regarding bias correction please see my answer to another question here, and hope this is helpful

best wishes

Pete

https://www.researchgate.net/post/In_hydroclimatology_researchers_often_compare_GCM_future_run_with_historical_observed_or_GCMfuture_run_with_GCM_control_run_Which_should_we_select#5937e264404854a6340b62fc

Thank you Sohrab, Nasrin and Pete for your responses......Thanks for the link Pete

Dear Pauline,

There are different tools you could use.

1. SDSM

2. Climdown package in R software

3. Cmhyd

4. Another MATLAB package . see https://meteo.unican.es/trac/ MLToolbox/wiki/Downscaling

or this https://github.com/FIDS-UWO/climate

All those are freely available

You can use CMhyd tool. The guide book and example data is available at

https://swat.tamu.edu/software/cmhyd/

Thanks guys.

Pauline Onjira Is there a need to Downscale RCM (CORDEX) data for our region?

Downscaled RCMs output is available at different CORDEX domains. But, you may need to bias correct the RCMs output.

Thank you Gebrekidan Worku Tefera

Gebrekidan Worku Tefera I have worked on the 0.5 degree resolution of observed data and Crodex provides the data of 0.44 degree resolution. can I use this data directly or I have to covert Cordex data to 0.5 degree? If yes, please suggest the methodologies involved there to convert them.

What matter is the resolution of your observed or re-analysis data. You can re-grid your RCMs data into the resolution you want. You can use the the bi-linear interpolation method on ArcGIS.

OK, Thank you Gebrekidan Worku Tefera

Gebrekidan Worku Tefera Dear Sir, Can you suggest any tutorial or way to re-grid the RCM data by Model building on ArcGIS or other than ArcGIS approach? I have number of data which can not regrid easily.

Hi Ankur,

the climate projection of RCM's are usually in netcdf format. There is a software package used by the community called Climate Data Operators (CDO) that you can use to regrid (among other stuff) any gridded data in necdf format. https://code.mpimet.mpg.de/projects/cdo

As for bias adjustment of CMIP5/6 or CORDEX projections you can used the CDFt method as explained in our technical report available hereTechnical Report Bias Adjusting Climate Model Projections

Harilaos Loukos I have extracted .nc data to Excel for my Coordinates. Can any other method applied to regrid them? (Specially for my extracted data).

Sorry Ankur Vishwakarma, I wouldn't know ho to do this in Excel. I do not recommend using excel when dealing with climate gridded data.

OK. Thank you Harilaos Loukos sir.

Harilaos Loukos Sir, I have query regarding CDO,

If dataset in curvilinear form, only I have to change remapbil to remapbic. Are there other changes needed to be followed (steps mentioned in bilinear interpolation)?

After opening the source file (Intermediate file), I got the file description two times, one in starting and another one at the end of the source file, So, is there a need to do changes in both ones or only in starting one?

One last thing, how to find xsize and ysize for a particular grid?

Hi Ankur Vishwaka, you just need to apply either one (remabil or remapbic), and the grid description (xsize y size) is in the netcdf file meta data.. Check the CDO documentation. All the best

I got the process. Thank you for suggesting CDO tool Harilaos Loukos

Thank you Muhammad Ali Hafeez

Harilaos Loukos I am getting a message while remapping the data on CDO as, Warning (cdf_set_var): Inconsistent variable definition for lon!

Warning (cdf_set_var): Inconsistent variable definition for lat!

(see the attachment). But it remap my data. So is there any problem with this warning or i can ignore it?

Hi All,

CMhyd is handy for bias correction of GCMs and RCMs.

https://swat.tamu.edu/software/cmhyd/

Simple and effective. Can be applied for both netCDF and ASCII format.

Engr. Muhammad Shafeeque, I am new in climate change analyses and want to evaluate that climate change impacts on water resources. I have read about downscaling and bias correction of GCM and RCM outputs but i am a little bit confuessed that if we do Bias correction for example in CMhyd, Does it mean that we did the downscaling or for that we should do other processes?

Please some one kindly help and correct me

Thanks in advance

Tooryalay Ayoubi Bias correction is synonmous to downscalling. First, the historical data of the model is compared to observations and delta values generated. The values are then used to correct future data.When you plot observations against the corrected observations, similarity index approaches 1.

Dear Kiprotich Paul;

Thank you so much for the good explaination. now it is abit cleared for me.

I have another question related to monthly output parameters for every station. What does this parameters mean? is it also correlation between observations against the corrected observations?

I am getting a warning massage after processing in CMhyd and it says some factors are extraordinary high/low. I have seen some months have really high values in deffirent months. Please see the example below and guide me.

Gauge number, Month, Factor

8, 04, 9.0946

8, 05, 4.6426

8, 06, 2.7943

8, 07, 2.8596

8, 08, 3.7973

8, 09, 2.7744

8, 10, 5.3810

8, 11, 3.6014

8, 12, 27.5163

9, 01, 32.6008

9, 02, 52.2894

9, 03, 60.8641

9, 04, 24.1673

9, 05, 14.3154

9, 06, 5.9278

9, 07, 4.1358

9, 08, 4.2006

9, 09, 4.6616

9, 10, 34.9687

9, 11, 6.7301

9, 12, 239.6639

10, 01, 32.2601

10, 02, 52.5875

20, 01, 44.1314

20, 02, 119.9376

20, 03, 116.9407

20, 04, 152.9020

20, 05, 317.0263

20, 06, 214.6479

20, 07, 208.8585

20, 08, 53.2755

20, 09, 18.6081

20, 10, 135.0588

20, 11, 18.2240

20, 12, 2176.4583

21, 05, 1.6198

21, 07, 2.8002

21, 08, 1.7191

21, 11, 0.4996

22, 05, 1.6198

22, 07, 2.8002

22, 08, 1.7191

22, 11, 0.4996

23, 07, 5.7564

Thanks in advance for your help

Tooryalay Ayoubi First, I would suggest you evaluate the models before downscalling in CMhyd. These warnings always come when you're using linear scaling meaning there is high descrepancy between observations and the model data. You can opt for distribution mapping (DM) for both rainfall and temperature. Recent literature shows that DM is best for use in CMhyd. Note that CMhyd do not include many calenders. I noted that there was mismatch between NetCDF timesteps and the CMhyd dates read from the same NetCDF. You can use CDO to check if you NetCDF file calender is either 365_DAY or Proleptic_Gregorian. You'll get some errors if your data calender 360_DAY especially in CORDEX RCMs.

Hello sir, I am also using CMhyd tool for bias correction , I have taken my observed data from 1975-2005 which is used for overlapping , then my cordex data is from 2006-2099. my problem is output of bias corrected data is from 2005 to 2036 not 2099. can someone please suggest me what i am doing wrong, i am new to this. Thank you in advance for help.

Kiprotich Paul I Would like to thanks for your help and suggestions.

I evaluated my data in CDO, as i am using 3 RCM Models (with 4 GCMs) from CORDEX (Please see the table below) and all models shows 365 days without REMO2009 which is 365 days proleptic-gregorian.

No, RCM Model, GCM_Driving_Model, Calender, Time_coordinate

1 , RCA4, CCCma-CanESM2, 365 Days, 1825 steps

2 , RCA4, MIROC-MIROC5, 365 Days, 1825 steps

3 , REMO2009, M-MPI-ESM-LR, 365 Days (proleptic_gregorian), 1826 steps

4 , RegCM4-4, NOAA-GFDL--ESM2M, 365 Days, 1825 steps

When I used these models in CMhyd using distribution mapping (DM) , The first 3 models showed error ( see attached error message) but the last one (4) was good and doesn't showed error.

So, What is the reason for the error?

and, How can I correct the error?

Kindly help, it is very urgent. Thank you

Kiprotich Paul , I forgot to say that I am using WAS Domain (South Asia Domian) and my historical observed data is from 01.01.1986 - 31.12.2005 and I have five leap years (1988, 1992,1996,2000,2004) with 366 days. As the models shows 365 days, Does CDO include leap years?

Hello everyone.

Please how can I use the CORDEX data as .nc file in the CMhyd tool. How to convert it to different files readable by CMhyd?

CMhyd uses the meta information of a netCDF file to find the climate model grid cells that overlay the gauge locations and convert the precipitation and temperature data into millimeters and degrees Celsius, respectively. Finally, CMhyd extracts time series of the relevant grid cells by reading from the netCDF files.

Hi everyone.

Can somebody help me how to do so for monthly RCM data? Because CMhyd works only for daily data, So how can I extract and bias correct monthly RCM outputs?

Try CMhydr