I am trying to do bias correction for rainfall data using the ‘qmap’ package in R.
My daily observed data is collected from 1981 to 2014 at the point (station) scale. To downscale the future data, I extracted the variable for corresponding locations from GCMs, including historical and future period (1981-2100).
The observed and GCMs data are detected the relationship and estimate the parameters of downscaling future data. But when I do that using qmap package, the results are poor. The output data is aggregated to monthly scale, and evaluated the performance of different GCMs with root mean squared error (RMSE) and correlation coefficient (r). From the results, I found that the bias correction even decreased the r. Additionally, I also provide the boxplot for month and annual precipitation. The above results are based on the fitQmapRQUANT method in the ‘qmap’ package. I also tried to other methods (fitQmapDIST, fitQmapPTF, fitQmapQUANT, and fitQmapSSPLIN), however, the results are still poor.
Does anyone get such results or I am doing it in the wrong way? How can I apply "qmap" to downscale daily precipitation data with my observed data?
Thanks in advance
Can you check these articles that might give you some direction
Article Climate change projections for the Texas High Plains and Rol...
Thesis Assessing the impacts of climate change on cotton production...
Hi, Naga Raghuveer Modala
Thanks for your answers.
I have read the articles you mentioned, and that's pretty good. The studies give me some ideas to improve the bias correction for daily climate data.
Thanks again.
Sincerely
you may check also the following paper (available in RG):
To bias correct or not to bias correct? An agricultural impact modelers' perspective on regional climate model data
Dear Jiongchao Zhao ,
Would you please summarize here some of the keypoints which helped you in improving the result of the bias correction ? As it might be really helpful to some others out there.
Regards,
Roland.
Hello. To the best of my knowledge and based on many studies, RQUANT is the best method of Quantile Mapping to correct the bias of rainfall data. But same as you, the accuracy of forecasts in my study didn't improve using Quantile Mapping and that means it decreased the accuracy.
If possible applie Quantile Mapping in daily value of precipitation, fir extreme value. For mean regimen use delta method
- Badges
- Science method