Hi Saha,

there are many tools and many approaches in literature for bias correction.

I think you have to specify your needs and also your objectives. It makes a different if you want to debias precipitation or temperature, for instance, also if you want to work on a daily, monthly or annual basis.

What is your purpose, using the bias corrected data to drive climate impact models (if so, which ones?)?

Cheers, Patrick

Hi Patrick,

Many thanks for your reply. My purpose is to identify the difference between observation and model output and how to remove those bias in future climate projection. I am interesting with statistical method. However, I am neither climate scientist nor modeler. I am looking for tools or approaches to do this in simplest way. I have downloaded SDSM software, but I am not sure that it is the suitable tool or not?

Satha

It sounds that linear scaling approaches might be useful here. Basically, you derive correction values based on long-tern observation means, and correct your model data based on the correction values. Note that in case of precipitation, the correction values are factors, for temperature or other variables, these correction values are added to or subtracted from the model data. Please have a look in the following papers for more details. It is very easy to do, you can simply use excel to do these corrections.  

Leander, R. and Buishand, T. A.: Resampling of regional climate model output for the simulation of extreme river flows, Journal of Hydrology, 332, 487–496, doi:10.1016/j.jhydrol.2006.08.006, 2007.

Lenderink, G., Buishand, a., and van Deursen, W.: Estimates of future discharges of the river Rhine using two scenario methodologies: direct versus delta approach, Hydrology and Earth System Sciences, 11, 1145–1159, doi:10.5194/hess-11-1145-2007, 2007.

Hay, L. E., Wilby, R. L., and Leavesley, G. H.: A Comparison of Delta Change and Downscaled Gcm Scenarios for Three Mounfainous Basins in the United States1, JAWRA Journal of the American Water Resources Association, 36, 387–397, doi:10.1111/j.1752-1688.2000.tb04276.x, 2000.

There are many ways for bias correction-scaling,standarization, empirical quantile mapping, gamma quantile mapping etc. Please see the following paper for more details-

Article CMIP5 ensemble-based spatial rainfall projection over homoge...

Hi Saha,

There are also databases available with already bias corrected/downscaled GCM output if that is helpful. Here are a few sources with collection of different bias corrected outputs:

Geo Data Portal: https://cida.usgs.gov/gdp/

Bureau of Reclamation: http://gdo-dcp.ucllnl.org/downscaled_cmip_projections/dcpInterface.html#Welcome

NASA NEX (under datasets -> climate): https://nex.nasa.gov/nex/

Best,

Anne

Many thanks for useful information Anne, Is there any bias corrected/downscaled GCM output outside the United States? I was trying to look for my area of interest.in Geo Data Portal: https://cida.usgs.gov/gdp/

Hi,

To me, the simplest way in correcting bias is delta change approach. Many publications on this are available. Once you are familiar with big data of model output, it is not really difficult to do the mentioned approach.

For global corrected GCM data, you can try to find in this link:

http://rda.ucar.edu/datasets/ds316.1/

https://catalog.data.gov/dataset/bias-corrected-spatially-downscaled-monthly-cmip5-climate-projections

Hopefully, it is useful for you

Hi Saha,

It looks like at Geo Data Portal the only area outside the US is Canada. At the Bureau of Reclamation website they do have global 1/2 and 1 degree monthly downscaled output. And the NEX-GDDP dataset from the NASA website has global 0.25 degree daily output.

Anne

can anyone help me suggesting a tool for bias correction using quantile mapping

you can use this tool following the methods mentioned by Javed Akhter

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

I tried that tool, but it is not improving my data

Visakh S.

Try different methods in CMhyd.

Delta change may serve your purpose, if just interested in magnitude of changes.

These papers may guide more

10.1016/j.jhydrol.2012.05.052

Article Bias Correcting Climate Change Simulations - A Critical Review

Try all the bias-correction methods, some will work well for you. In a study area in Africa, Distribution mapping improved my data significantly.

You can try the CDF matching bias correction approach for minimising the systematic errors. You can refer to the publication below (You can also download the code)

Publication: Article Estimation of Soil Moisture Applying Modified Dubois Model t...

Code:

https://www.mathworks.com/matlabcentral/fileexchange/78784-cdf-matching-bias-correction-method-in-matlab

Alfred Awotwi I tried delta and distribution mapping both in CMhyd tool to correct bias, the problem is delta change is correcting my data too well, it basically made all historical precipitation and temperature data of different RCM same as my observed data so it become useless for me, then i used distribution mapping, it does not improve my historical RCM data that much. can you suggest what should i do?