I am using the topographic position index (TPI), which is derived from high-resolution DEM. This index shows the difference between a central point and the mean elevation in the neighbourhood around the point. I have this value for different radiuses, so for example for 100m, 200m etc till 5000m. I made a composite file (multiple band raster file) from the summation of TPI of several radiuses, to obtain TPI values ranging from 0 to 100. Thus every band represents the value of the TPI of a certain radius. In attachment I added a picture of the TPI of the meso-scale (100-1000m). The lower TPI values (in blue-green) show areas that are lower situated than the higher TPI values (in orange-red).
Is there a way to find/detect/identify patterns in this multiple band raster file?
I am interested particularly in depressions which actually show wetlands, as these TPI index shows infiltration-seepage patterns. Is there a way to detect these depressions (perhaps based on a threshold value, just thinking out loud)?
And how can you assign a 'signature' on such a kind of pattern, for example based on the values of the different bands? As you see in attachment the graph shows the TPI values of 1 pixel for different bands (radiuses). It can be interpreted as a reaction of the groundwater level on a rainfall event. Of course this is different for every pixel, but it would be nice to have something that 'characterises' a wetland/depression/the pattern that were found.
(I am using ArcGIS.)