Dear all,
I am going to derive the precipitation data from NETCDF files of CMIP5 GCMs in order to forecast precipitation after doing Bias Correction with Quantile Mapping as a downscaling method. In the literature that some of the bests are attached, the nearest neighborhood and Inverse Distance Method are highly recommended.
The nearest neighbour give the average value of the grid to each point located in the grid as a simple method. According to the attached paper (drought-assessment-based-on-m...) the author claimed that the NN method is better than other methods such as IDM because:
"The major reason is that we intended to preserve the
original climate signal of the GCMs even though the large grid spacing.
Involving more GCM grid cell data on the interpolation procedure
(as in Inverse Distance Weighting–IDW) may result to significant
information dilution, or signal cancellation between two or more grid
cell data from GCM outputs."
But in my opinion maybe the IDM is a better choice because I think as the estimates of subgrid-scale values are generally not provided and the other attached paper (1-s2.0-S00221...) is a good instance for its efficiency.
I would appreciate if someone can answer this question with an evidence. Which interpolation method do you recommend for interpolation of GCM cmip5 outputs?
Thank you in advance.
Yours,
Hi Farzad,
Spatial interpolation of climate variables is not a trivial problem because the interpolation depends on many other characteristics of the problem (altitude, location, vegetation, etc). Modern techniques recommend the incorporation of a great deal of known gridded variables (i.e. auxiliary variables) in order to improve the results. In this case it is necessary to use advanced statistical techniques to find efficient, accurate, solutions to the problem. An example of such techniques can be found in:
https://www.researchgate.net/publication/257298539_Spatial_Interpolation_of_McArthur's_Forest_Fire_Danger_Index_accross_Australia_Observational_study
Hope this helps,
Augusto
Article Spatial Interpolation of McArthur's Forest Fire Danger Index...
Hi Farzad
I think IDM is the best method compared to other methods. It includes the weighting of other grids points when estimating climate variable at specific station. The nearest neighbor consider only one grid point but for climate variables with high spacial variability may not provide realist results. For the RCM I found that there are small difference between different interpolation methods, this mean there is no significant difference because we dont expert strong spacial variation between grid-point within the models due to little physics included in the models. Therefore for RCM one can use any interporation method. For GCM I think you can use the IDM to include the spacial variability of the variable from different grids to respective station.
Article Evaluation of the performance of CORDEX regional climate mod...
Dear Farzad Emami,
as to me the www.gmdhshell.com (sciense variant) are more sutable in yours case.
- Badges
- Science topic
- Similar topics
- Geoscience
- Atmospheric Sciences
- Climate Science
- Climate Change