Hi gourav panchal and welcome to the Coordinate Reference Systems Place!
To determine the accuracy of a DEM requires some "true" elevation values, like point locations you know are accurate so that you can measure the error of a DEM vs the true values.
The process would include the following process:
- Compute the root mean squared error (RMSe) between the DEM elevations and the "true" elevations at your reference point -- the RMSe is the "average error."
- Produce an error map that presents the RMSe values at each reference point.
- Use these reference points to create an interpolation (kriging, IDW, etc) to estimate errors at non-reference points.
Or a cross validation technique.
However, I am not sure if that is the question you are asking. In your example above you explain how you have delineated the boundary of a watershed per DEM. So you should have 4 DEM delineations. You want to know which one of these is the most accurate.
Well, I would assume you have calculated these directly from the DEM and therefore the accuracy of the watershed will be dependant on the resolution of the DEM (think 30m is worse because it is coarser than 10m) and the technique used to generate that DEM.
For example, ASTER is based on optical data and SRTM is based on radar interferometry (both have a 30m resolution). This has consequences on the surface produced. ASTER is typically noisy in areas of strong topography vs SRTM which can be noisy in areas of low topography. On the other hand Cartosat was compared with SRTM in a paper listed below where it was determined that the higher resolution did not matter much in the generation of a stream profile.
Essentially, I think you need to research the most appropriate DEM type based on how it was made as well as its resolution to make an informed decision on which watershed delineation is most appropriate for the question/analysis you are trying to answer. This in my opinion is not about positioning or vertical measurements but rather which is the most appropriate dataset being used on that note I have shared the question with the Analysis and Imagery and Remote Sensing
See this Comparison of ASTER to SRTM
https://opentopography.org/blog/comparison-aster-gdem-srtm
See this Comparison of Cartosat and SRTM
http://technical.cloud-journals.com/index.php/IJARSG/article/download/Tech-480/pdf
