Hi,
Using gpprofile it is possible to compute a mean profile from a report file (right?), with a carefully choose file.report.
I would like to know how is it possible to also compute the standard deviation?
Thanks
regards,
Heloise
gpprofile: mean profile
Hi Heloise,
To produce ground models out of a report, with a misfit less or equal to 1:
Then, you can chain this output to gpprofile to extract Vs profiles, for instance:
This output a 'ready to plot' curve. For statistic computation, this is not exactly what we want. Instead, I suggest to re-sample the profile, let's say every 1 meter down to 50 meters:
Next we compute the statistics over this ensemble of profiles.
The grep and awk are mandatory to remove comment lines and to swap first and second columns. gphistogram constructs one histogram per x bin. Hence, if not swapping, you'll obtain one histogram for each velocity rather than for each depth.
Finally, you can extract the mean, median, mode,.. with its deviation thanks to option '-mean', '-median',...
To produce ground models out of a report, with a misfit less or equal to 1:
Code: Select all
gpdcreport my_run.report -gm -m 1
Code: Select all
gpdcreport my_run.report -gm -m 1 | gprofile -vs
Code: Select all
gpdcreport my_run.report -gm -m 1 | gprofile -vs -resample -d 50 -n 50
Code: Select all
gpdcreport my_run.report -gm -m 1 | gprofile -vs -resample -d 50 -n 50 | grep -v "#" | awk '{print $2,$1}' | gphistogram -y-min 150 -y-max 3500 -x-min 1 -x-max 50 -x-count 50
Finally, you can extract the mean, median, mode,.. with its deviation thanks to option '-mean', '-median',...
Code: Select all
gpdcreport my_run.report -gm -m 1 | gprofile -vs -resample -d 50 -n 50 | grep -v "#" | awk '{print $2,$1}' | gphistogram -y-min 150 -y-max 3500 -x-min 1 -x-max 50 -x-count 50 -mean
With latest snapshots (<200907), you will get all value twice out of gpprofile -resample. The next snapshot will be soon available. Meanwhile you can filter out this double values, for instance with the following script:
Code: Select all
gpdcreport my_run.report -gm -m 1 | gprofile -vs -resample -d 50 -n 50 | grep -v "#" | awk 'BEGIN{d=-1}{if($2!=d){print $2,$1;d=$2}' | gphistogram -y-min 150 -y-max 3500 -x-min 1 -x-max 50 -x-count 50 -mean
Re: gpprofile: mean profile
Hy Marc,
I am a new user with Dinver and I have been trying to get the mean profile of my inversion.
I have followed the command lines that you indicate here, but i can't use the grep and awk.
I am using Ms-dos and the code works just fine until this part:
Thank you for your time.
Francisco
I am a new user with Dinver and I have been trying to get the mean profile of my inversion.
I have followed the command lines that you indicate here, but i can't use the grep and awk.
I am using Ms-dos and the code works just fine until this part:
Should i use another command line instead of Ms-dos? Is Matlab a better one? Or is it a code problem?Code: Select all
gpdcreport my_run.report -gm -m 1 | gprofile -vs -resample -d 50 -n 50 | grep -v "#" | awk '{print $2,$1}' | gphistogram -y-min 150 -y-max 3500 -x-min 1 -x-max 50 -x-count 50
Thank you for your time.
Francisco