Comparison of different statistical downscaling methods for climate change rainfall projections over the Lake Victoria basin considering CMIP3 and CMIP5
MetadataShow full item record
In this study, outputs of three statistical downscaling (SD) methods including the change factor (Delta), simplified (simQP) and advanced (wetQP) quantile-perturbation-based approaches were compared based on daily rainfall series at 9 meteorological stations in the Lake Victoria basin (LVB) in Eastern Africa. The comparison was made considering phase 5 and phase 3 of the Coupled Model Inter-comparison Project, i.e. CMIP5 and CMIP3 respectively. For the CMIP5 (CMIP3) at each station, there were a total of 7 (14) GCMs, 18 (20) daily historical (control) simulations over the period 1961–2000, and 35 (49) daily future projection series of the periods 2050s and 2090s. The ensemble mean of the GCMs’ Bias in reproducing rainfall extremes for return periods in the range of 1 to 40 years for the CMIP5 (CMIP3) varied from −19.05% to 3.11% (−65.85% to −4.86%). For the high greenhouse gas scenario rcp8.5 (A2) of the CMIP5 (CMIP3), the ensemble mean of the projected changes over the LVB in the 10-year rainfall intensity quantile obtained from the Delta, simQP, wetQP SD goes up to 5.8, 10 and 22.4% (11.7, 15.9 and 43.6%) in the 2050s and 8, 11.4, and 25.4% (14.2, 23.3 and 40.6%) in the 2090s. Rainfall totals of the main wet (dry) season are generally projected to increase (decrease) in both the 2050s and 2090s. Because the outputs from the three SD methods captured well the pattern of monthly rainfall totals, the difference between the projected changes of seasonal or annual rainfall totals from the Delta, simQP and wetQP was shown to be insignificant. However, the differences in the results from the Delta, simQP and wetQP methods with respect to the projections of rainfall quantiles indicate that the choice of the SD method can be made on a case by case basis in line with the objectives of the climate change impact study, e.g. the Delta does not capture well the changes in rainfall extremes, whereas the wetQP is suitable for both rainfall extremes and rainfall totals at both seasonal and annual time scales. The findings of this study also show the need to consider evaluations of the inter-GCM differences in the LVB as a data scarce region in assessing the discernible impact of climate change on rainfall extremes and/totals for decision making related to water resources management and engineering.
- Natural Sciences