achieving a weighted inpolygon function

Hi,

I have a n x m array, which represents a space of discrete cells, all equal to each other and with a defined spatial size, dx x dy. In each entry of the array I can have a certain parameter, that I will later need to evaluate.

I then have a boundary curve, defined by x and y vectors, that I use to define a study region, closed and spatially homogeneous with the sizes of cells, so that my coordinates are in a range of 0 to n*dx and 0 to m*dy.

Creating the meshgrids, I can succesfully use the inpolygon function to obtain the matrix of internal cells.

What I would also like to obtain is a matrix of weighted membership of each cell to the region defined by the boundary curve, so that I have a number from 0 (fully outside) to 1 (fully inside) that gives me essentially the percentage of area included inside that region for each cell.

What are the possibile ways to achieve that?

Thank you in advance

Answer 1

[Matt J]

If the cells and the study region are represented by polyshape objects,

https://www.mathworks.com/help/matlab/ref/polyshape.html

you can use the polyshape.intersect() and polyshape.area() methods to find the intersection areas, and percentage areas.

[PIETRO DEVÒ]

This definitely did the trick!

Being short: starting from an NC file, with longitude, latitude, time and my parameter (precipitation), I produced a normalized x-y cell system of my problem, like that:

Where there is the little polyshape of my study region.

I essentialy used inpolygon function to identify firstly the "strictly included" barycentres of cells, then a combination of inpolygon to relative offsets of study region to add the "partially included" barycentres, obtaining:

Now, indexing that involved cells in a vector let me iterate cells in the original array of data to extract the parameter of interest. The intersect and area function, applied to each cell-region combination, evaluate the weight, achieving the objective.

Thank you very much.

Answer 2

[Image Analyst]

Not sure I understand. A diagram would help.

All I can guess is that the bwdist() function, to get the Euclidean distance from a point to the edge/boundary, or the regionfill() function might help.

[PIETRO DEVÒ]

Thank you for the answer, and sorry for not being clear.

This is an interesting method, I will try to implement also that way!