Agglomerative Clustering Analysis of North American Monsoon Using GFDL CM4 Climate Model Simulations

The North American Monsoon (NAM) is an annual precipitation maximum that occurs during the summer in the American Southwest, a region including Arizona, New Mexico, and North-Western Mexico. This seasonal disturbance is responsible for greater than 50% of annual precipitation in portions of Mexico, with slightly lower contributions in New Mexico and Arizona. As a result, it is of particular interest to develop a comprehensive understanding of the precipitation characteristics exhibited by the NAM and how these could change in future climate regimes. In this study, an agglomerative clustering analysis technique is applied to daily precipitation values from NOAA CPC gauge-based reanalysis and NOAA GFDL CM4 modeled datasets. Precipitation patterns are compared between reanalysis and historical CM4 simulation to assess model fidelity. Then, historical simulation data are compared to a simulation forced by an abrupt quadrupling of CO2. Finally, associated three-dimensional dynamic and thermodynamic fields are compared to determine mechanisms leading to the observed precipitation patterns. Agglomerative clustering produced three clusters indicating varying northward propagation of NAM precipitation. Results suggest acceptable agreement between observed and simulated precipitation patterns, with varied agreement on the characteristics of resolved clusters. Model comparison indicates a 31% reduction in NAM-associated precipitation, with analogous reductions over each resolved cluster. Based on these results, contributions of dynamic and thermodynamic forcing mechanisms to drying varied depending on the precipitation pattern in question. Continued work aims to further elucidate the impacts of various forcing mechanisms in these model simulations.