Project

General

Profile

Forums » Open discussion »

Downscaling simulated data on CDO

Added by ibrahim erkol over 3 years ago

Have a nice day everyone,
I need to downscale my simulated GCM data using Quantile mapping method.
Is there anyway I can conduct this process on CDO and if not, I'd be so glad if you shared with me a way that doesn't require code writing so much for performing quantile mapping downscaling/bias correction.

I have two nc files. One of them is observation data extending from 1980 to 2018 and the other one is simulated data extending from 1980 to 2100.

Thanks in advance

Best of luck


Replies (6)

RE: Downscaling simulated data on CDO - Added by Ralf Mueller over 3 years ago

Hi Ibrahim!

As far as I know, quantile mapping is not implemented in CDO, sorry.

cheers + a good weekend!
ralf

RE: Downscaling simulated data on CDO - Added by ibrahim erkol over 3 years ago

Hi Ralf,
Thanks for your kind return and everything.
Can you suggest any different way for performing Quantile Mapping downscaling which will not require me to write some lines of code from scratch.

Thanks in advance
Best of weekends :)

Ibrahim Erkol

RE: Downscaling simulated data on CDO - Added by Ralf Mueller over 3 years ago

sorry - the mapping technique is completely new to me. The method seems to related to histograms.

From what I understood so far it takes values and maps them into some classes, which are set up to cover certain value ranges.
So my first guess would be to analyze the histogram of the data set to come up with a useful classification, or setting of reasonable bins (using histogram vocabulary). Them I would

  • (a) mask out the data for each range using gtc and ltc operators with the corresponding values of the each class
  • (a) multiply these mask with a certain number uniq for each class
  • (a) use add for addition of all masks - by setting everything else to zero this should leave all classes intact (X-checks needed for validation)

or

  • (b) use the expr operator and come up with a very long ternary-statement, that covers all classes and
  • (b) sets the class identifiers right away

here is the docu about it.

hth
ralf

RE: Downscaling simulated data on CDO - Added by ibrahim erkol over 3 years ago

Thanks for the resource suggestions but this "there will be some coding involved" issue is a big deal for me. That's why I was looking another way around. Still, thanks a lot for the resources

Best regards

RE: Downscaling simulated data on CDO - Added by Brendan DeTracey over 3 years ago

Yes, but one of the above resources has a TK toolkit i.e. a GUI. I have not tried it but maybe it will help you.
https://github.com/AgroClimaticTools/CDBC

    (1-6/6)