Suitability Analysis and Fuzzy Logic
General overview
Part 1
- How to deal with diverse cell size and CRS?
- What about the extent of the study area?
- i.e. Physical global datasets in degrees (WGS 84) [resolution: 0.9 arcsec]
- i.e. Socio economic local indicators in metres (Pseudo-mercator) [resolution: 1Km]
Resampling
What we need as output…
- We need to transform vector layers to raster…
- We need to use the same spatial resolution (cell size)…
- We need to clip and align to the same number of columns and rows…
Why?
Raster Algebra!
Part 2
- How to compare layers from different aspects?
- Ranges in data? Units? Relationship?
- i.e. Coronavirus: 106 cases per 10K people.
- i.e. Temperature: -20.7 celcius degrees.
Linear Fuzzification
What we need as output…
- Each layer has to be ‘fuzzified’ to a range from 0 to 1 …
- No more units, just how suitable each aspect is…
- At this point we should have already dealt with outliers and null values…
Why?
Raster Algebra!
Part 3
- Assign weights to each layer if needed.
- Apply fuzzy overlay operations to aggregate the layers.
- Filter regions with higher potential.
Raster Algebra
What we get as output…
- A suitability map with values between 0 and 1 (we usually normalize it)…
- Can be filter out according to a threshold…
- Can be classified according to a criteria…
General Overview
