Reiner Palomino-Lemusa, b, Samir Córdoba-Machadoa, b, Sonia Raquel Gámiz-Fortisa, Yolanda Castro-Díeza, María Jesús Esteban-Parraa,

a Department of Applied Physics, University of Granada, Granada, Spain

b Technological University of Chocó, Colombia



In this study, statistical downscaling (SD) models have been built using principal component regression (PCR) for simulating summer precipitation in Colombia during the period 1950–2010, and climate projections have been made for the period 2071–2100 by applying the previous SD models to the SLP outputs of five GCMs. For this, the principal components (PCs) of the SLP reanalysis data from NCEP were used as predictor variables and the observed gridded summer precipitation as predictands. The period 1950–1993 was used for calibration and 1994–2010 for validation. Bootstrap with replacement was applied to provide estimations of the statistical errors. All SD models performed reasonably well at regional scales, and the spatial distribution of the correlation coefficients between the predicted and observed gridded precipitation values shows high significant values (between 0.5 and 0.93) along the Andes range, north and north Pacific of Colombia.


The ability of the GCMs to simulate the summer precipitation in Colombia, for the present climate (1971–2000), has been analyzed by calculating the differences between the simulated and observed precipitation values, with the result that the precipitation simulations made for the GCMs show strong biases. However, SD models applied to the SLP output from GCMs demonstrate their ability to faithfully reproduce the rainfall field. Finally, for summer precipitation projections in Colombia for the period 2071–2100, the SD models, recalibrated for the total period 1950–2010, have been applied to the SLP output from GCMs under the RCP2.6, RCP4.5, and RCP8.5 scenarios.


The SD estimations show considerable differences with respect to SD present values, generally towards precipitation increases. The SD MIROC5, HAdGEM2-AO, and CESM1(CAM5) present significant changes in all the regions for both the RCP4.5 and RCP8.5 scenarios. So, for the RCP8.5 these models project changes between 12.85% and 18.32% for the NWC region, 6.73% and 10.02% for the SWC region, 6.16% and 14.76% for the NCC region, and 8.31% and 23.11% for the NC region. This last region, that presents the greatest increase (23.11%) for the SD MIROC5 also presents the greatest decrease (− 7.8% under the RCP8.5) for SD MPI-ESM-LR. For the RCP2.6, significant increases are found only for the NWC region with SD HadGEM2-AO (4.5%) and for the NC region with the SD CESM1(CAM5) (8.33%).


Reiner Palomino-Lemus, Samir Córdoba-Machado, Sonia Raquel Gámiz-Fortis, Yolanda Castro-Díez, María Jesús Esteban-Parra.Summer precipitation projections over northwestern South America from CMIP5 models, Global and Planetary Change, Volume 131, August 2015, Pages 11-23, ISSN 0921-8181.


Energías Renovables y Meteorología