In tropical regions, particularly in Central and South America (CSA), the projections of climate seasonality under climate change are still uncertain. This is especially true for ecologically-relevant variables such as precipitation and temperature. However, assessments of model-based projections of seasonal climate for this region are scarce. We analyzed the simulation of seasonal precipitation and air surface temperature in CSA and six sub-regions within from 49 models included in the Coupled Intercomparison Project Phase 5 (CMIP5) and 33 models from CMIP6. In general, continental patterns and seasonality of both variables are moderately well resembled, while most models show systematic biases over the oceans, producing unrealistic spatial patterns. To quantify how well CMIP5/CMIP6 models simulate these variables, we used Taylor diagrams with respect to TRMM for precipitation and ERA5 for temperature.Precipitation shows the largest spread among models. Conversely, temperature shows a better simulation. CMIP5/CMIP6 models exhibit a better performance simulating both variables during December-January-February and March-April-May than during the other seasons. This is partly due to the reduced model biases in representing the Intertropical Convergence Zone during these two seasons. In general, biases are reduced in the CMIP6 models with respect to CMIP5. Regarding regional evaluations, precipitation patterns for Mesoamerica, Cerrado and Chaco regions are better reproduced compared to TRMM, while the annual cycles for the Andes hotspot, Central Chile and Guianas are not well simulated, mainly during their wet seasons. However, these biases are reduced in CMIP6 models. In regard to precipitation projections, models only agree over most of the regions with decreasing precipitation. Conversely, temperature exhibits a general consensus on persistent warming even during the historical period, with an average increase of 6 C by the end of the century, according to the CMIP6 models.

Northern South America is among the regions with the highest vulnerability to climate change. General Circulation Models (GCMs) are among the different tools considered to analyze the impacts of climate change. In particular, GCMs have been proved to provide useful information, although they exhibit systematic biases and fail in reproducing regional climate, particularly in terrains with complex topography. This work evaluates the performance of GCMs included in the fifth and sixth phases of the Coupled Model Intercomparison Project (CMIP), representing the annual cycle of precipitation and air surface temperature in Colombia. To evaluate this, we consider different observational and reanalysis datasets, including in situ gauges from the Colombian Meteorological Institute. Our results indicate that although the most recent generation of GCMs (CMIP6) show improvements with respect to the previous generation (CMIP5), they still have systematic biases in representing the Intertropical Convergence Zone and elevation-dependent processes, which highly determine intra-annual precipitation and air surface temperature in Colombia. In addition, CMIP6 models have larger biases in temperature over the Andes than CMIP5. We also analyze climate projections by the end of the 21st century according to the CMIP5/CMIP6 simulations under the highest greenhouse gases emission scenarios. Models show projections toward warmer air surface temperatures and mixed changes of precipitation, with decreases of precipitation over the Orinoco and Colombian Amazon in September-November and increases over the eastern equatorial Pacific during the entire year.

No abstract