From the histograms of the the three DEMs, in SRTM-X there are distinct repetitive peaks. Do you have an explanation for these peaks?
Attached figure shows histograms for four different DEMs ( Photogrammetric DEM, ASTER GDEM, SRTM X DEM)
SRTM tends to over estimate the height of valley floors (Hayakawa and Oguchi, 2008)
I think you will not expect that DEMs obtained with different methodologies will have exact statistics. Each method have its own cons and pros. p.e. ASTER DEM is based on automatic detection of pixel brightness, and use to fail in very bright or uniform areas. STRM will probably have some problems in radar-shadowed areas. ASTER and STRM also have some problems detecting elevation in forest areas. In those places you have the trees DEM = tree's height + elevation. Photogrametric method could probably be much precise.
There are another parameter that affects the values that is pixel or cell size. If you want to compare those DEMs among themselves a further correction must be done. A correct comparison between DEMs must involve the standarization of values, that means to use same pixel size. In order to do that I suggest to reduce information using majority filter.
SRTM detects height values from Radio detection and Ranging (RADAR), which literally translates to measurement of the time it takes for the Radio or microwave to travel from the antenna to the ground and back to the receiver, it simply measures distance, and such distance can be used to determine relief of an area particularly in forested areas more accurately when compared with ASTER as RADAR can penetrate forest canopy to get to the terrain. This can be the reason for the difference in peaks for the three histograms. Also various factors influence the accuracy of all three sensors, perhaps you should take those in to consideration.
Hi, It looks like a bi-modal histogram... This is a typical situation for a histogram of SRTM-X, SRTM-C and ASTER, when the terrain is partially under forest and partially not. One pick in the histogram represents bare land, and the another represents forested area. This is because SRTM-C/X and ASTER do not represent terrain elevations in forested areas. There is always a positive bias in forested areas!
Regarding the photogrammetric DEM: was tree heights subtracted from this DEM? Most likely not. So, this is also a DSM, and not DTM.
BUT, considering the elevations shown on your histograms, the origin of the two local picks most likely correspond to two types of mountainous terrain presented in your AOI. See attached figure and the publication. Regards.
Article Assessing Global Digital Elevation Models Using the Runway M...
Geomorphologically these are planar surfaces expressing erosion surfaces.
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing