Can Significant Trends in Surface Temperature and Precipitation be Detected over South America?

Abstract:

This paper explores the existence of significant trends in near-surface temperature and precipitation over the South American continent by using observational data and estimates of natural variability based on simulations with numerical climate models. Trends are computed from three observational datasets in the period 1975-2004 for temperature and 1955-2004 for precipitation. Significance of the trends is tested against the null hypothesis that they arise from natural variability alone, which is estimated from the output of a suite of CMIP5 pre-industrial control experiments. Trends obtained from the observational datasets are compared with those simulated by CMIP5 historical runs, in which observed external transient forcing is imposed, and with those from simulations with natural-only forcing. In the case of temperature, an overall warming trend is found over the entire South American continent (0.23 C per decade). Significant trends (at the 95% level) are found in a region that corresponds roughly to Brazil with maximum warming over the north-central part. The average trends over South America in the observations broadly agree with those in the CMIP5 historical simulations for all seasons. This agreement is less close for the natural-only forcing simulations. The maximum warming over north-central Brazil is generally underestimated by the models. In the case of precipitation, trends vary both in sign and intensity according to region and season. The only significant trends in precipitation are obtained in La Plata Basin. Over the southern part of the basin (south of the Tropic of Capricorn), a significant decrease in precipitation is found during winter (-1.6 mm/month per decade) and an increase in all other seasons (4.2 mm/month per decade during summer). Over the northern part of La Plata Basin, the only significant trend in precipitation is a decrease during winter (-1.2 mm/month per decade).