SpatialMapsComparison

Reproducible workflow in R using spdep to compute Bivariate Moran's I for grid-cell data.

Add spatial data to be analysed in folder dataInput.

Input data requirements: It can be raster or csv or shapefile.

If input data is Shapefile with point geometry, you should have values to compare in a single shapefile (e.g., posES and negES).

• grid.shp — geometry
• grid.dbf — attribute table with NegES and PosES columns' values
• grid.shx — shape index
• grid.prj — coordinate reference system (CRS)
• No missing values (or handle them explicitly) (zeros are accepted)

Analytical logic:

The analysis follows three sequential steps of increasing spatial specificity:

1. Co-occurrence (OR condition to cells selection) — do NegES and PosES share the same space?
2. Pearson correlation (AND condition to cells selection) — within shared space, are values correlated?
3. Bivariate Moran's I (AND condition to cells selection) — is that correlation spatially structured?

Summary of steps in script bivariateMoranIndex.Rmd, with indication of code chuncks that need user minimum edits (e.g., file name or tailor options for analyses):

1. Input data — EDIT THIS
2. Load & reproject
3. Choose neighbour method — EDIT THIS
4. Co-occurrence summary (OR condition — descriptive)
5. Pearson correlation (AND condition)
6. Build spatial weights (AND condition)
7. Global Bivariate Moran's I
8. Local Bivariate Moran's I (LISA)

LISA maps produced will be automatically saved in folder outputs, in 2 formats:

• TIFF (300 dpi, CMYK-compatible) and
• PDF vector format, ideal for resizing without quality loss.