CORDEX‐NA: factors inducing dry/wet years on the North American Monsoon region

The output of four regional climate models (RCMs) from the Coordinated Regional Climate Downscaling Experiment (CORDEX)-North America (NA) region was analysed for the 1990-2008 period, with particular interest on the mechanisms associated with wet and dry years over the North American Monsoon (NAM) core region. All RCMs (RCA3.5, HadGEM3-RA, REMO, and RegCM4) were forced by the ERA-Interim reanalysis. Model precipitation was compared against several observational gridded data sets at different time scales. Most RCMs capture well the annual cycle of precipitation and outperform ERA-Interim, which is drier than the observations. RCMs underestimate (overestimate) the precipitation over the coastal plains (mountains) and have some problems to reproduce the interannual variability of the monsoon. To further investigate this, two extreme summers that showed the largest consistency among observations and RCMs were chosen: one wet (1990) and one dry (2005). The impact of the passage of tropical cyclones, the size of the Western Hemisphere Warm Pool (WHWP), the Intertropical Convergence Zone (ITCZ) position, and the initial intensity of the land-sea thermal contrast (LSTC) were analysed. During the wet year, the LSTC was stronger than the 2005 dry monsoon season and there were a larger number of hurricanes near the Gulf of California, the WHWP was more extended, and the ITCZ was located in a more northerly position than in 2005. All these processes contributed to a wetter NAM season. During the dry year, the LSTC was weaker, with a later onset, probably due to a previous very wet winter. The inverse precipitation relationship between winter and summer in the monsoon region was well captured by most of the RCMs. RegCM4 showed the largest biases and HadGEM3-RA the smallest ones.