Assessing the representation of teleconnective drivers of rainfall over Eastern Africa in global and regional climate models and projected future changes
Doctoral Thesis
2017
Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
University of Cape Town
Faculty
License
Series
Abstract
Climate variability is an important characteristic of regional climate, and a subject to significant control from teleconnections. An extended diagnosis of the capacity of climate models to represent remote controls of regional climate (teleconnections) is vital for assessing model-based predictions of climate variability, understanding uncertainty in climate projections and model development. An important driver of climate variability for Africa is the sea surface temperature (SST) - rainfall teleconnection, such as the El Ni˜no/Southern Oscillation (ENSO) and Indian Ocean Dipole (IOD). In this study, an assessment of the teleconnection between tropical SSTs and Eastern African rainfall in global and regional climate models is presented, with particular attention paid to the propagation of large-scale teleconnection signals (as represented by model reanalyses and Coupled Global Climate models (CGCMs)) into the domain of the Regional Climate Models (RCMs). The teleconnection-rainfall relationship with the Eastern Africa region is assessed in two rainfall seasons (June-July-August-September and October-November- December) under present and future periods. Evaluation runs (RCMs driven by reanalysis datasets) and historical simulations (RCMs driven by CGCMs) are assessed to quantify the ability of the models to capture the teleconnection relationship. The future analysis is performed for two Representative Concentration Pathway scenarios (RCP4.5 and RCP8.5) to assess future change in this relationship as a result of global warming. Using ERA-interim reanalysis as perfect boundary conditions, the RCMs adequately simulate the spatial and temporal distribution of rainfall in comparison with observations, although the model performance varies locally and seasonally within the region. Furthermore, the RCMs correctly capture the magnitude and spatial extent regional-scale seasonal rainfall anomalies associated with large-scale oceanic modes (ENSO and IOD). When the lateral boundary conditions are provided by CGCMs, RCMs barely capture the regional teleconnection patterns associated with large-scale modes, and mostly depend on the selection of the driving CGCM. Comparison of the CGCM-driven RCM simulations with the reanalysis-driven RCM simulations revealed that most of the errors in teleconnection found in the RCM simulations are inherited from the host CGCMs. The ERA-Interim driven downscaled results show better agreement with observed spatial teleconnection patterns than the CGCM driven downscaled results. Analysis of the CGCMs and corresponding downscaled results showed that in most cases both the CGCM and the corresponding downscaled simulations had similar teleconnection patterns, but in some cases the RCM results diverge to those of the driving CGCM results. It has been demonstrated that similarities in SST-rainfall teleconnection patterns between the RCM simulations and respective driving CGCM simulations are noted over the equatorial and southern part of the region during OND season, where the rainfall is primarily controlled by large-scale (synoptic-scale) features, with the RCMs maintaining the overall regional patterns from the forcing models. Di↵erences in RCM simulations from corresponding driving simulations are noted mainly over northern part of the domain during JJAS, which is most likely related to mesoscale processes that are not resolved by CGCMs. Looking at the model projections of the future, although the spatial pattern of teleconnections between ENSO/IOD and rainfall still persist, important changes in the strength of the teleconnection have been found. During JJAS, ENSO is an important driver of rainfall variability in the northern parts of the region where dry anomalies are associated with El Ni˜no and wetter anomalies with La Ni˜na. Both regional and global ensemble projections show higher rainfall during La Ni˜na and lower rainfall during El Ni˜no over the northern part of the region compared to the present period. During OND, the teleconnection between ENSO/IOD and rainfall is projected to strengthen (weaken) over Eastern horn of Africa (southern parts of the region) compared to the present period. This implies heavy seasonal rains associated with positive phases of ENSO and IOD will increase in future across the Eastern horn of Africa. The change OND rainfall teleconnections are stronger and also more consistent between the models and scenarios as compared to the change in JJAS teleconnections. These findings have an important implication for the water and agricultural managers and policies in the region to tackle the anticipated droughts and floods associated anthropogenic climate change. Finally, the analysis demonstrated that the largest source of uncertainty in the regional climate model simulations in the context of teleconnective forcing of rainfall over Eastern Africa is the choice of CGCM used to force the RCMs, reinforcing the understanding that the use of a single GCM to downscale climate predictions/projections and using the downscaled product for assessment of climate change projections is insufficient. Simulations from multiple RCMs nested in more than one GCM, as is undertaken in the Coordinated Regional Downscaling Experiment (CORDEX), are needed to characterize the uncertainty and provide estimates of likely ranges of future regional climate changes.
Description
Reference:
Endris, H. 2017. Assessing the representation of teleconnective drivers of rainfall over Eastern Africa in global and regional climate models and projected future changes. University of Cape Town.