Jump to content

 
Photo

Slope having error generated by SRTM 90 m DEM

- - - - -

  • Please log in to reply
9 replies to this topic

#1
himanshupahava

himanshupahava

    Newbie

  • New Member
  • Pip
  • 6 posts
  • India

Hi,
I am having SRTM 90 m DEM. When i am generating slope in ERDAS 9.1 and ENVI 4.5 using this DEM, this slope image is having error in the form of strip line. Same problem i m facing for ASTER 30 and CARTOSAT DEM also.
Is anybody having idea why it is happening and how to remove this error. This error also comes fro Aspect image.

#2
frax

frax

    Hall of Fame

  • Associate Admin
  • PipPipPipPipPipPipPip
  • 2,310 posts
  • Gender:Male
  • Location:Stockholm, Sweden
  • Interests:music, hiking, friends, nature, photography, traveling. and maps!
  • Sweden

Not sure what you mean by "in the form of a strip line". Are you sure this is really an error... ?
Hugo Ahlenius
Nordpil - custom maps and GIS
http://nordpil.com/
Twitter

#3
Gretchen Peterson

Gretchen Peterson

    Master Contributor

  • Validated Member
  • PipPipPipPip
  • 238 posts
  • United States

Can you post a picture of the line? Do you have multiple DEM datasets that overlap at exactly this line?

#4
himanshupahava

himanshupahava

    Newbie

  • New Member
  • Pip
  • 6 posts
  • India

hi,
I have attacheded two sample slopes of same area generated by SRTM and Aster. Now u would have a better idea of my problem. Plz zoom in to have a better look of those lines.

Attached Files


Edited by himanshupahava, 12 January 2010 - 01:10 AM.


#5
honeybee77

honeybee77

    Newbie

  • New Member
  • Pip
  • 8 posts
  • United States

Have you done any resampling prior to calculating slope/aspect? I had a similar output happen when converted a DEM in lat/long coordinates to UTM coordinates and selected "nearest neighbor" for the resampling method. Once I selected bilinear interpolation or cubic convolution for resampling method, the slope result was much more pleasing.

#6
himanshupahava

himanshupahava

    Newbie

  • New Member
  • Pip
  • 6 posts
  • India

Have you done any resampling prior to calculating slope/aspect? I had a similar output happen when converted a DEM in lat/long coordinates to UTM coordinates and selected "nearest neighbor" for the resampling method. Once I selected bilinear interpolation or cubic convolution for resampling method, the slope result was much more pleasing.


Thanks..It worked but would you plz tell me why it is the case for all satellite data derived DEMs. I have contour generated DEM and slope n aspect of this DEM come good.

#7
honeybee77

honeybee77

    Newbie

  • New Member
  • Pip
  • 8 posts
  • United States

Thanks..It worked but would you plz tell me why it is the case for all satellite data derived DEMs. I have contour generated DEM and slope n aspect of this DEM come good.


I believe it has more to do with the technique used for resampling rather than the source of the data. When I had a similar problem, I was using radar derived DEM.

A possible explaination as to why this occured in your situation:
It may appear that the contour-derived DEM did not show this result because your beginning and ending coordinate systems were similar and thus nearest neighbor resampling did not have as large an effect on the results. However, in the case of the satellite derived DEM, perhaps you were changing to a completely different projection, with a new size and shape, so when nearest neighbor resampling was used, the results were poor. This is o nly a guess as I don't know your actual situation.

I hope that helped answer your question.

Best,
Melisa

#8
himanshupahava

himanshupahava

    Newbie

  • New Member
  • Pip
  • 6 posts
  • India

Thanks..It worked but would you plz tell me why it is the case for all satellite data derived DEMs. I have contour generated DEM and slope n aspect of this DEM come good.


I believe it has more to do with the technique used for resampling rather than the source of the data. When I had a similar problem, I was using radar derived DEM.

A possible explaination as to why this occured in your situation:
It may appear that the contour-derived DEM did not show this result because your beginning and ending coordinate systems were similar and thus nearest neighbor resampling did not have as large an effect on the results. However, in the case of the satellite derived DEM, perhaps you were changing to a completely different projection, with a new size and shape, so when nearest neighbor resampling was used, the results were poor. This is o nly a guess as I don't know your actual situation.

I hope that helped answer your question.

Best,
Melisa


Thanks Melisa, but if u get any technical doc related to this problem would u plz update me?? that would help me a lot..
Have a nice day

Regards,
Himanshu

#9
Crystal

Crystal

    Newbie

  • New Member
  • Pip
  • 2 posts
  • United States

This article may help explain or understand the artifacts that can occur in the data collection phase:

http://geomorphometr...morphometry.pdf

It is a possible explanation. I experience this type of issue with the much older USGS United States 90 Meter and some 30 Meter DEMs. It is often difficult to see the striping on the source data, but once you begin to create derivative products, such as slope, aspect, hillshade...those results show the striping much more prominently b/c of the propagation of error with the processing algorithms.

#10
honeybee77

honeybee77

    Newbie

  • New Member
  • Pip
  • 8 posts
  • United States

Crystal,
Although I cannot open the document you linked to (blocked on my work computer for some reason), I believe Himanshu is having the opposite problem, where the satellite derived DEMs performed more poorly than the contour derived DEMs. Of course, the possible errors in contour-derived DEMs is still good to know.

Himanshu,
This document may be a starting point for some information on the subject. You may have to do some searching on your own if you would like learn more.
ON SOME LIMITATIONS OF SQUARE RASTER CELL STRUCTURES FOR DIGITAL ELEVATION DATA MODELING
Ashton M. Shortridge and Keith C. Clarke
In: Spatial Accuracy Assessment: Land Information Uncertainty in Natural Resources
K. Lowell and A. Jaton (Editors). Ann Arbor Press: Ann Arbor, Michigan. 1999.
https://www.msu.edu/...s/rasterdem.pdf

The book Principles of Geographic Information Systems by P.A. Burroughs and R.A. McDonnell may be another valuable source as it was referred to in several searches I conducted.

Best, Melisa




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

-->