I am looking at the spatial patterns of 7 different landcover types and would like to know if they are arranged randomly or not. I was hoping I would be able to run Moran's I on my data but have a few concerns:
- I originally had a raster map with 7 different landcover types, which I converted to polygons
- The only values associated with each landcover type are the codes denoting each of the 7 types (1, 2, 3, ..., 7)
- Moran's I computes the mean and variance for the attribute being analysed to get Moran's I Index value
If I run this spatial autocorrelation tool on my data, I am concerned that I will not get an accurate output, as the files only have code values, not a measurement of anything. The values I get are:
- Index: 0.026753
- ZScore: 4.32751
- PValue: 0.000015
An index value of 0 indicates randomly spaced attributes. A postive value indicates a tendency for clustering. Since I have a significant pvalue, I can reject H0 (that feature values are randomly distributed across the study area). And since my pvalue is statistically significant, then my positive zscore tells me that my features are more clustered than would be exptected if underlying spatial processes were random. Am I interpreting this correctly?
More importantly, are these calculations valid with my data type since I only have code values (1 to 7)?
