South American precipitation changes simulated by PMIP3/CMIP5 models in past and present climates

VERA CAROLINA; DÍAZ LEANDRO

Paris

Conferencia; Our Common Future Under Climate Change; 2015

Significant precipitation changes have been identified in many regions of South Americausing both, paleoclimatic proxy records of the last 700 years and instrumental observations ofthe more recent decades. Such changes are not clearly understood yet, besides that theylargely impact the socio-economic activities of the countries that the regions encompass. Theavailability of the climate simulations included in the Fifth Phase of the World Climate ResearchProgram-Coupled Model Intercomparison Project (CMIP5) and the Third Phase of thePaleoclimate Modelling Intercomparison Project (PMIP3) provides an excellent opportunity toassess the ability of current models in representing those precipitation changes. Therefore, thatassessment is briefly discussed here as well as the exploration of the main dynamicmechanisms that might explain them.The following experiments of the PMIP3/CMIP5 set were considered: the pre-industrialexperiment, which not include any external forcing; the Historical experiment, which wasobtained forcing models by both, natural and anthropogenic sources, observed between 1850and 2005; and the Last Millennium which which span the period from 850 to 1850 and isobtained considering the natural forcing estimated for that period. Multi-model ensemble means(MEM) were computed for each experiment over the periods under study.The Andes Mountains are one of the regions in South America in which paleoclimatestudies have been focused on. In particular, in both, the Altiplano, a high-level plateau (around3800 m), located in the Andes between 15oS and 21oS, and the subtropical Andes, located incentral Chile at around 33oS, wetter-than-normal conditions were identified during the 17 thcentury within the period known as Little Ice Age (LIA). On the other hand, drier-than-normalconditions were detected at both regions in the second part of the 20 th century in associationwith the more recent global warming period (GWP). Although MEM are capable of representingthe thermal changes in South America estimated in both periods, they do not properly representthe expected precipitation changes, except in the subtropical Andes during the GWP. The largeuncertainties associated with the model simulations might be due to their limitations inreproducing the regional precipitation, especially over complex topography. On the other hand,models seem able to represent the large-scale circulation changes that would explain theprecipitation changes observed at those two regions in a physically consistent way. It was foundthat in the Altiplano, wetter (drier) summers in LIA (GWP) seem to be related with a stronger(weaker) upper-level eastward zonal flow, well represented by the MEM. On the other hand,models consistently show that wetter (drier) winters in the subtropical Andes in LIA (GWP),seem to be associated with a stronger (weaker) westward zonal flow in the low troposphere,induced in turn by hemispheric changes related to a negative (positive) phase of the SouthernAnnular Mode (SAM).Southeastern South America (SESA), encompassing the most productive economic areas of6 different countries (Argentina, Bolivia, Brazil, Chile, Paraguay, and Uruguay), is another regionin which available records show a significant precipitation increase by the end of the XX centurynot only as as compared to that observed at the beginning of that century, but also to thatestimated by paleoclimatic records at around 700 years ago. Both precipitation and circulationchanges simulated by the models at that early period are highly uncertain. However, the MEMfor the Historical experiment are able to reproduce precipitation changes observed in SESAduring the GWP, although they are weaker than observed. In fact, most of the simulationsreproduce the right sign of the precipitation changes in SESA during that period. However,associated uncertainty ranges (due to both inter-modeldispersion and internal climatevariability), are still large. In addition, it was found that mean positive precipitation trends inSESA for the Historical experiment are statistically distinguishable from those obtained for thenatural-forcing-only experiment made over the 20 th century, which exhibit negligible meanvalues. Results allow concluding that the anthropogenic forcing has at least a partialcontribution in explaining the precipitation changes observed in both SESA during the GWP.