hey guys. Could you help me ?
i am working with sar simulation terrain correction to know the value when geometry correction is done. at the end of processing there is a notepad that indicates the rmse value at the time of processing. but I am still confused in reading what is meant in the article. on the processing of one polarization there are 6 parse with different rmse values. what is meant by the parse? and why is the rmse value different? and i should use which one?
below I attach the notepad generated during processing
subset_3_of_S1A_IW_GRDH_1SDV_20180203T221729_20180203T221754_020448_022F63_96DB_Orb_Cal_Spk_residual.txt (733.7 KB)
hey guys. Could you help me ?
The initial 2000 GCPs are randomly placed in the image so you might get slightly different results with each run.
What you see are the remaining points where good statistical correlation between the images is found as well. They are then used for polynomial transformation of all images into the same geometry. The errors indicate how well the polynomial fits the input images at the corresponding point.
Thank you Mr. Abraun. If the Sar Simulation terrain correction on one of the VH polarization I run produces parse 0 to parse 5 with various RMSE. Which RMSE should I write on my discussion sheet later. Do I have to write the best RMSE values out of the five RMSEs out and assumed to represent all the RMSEs in this VH polarization processing. Or the five RMSEs become one unity?
Below The Parse after my processing and the RMSE
Parse 0 = RMSE mean 2,9
Parse 1 = RMSE mean 1.2
Parse 2 = RMSE mean 0.8
Parse 3 = RMSE mean 0.64
Parse 4 = RMSE mean 0.50
Parse 5 = RMSE mean 0.39
The interation is repeated several times in order to increase the RMSE of the final result. So the last GCPs remaining (often noch very much) and their RMSE are the ones being used for the transformations.
Thanks mr abraun with your answer.
I am also still confused as to whether there is any difference between doppler terrain correction range and sar simulation terrain correction. when should I use RDTC and when should I use SAR Simulation TC for processing?
if I see the results of both processing there is no significant difference between the two (both pixels and geoloaction).
They both have the same result but there are cases where each is more suitable:
For normal SAR data wich good orbit information and geocoding, Range Doppler Terrain Correction works the fastest and gives good results.
If your data lacks of good geocoding or you experience shifts in your data, you can apply SAR Simulation Terrain Correction which uses the variation in backscatter intensity caused by topography in your image to fit it to the earth’s surface by simulating the terrain effects based on the incidence angle.
Have a look at Figure 2c (https://www.gamma-rs.ch/uploads/media/2013-2_PixelArea.pdf) to see how this works. You however need pronounced landforms and a DEM of suitable quality. Otherwise there is no pattern to match in your image.