Historical meteorological droughts over the CORDEX‐CAM (Central America, Caribbean and Mexico) domain: Evaluating the simulation of dry hot spots with RegCM4.

Historical meteorological droughts are analysed over the Coordinated Regional Downscaling Experiment‐Central America, Caribbean and Mexico (CORDEX‐CAM) domain during 1981–2010, with particular emphasis on the North American monsoon (NAM) and the mid‐summer drought (MSD) regions. We analyse droughts based on the standardized precipitation index (SPI‐12) and the standardized precipitation‐evapotranspiration index (SPEI‐12) using observations from CRU, CHIRPS, GPCP and ERA5‐Land reanalysis (ERA5), and assess the skill of the regional climate model RegCM4 (version 7) at 25 km resolution driven by ERA‐Interim (Reg‐ERA) and by three global climate models (Reg‐GCMs: Reg‐Had, Reg‐MPI and Reg‐GFDL). Observational data sets show large spatial variations in drought frequency, and both Reg‐ERA and Reg‐GCMs have difficulties simulating it. RegCM4 shows positive precipitation and water balance biases over mountain regions and negative ones over Central America, possibly due to the complex terrain and poor observational data coverage. Despite differences among observations, the trend in droughts, duration and severity show consistent dry hot spots (regions with long‐duration severe droughts) over the western United States, the United States‐Mexico border region, the NAM, the Yucatan Peninsula and northern Central America, with stronger values of SPEI‐12 than SPI‐12, particularly over the subtropical regions. Reg‐ERA and ERA5 show similar spatial patterns and similar positive and negative spatial biases relative to observations. Reg‐ERA and Reg‐Had adequately simulate the spatial patterns of the trend, duration and severity of droughts, with smaller biases in SPI‐12 than SPEI‐12; in contrast, Reg‐MPI and Reg‐GFDL overestimate the trend biases over northwest CAM. Observations, reanalysis, and RegCM4 capture an inverse drought response between the NAM and the MSD regions linked to climate teleconnections; however, a stronger drought signal is observed in the NAM, which appears to be linked to decadal variations from negative to positive phases of the Atlantic Multidecadal Oscillation combined with La Niña conditions (negative El Niño 1+2 phase). [ABSTRACT FROM AUTHOR]