Hi again,
101 views and no reply, is that because:
1. The answer is obvious and the question embarrasing ;
2. It is the wrong question ;
3. It can't be done ; or
4. Nobody knows how to?
Either way, I have tried to solve it myself and perhaps my solution will spark an interesting thread:
(i) Use "topographic modeling" under the 'Topographic' menu to generate an image containing one band that represents slope. By default this results in an image with pixels the same size as the input DEM.
(ii) Open the vector file representing the area for which you wish to calculate the actual ground surface area for, and export it as a region of interest (ROI) using the 'File' menu in the "Available Vectors List" browser.
(iii) Open the "ROI Tool", select the ROI created above and in the 'File' menu, select "Output ROIs to ASCII...".
(iv) Select the file generated in step (i) above as the input file for the ROI data, click ok, then in the next dialog box, make sure you click "Edit Output ASCII Form..." and for the band values; set precision to 1. Note: If you deselect everything but band values, you will end up with a much more manageable text file with each row holding the slope value for a pixel. Setting the precision to 1 means that you extract the values for all pixels within the ROI.
(v) Open the text file as a comma delimited file in excel and calculate the actual ground surface area represented by each pixel using the below formula, then simply sum them all to get an approximation of the ground surface area of your ROI:
x2 / cos(s), where x2 is the horizontal area represented by a pixel and s is the slope in radians.
- I would love feedback on how well you think this formula approximates the ground surface area, my concern is that as the aspect changes, while the north-south orientation of a pixel remains the same, the actual ground surface area is best represented by a parallelogram and not a rectangle as is assumed here and I might be underestimating the ground surface area.
Clearly, if there is a way (?) of calculating the sum of band values for a set of pixels within an image using, say, “band math” – then there would be no need for exporting the pixel values to excel.
Thanks,
Martin Elin
|