Number of areas with a specific site density not working

[Mono2]

The calculation of the number of areas with a specific site density does not work in the right way:

The area marked in the picture counts as 4 areas. I think they should count as one.

[RobinCGN]

Maybe delete / or not count polygon part that have an area equal to the cell size (your_grid_spacing, Pixel size).

[Mono2]

I think if we delete or not count polygon parts that have an area equal to cell size, we will loose (maybe) too much information.
I would suggest to create a new SpatialPolygonsDataFrame (isoline_merged). Within this SPDF polygons are merged, meaning that polygons of a certain site distance enclose the hole area and not only the area of that specific site distance.
Example: The new polygon of site distance 1000 is the old polygon of 1000 + the polygon of 500.

There will be still some errors and we could delete these remaining errors.

I have implemented this function.

What do you think?

[RobinCGN]

If this solves the number of areas problem. I'm fine.

But I had to change the script a bit, it corrupted the isoline_polygons spdf (only one feature left):

And is it correct that isoline_polygons merge contains 61 objects?

[RobinCGN]

Regarding the performance of the merge process: I have tried an alternative:

`sample <- isoline_polygons

x <- your_isoline_steps[-1]
iso_n = nrow(sample)
s <- NULL

for (i in seq(1:(length(your_isoline_steps)-1))){
s = c(s, x[i:(length(your_isoline_steps)-1)])}

st <- isoline_polygons[c(rep(seq(1,iso_n),seq(iso_n,1))),]
st@data <- cbind(st@data, x = s)
st@data
table(st$z)
st@data[st@data$x=="2500",]

isoline_polygons_merge_test <- raster::aggregate(st, by="x")
`

but your first approached performed much better:

[Mono2]

I am not sure if it solves the Problem.
I added a section, that after the merging only polygons with an area larger than the cell size are counted (this is in 02_ODI_selection).
Because I am not quite sure what the best solution is we now have two entries in Isolines_stats one for the "raw" data or the old method (number_Area) and one for the merged and omitted (cell size) polygons (number_Area_merged).
All other statistical properties are based upon isoline_polygon and not upon isoline_merged.

And is it correct that isoline_polygons merge contains 61 objects?

When I run the script isoline_merged only has 60 objects as it is the case for isoline polygons

[Mono2]

Playing with my Rössen data, I have recognized that the grid spacing also affect this problem. If a grid with a very fine grid spacing is chosen this problem does not occur.

[RobinCGN]

In my opinion experimental parts like "Merge polygons" should be moved to a fork and once they are ready can be merged with master. Would that be okay?

This is of course not a very urgent issue, but every time i run "01_kriging.R" it takes ages because of "Merge polygons" or I have to comment it out.

[RobinCGN]

I have revised the "Merge Polygons" Function. It is a bit faster now. Number of Polygons is the same as before. Please review if you got the time.

[Mono2]

I like your new update on "Merge Polygon". Introducing the possibility of using it or not was exactly what I was thinking about! I, however, prefer to use it and I would always recommend to use it!