Currently that value can be get just in Surface Difference tool's report but to me smartest place would be simply in Spatial Modeller's Statistics (Global-Statistics) function
Having that and maybe other similar factors in that so that all possible Statistical values can be exported as a single numbers would be easiest for end user. So do not stick just my LE90 need but think more generally capabilities of Statistics function and add there as many statistical parameters as you easily can.
Not sure that the Statistics operator is the correct place for an LE90. Statistics takes a single raster stream (which can potentially be of multiple bands) and calculates statistics based on the DNs of each band. It's totally generic to the type of input data (other than the input is generally assumed to be athematic).
Whereas to generate an LE90 you would have to assume that the input represents the difference of two DEMs. I.e. LE90 is not a generally applicable statistic.
However LE90 could be easily derived from the absolute height difference of two DEMs if you had, for example, a "Get Raster Values by Percentage" operator. With such an Operator you could set it to tell you the 90th percentile DN value (i.e. the value below which 90% of the height difference values fall).
Would that fulfill your requirement (and provide flexibility to perform other operations)?
I absolutely agree with you. Making statistics just for some value like 90 given here is too limited solution as percentage could be anything else. So your suggestion is exactly what I need. I can run image difference easily but at the moment I found no easy way return data that falls under certain percentile in Imagine. So I might want 70 or 95 and that should be users own choise.
So the function you suggested would solve this problem totally. SO my vote comes for Get Raster Values by Percentage operator in spatial modeller. I think this functionality lies not in Imagine anywhere easily yet - unless you build your own model for it - Am I right?
The preferreed solution would be to add a "Get Raster Values by Percentage" operator, which thereby provides a more generic capability.
Delivered as part of ERDAS IMAGINE 2018, released March 1st, 2018
https://download.hexagongeospatial.com/downloads/imagine/erdas-imagine-2018