Understanding the causes of the biases that determine the onset of the rainy season in Amazonia in climate models using GoAmazon-CHUVA measurements

Abstract

The onset of the Amazon rainy season has a large temporal and spatial variability, and with this a strong impact on variability of the distribution of aerosols, ecosystems, fire and carbon fluxes, dry season length, agriculture, and hydropower. Changes of the length of the dry season, and increased dry season resilience due to CO2 fertilization, are the two leading factors from climate change for determining the risk of Amazon rainforest die-back in the 21st century. Two major droughts occurred in the region in the last 20 years (2005 and 2010) and the dry season has increased in length by about one month. However, evaluations of the CMIP3 and CMIP5 global climate models suggest that most of these models underestimate the past variability of the rainy season onset, thus dry season length, and also project virtually no future change of these variables even when the models are forced by strong greenhouse forcing under the RCP8.5 scenario. Why the models underestimate the variability of the rainy season onset, and whether this bias implies an underestimate of the climate sensitivity of the dry season length to anthropogenic radiative forcing remain unclear. This proposal aims to explore use of the measurements provided by GoAmazon and Brazil CHUVA Field Experiments, along with regional model experiments, to explore the processes that contribute to this underestimation of the modeled variability of rainy season onset, with focus on the WRF and CESM from the US investigators and the Eta regional Model and the BESM (Brazilian Earth System Model) from the Brazilian partners. (AU)