I was wondering if there's any tutorial or scripts that can be used to directly downscale and correct the bias in a reanalysis data available as NetCDF such as those provided by CRU or GPCC?
Also, for the downscaling part, does this mean we're just interpolating to lower mesh or is there an actual algorithm used to downscale the data in CDO?
Thanks in advance!
Hi,
If you use CDO, it's just interpolation.
To downscale, you need a physical-based numerical model, such as WRF.
Hope this helps!
Cheers,
Kishore
With CDO, you will only be able to extrapolate the data over space, which is very different from what would be called a downscaling. Extrapolation will only be made from statistical averages based on the initial data. A proper dynamical downscaling should be made with a climate (or atmospheric I guess) model that uses the actual model physics to go from low to high resolution.
If you want to extrapolate/interpolate any global data on a, for instance, 1x1 regular grid, you can simply use "cdo remapbil,r360x180 file_in.nc file_out.nc".
If you want it to be 0.5x0.5, replace "r360x180" by "r720x360", and etc.
Hope that helps.
Cheers,
Simon
Hello!
With CDO you can just interpolate. It just regrids your data to higher reaolution by taking statistical avgerage of the value of each grid point. To perform statstical downscaling you can take the help of other softwares. There are various packages in Rstudio which can help. ClimDown, downscaleR are some of this.
Cheers,
Afifa
Hi Afifa Talukder ,
what Ahmed Albabily mentioned in the query looks like he is interested in downscaling the precipitation data. I seriously suggest against statistical downscaling, especially for precipitation. Though physical-based numerical models are better for downscaling, the present employed methods can not downscale the precipitation reliably.
Cheers,
Kishore
- Badges
- Science topic
- Similar topics
- Philosophy
- Moral Philosophy
- Free Will