no information in the attribute table beside the ID, and zero topology problem) but Invest still spends a lot of time "Attempting to fix the geometry of predictor xxx". Do you know what may be causing this? I'd like to avoid that delay...

Perhaps it comes from the "holes" in the shapefile? But if I have to provide a full shapefile, I will have to provide a reclassification table as well to specifiy yes/no polygons, won't I?

Also, I read in the forums (http://forums.naturalcapitalproject.org/index.php?p=/discussion/comment/1525/) and in the user guide that the model does not handle quantitative data in the input attribute tables. However it is also written that in case of point shapefiles the predictor is processed by counting the number of points per grid cell.

So I tried to bypass that problem by (1) running a fake regression to retrieve the Invest-processed grid, and then (2) produced point shapefiles instaed of polygons, where the number of points within each grid cellstands for the value of my input quantitative data (for instance to simulate population density, the number of points stands for the density of inhabitants within the grid cell; but I used it to simulate other quantitative information as well).

What do you think, can it do the trick or is it too dirty?

Thanks in advance for the help :)

Coline

I am trying to run the recreation model over a 4.500km² area located in the French Alps at a 1km² resolution, and I have a few questions regarding the user-provided inputs. If someone feels like helping me, it would be greatly appreciated :)

I tried to provide the Invest prompt with flawless shapefiles (I've tried to input shapefiles with all polygons merged as one, or all polygons as single part polygons, with absolutely no information in the attribute table beside the ID, and zero topology problem) but Invest still spends a lot of time "Attempting to fix the geometry of predictor xxx". Do you know what may be causing this? I'd like to avoid that delay...

Perhaps it comes from the "holes" in the shapefile? But if I have to provide a full shapefile, I will have to provide a reclassification table as well to specifiy yes/no polygons, won't I?

Also, I read in the forums (http://forums.naturalcapitalproject.org/index.php?p=/discussion/comment/1525/) and in the user guide that the model does not handle quantitative data in the input attribute tables. However it is also written that in case of point shapefiles the predictor is processed by counting the number of points per grid cell.

So I tried to bypass that problem by (1) running a fake regression to retrieve the Invest-processed grid, and then (2) produced point shapefiles instaed of polygons, where the number of points within each grid cellstands for the value of my input quantitative data (for instance to simulate population density, the number of points stands for the density of inhabitants within the grid cell; but I used it to simulate other quantitative information as well).

What do you think, can it do the trick or is it too dirty?

Thanks in advance for the help :)

Coline

Tagged:

## Comments