How climate model ITCZ biases depend on weather: case studies in the east Pacific Ocean

Abstract:

The intertropical convergence zone (ITCZ) is a fundamental feature of the atmosphere. It has a high impact on other large-scale events such as atmospheric rivers and tropical storms which influence large biospheres and the lives of millions of people. However, many climate models overproduce one ITCZ in each hemisphere when there should only be one ITCZ in the northern hemisphere, called the double ITCZ bias. This study presents an analysis of the accuracy of the Community Atmosphere Model (CAM) in its hindcast mode in simulating sub-monthly shifts of the eastern Pacific Ocean ITCZ during boreal spring. Precipitation and horizontal surface wind velocities were compared against observational datasets such as NASA’s TMPA and the European Centre’s ERA5 reanalysis. Spring 2010 case studies of observed northern ITCZ events show that CAM produces significant biases in the Southern Hemisphere in as little as one week of lead time. Due to this double ITCZ bias, the precipitation modeled by CAM just south of the equator in the east Pacific Ocean erroneously reached up to 30% of the precipitation observed by TMPA just north of the equator. The southerly winds modeled by CAM in the Southern Hemisphere were also weaker than what was observed. These relatively weak winds allowed for the stronger northerly winds in the Northern Hemisphere to advect moisture from the northern ITCZ and create a second convergence zone in the Southern Hemisphere; thus, a double ITCZ was incorrectly modeled. In addition, strong storms modeled in one hemisphere created a disconnect in the zonally elongated ITCZ in the other hemisphere along the same longitude. Five more years of CAM hindcast simulations will allow for an analysis of the interannual variability of the double ITCZ bias as it develops on sub-monthly time scales. To better show long term trends, multi-annual averages of the biases in CAM will be computed. Furthermore, the scope of variables investigated in the work will be expanded to include low-level horizontal pressure gradients and surface latent and sensible heat fluxes.