An Evaluation of the Seasonal Caribbean Hydroclimate in Low and High-Resolution CESM and other CMIP6 Models

Abstract

Abstract The Caribbean and Central American hydroclimate is understudied and complex in part due to its data sparsity, varied topographies, and multi-faceted interactions with the tropics and mid-latitudes. Recent work developed a refined and comprehensive understanding of the observed hydroclimate that has yet to be explored in global circulation models. This study investigates the simulation of the Caribbean hydroclimate using a suite of station and gridded observational datasets, the Community Earth System Model version 1 (CESM1) at high (0.25x0.25°) and low (0.9x1.25°) resolution, CESM2 at low (0.9x1.25°) resolution, the Coupled Model Intercomparison Project phase 6 (CMIP6) High-Resolution Model Intercomparison Project (HighResMIP), and the Geophysical Fluid Dynamics Laboratory Seamless System for Prediction and Earth System Research (GFDL-SPEAR). The simulated climatologies (1983-2014) of the annual rainfall cycle and total moisture fluxes, and climatological correlation coefficients of sea surface temperatures (SST), sea-level pressure (SLP), and zonal/meridional low-level winds onto indices of seasonal Caribbean rainfall totals are calculated to investigate inter-model differences. Generally, fully coupled CESM, GFDL-SPEAR-MED, and CMIP6 simulations underestimate precipitation across the Caribbean, with some improvements using high-resolution (<0.5°) simulations. The underestimations are largest during the Early-Rainy Season (ERS; mid-April to mid-June). Coupled models also show a moisture divergence bias associated with a stronger/west-displaced North Atlantic Subtropical High (NASH), a weaker / southward displaced Intertropical Convergence Zone (ITCZ), and stronger Caribbean Low-Level Jet (CLLJ). Precipitation and large-scale dynamic biases in experiments with observation-based SSTs are smaller, regardless of their spatial resolution, suggesting SST biases in coupled models may contribute to precipitation and dynamic biases. The findings emphasize the importance of both high-resolution and accurate simulation of coupled dynamical interactions in global circulation models to accurately simulate the Caribbean’s hydroclimate, and, therefore, provide reliable future climate projections for the region.