Climate change models Climate change scenarios

3. Climate change scenarios

3.1 Climate change models

The Amazon has a critical role in the global carbon balance with high net primary productivity and as a huge carbon store, in both plant biomass and soil. It also plays a crucial role in the climate regulation and moisture recycling and transport in South America through its effect on the local and regional water cycle. Downscaling projections from Global Circulation Models for climate change in the Amazon indicate an increase in temperature ranging from 0.5 to 8 o C during the 21 st century and a reduction in precipitation varying between 20 and 50 depending on the IPCC emission scenario used Marengo et al., 2011c. More detailed studies using higher resolution climate change scenarios, at 40 x 40 km, derived from the regional Eta Model run with the boundary conditions of the HadCM3 global model CMIP3 model indicate important changes in climate in the region up to 2100, including rainfall reduction in Amazonia by about 30-40 and warming of about 4-5 o C Chou et al., 2011, Marengo et al., 2011c. This report assesses future climate risks for South America using the new projections from the models available at the CMIP5 Coupled Model Intercomparison Project phase 5. These models will be presented in the next IPCC report IPCC AR5 and are compared to the outputs of the CMIP3 models used in the previous IPCC report, IPCC AR4 in figures 9, 10, 11 and 12. Figures 9 and 11 show average temperature and precipitation changes for 2015-2034 from 15 CMIP3 models, while Figures 10 and 12 show the mean temperature and rainfall anomalies from 9 models of CMIP5. For CMIP3 models the A2 emissions scenario of high GHG emissions atmospheric CO 2 concentration is 435 ppm; IPCC, 2007 is used in the simulations and for CMIP5 models only one Representative Concentration Pathway RCP is shown; 8.5 Wm 2 the atmospheric CO 2 concentration in the period is 431 ppm. The projected temperature warming derived from the CMIP3 global models for Amazonia range from 0.5 to 3°C for 2015-2034, but all models show the same tendency, i.e. warming Figures 9 and 10. The analysis is much more complicated for rainfall changes Figures 11 and 12. Different climate models show rather distinct patterns, even with almost opposite projections. In sum, current GCMs do not produce projections of changes in the hydrological cycle at regional scales with confidence. That is a great limiting factor to the practical use of such projections for active adaptation or mitigation policies. The CMIP5 models project an even larger expansion of the South American Monsoon over southern Amazonia Kitoh et al., 2011. In this study, eight CMIP3 and CMIP5 models were compared to identify improvements in the reliability of projections, and while no significant differences are observed between both datasets, some improvements were found in the new generation models. For example, in summer CMIP5 inter-model variability of temperature was lower over north-eastern Argentina, Paraguay and northern Brazil in the last decades of the 21st century. Although no major differences were observed in both precipitation datasets, CMIP5 inter-model variability was lower over northern and eastern Brazil in summer by 2100 Blazquez and Nunez, 2012. On El Nino simulations and projections there are indications that ENSO may become more frequent in a warmer climate, however, the confidence is low because of large natural modulations of El Niño patterns, and there is no consistent indication of discernible changes in projected ENSO amplitude or frequency in the 21 st century in CMIP5 models. Furthermore, the study has found that there is robust evidence that the simulation of the ENSO has improved from CMIP3 to CMIP5, with several models now realistically simulating the ENSO frequency spectrum and amplitude in sea surface temperature. Both CMIP3 and CMIP5 models tend to do somewhat better Coelho and Goddard, 2009 at precipitation reductions associated with El Niño over equatorial South America. Figure 9: Climate change projections for 2015-2034 of near surface temperature anomalies C for 15 CMIP3 Global Climate Models with respect to each model’s average temperature for the base period 1961-1990 for emissions scenario A2. Source: IPCC-AR4, 2007. Figure 10: Climate change projections for 2015-2034 of near surface temperature anomalies C for 9 CMIP5 Global Earth System Models with respect to each model’s average temperature for the base period 1961-1990 for RCP 8.5. Source: CMIP5, 2012 and Sampaio et al., 2013 – to be submitted. Figure 11 - Climate change projections for 2015-2034 of precipitation anomalies mmday for 15 CMIP3 Global Climate Models with respect to each model’s average precipitation for the base period 1961-1990 for emissions scenario A2. Source: IPCC-AR4, 2007. Figure 12 - Climate change projections for 2015-2034 of precipitation anomalies mmday for 9 CMIP5 Global Earth System Models with respect to each model’s average precipitation for the base period 1961-1990 for RCP 8.5. Source: CMIP5, 2012 and Sampaio et al. – not published. To model the complex climate system a climate model requires a very large amount of computer resources, which places a limit on the number of calculations that can be made and hence the size of the grid. Grid boxes within a global climate model are currently fairly coarse - to the order of 100-300 km square. Even at this resolution they give a valuable picture of how large-scale changes may be manifest. But to see how country-level changes may occur, and how different levels of concentrations of greenhouse gases may affect any changes, there is a need for finer-scale information. One way this can be achieved is through increasing the spatial resolution of the climate model in the region of interest, such as South America, which is computationally feasible because of the limited size of the region. The finer spatial resolution allows a more realistic representation of features such as the coastline and mountains, and of smaller-scale atmospheric processes. Thus, a regional climate model should provide a better representation of a particular country’s climate than a global model. This is why we used the Eta regional model from INPE run into the HadCM3 global model, for the present 1961-1990 and future 2010-2100, for various realizations of the A1B emission scenario. Changes in rainfall and temperature in the South America region projected from the Eta- CPTEC high-resolution climate model over the 21 st century are shown in Figure 13. As we move through the century, the projected changes become larger. Over the South America domain, there are areas predicted to become wetter in the future and other regions that are predicted to become drier Figure 13a-c. On a finer scale, the Eta model also projects large percentage decreases in rainfall and increases in air temperatures over the Amazon, with the changes becoming more pronounced after 2040. For temperature Figure 13 d-f the projected warming in the tropical regions varies from 0.5 - 3 °C from 2010-40 to 6-8 °C by 2071-2100, with increases being largest in the Amazon region. In addition to changes in temperature, information about possible future changes in rainfall with its implications for water resources is critically important in climate change management decisions. The direct output from this particular model Figure 14 indicates substantial percentage decreases in summer December-February rainfall by the end of the 21 st century. However, decreases in rainfall are projected throughout the year, not just in summer. It is always important to put the results in the context of other model projections, and it should be noted that the HadCM3 driving model simulates strong drying over Amazonia over the 21 st century, while other GCMs do not. HadCM3 lies on the extreme drying end of the multi-model group of projections Marengo et al 2011 a. We can say that in general, CMIP models still show uncertainties in rainfall projections for Amazonia, but most of the models agree in rainfall reductions in eastern Amazonia. Eastern Amazonia is the region that shows more impacts due to the extremes of climate variability and climate change, and perhaps could be considered as a climatic “hotspot”. Figure 13 - Changes in rainfall a-c, and in air temperature d-f, °C in South America for December-January-February 2010-40 column 1, 2041-70 column 2 and 2071-2100 column 3 relative to 1961-90 derived from the downscaling of HadCM3 using the Eta-CPTEC 40 km regional model. Maps represent the mean of 4 of the 17 scaled regional projections of change. Source: Marengo et al. 2011c. Figure 14 - Projected climate change over Brazil and the Amazon, Sao Francisco and Parana river basins by 2011-40, 2041-70 and 2071-2100 relative to 1961-1990 associated with different levels of global warming and CO2 concentrations. Direction of the changes in rainfall is indicated by arrows, and the regional warming is also shown in the figure. Source: Marengo et al. 2011c.

3.2 Climate extreme events