This is the implementation and extension of the Joint Variables Spatial Downscaling (JVSD) algorithm - a new statistical technique of downscaling gridded dateset.
Joint Variables Spatial Downscaling (JVSD), a new statistical technique of downscaling gridded climatic variables, is developed for the purpose of watershed hydrological modeling and water resources assessment. The proposed approach differs from other existing statistical downscaling methods in that multiple climatic variables are downscaled simultaneously and in a consistent way to produce realistic climate projections.
In steps of bias correcting, the joint variables quantile-based mappings (also known as joint empirical cumulative density functions) are used to adjust differentiated time sequences for Global Circulation Model (GCM) outputs instead of treating the different variables individually. The spatial disaggregating procedure is also distinct to other approaches. The simulated large scale spatial patterns are matched with observed small scale patterns directly instead of using any spatial interpolation approaches or random field generating techniques.
Analysis and comparisons are made by applying the proposed techniques on outputs of GCM simulations for 20th Century Climate in Coupled Models (20C3M), which are broadly available for most GCMs. Results show that the proposed downscaling method is able to reproduce the subgrid climatic features as well as the temporal/spatial variability within historical periods. Results from watershed hydrological simulations also indicate that the downscaled precipitation and temperature fields are consistent and thus suitable for hydrological modeling and for future regional water resources assessment of potential climate changes.
Read more: F. Zhang and A. Georgakakos, 2011, "Joint Variable Spatial Downscaling," Climatic Change, doi:10.1007/s10584-011-0167-9. http://link.springer.com/article/10.1007/s10584-011-0167-9