Publication Date



Open access

Embargo Period


Degree Type


Degree Name

Master of Science (MS)


Meteorology and Physical Oceanography (Marine)

Date of Defense


First Committee Member

Shuyi S. Chen

Second Committee Member

Chidong Zhang

Third Committee Member

Arnold L. Gordon


The Madden-Julian Oscillation (MJO) is the leading source of predictability on seasonal and subseasonal scales in the tropics, and is one of the least understood phenomena of tropical meteorology. Although extensive research has been done on the topic of MJO initiation and its eastward propagation, there is not a single widely-accepted theory that explains the phenomenon. The lack of understanding is reflected in the poor representation of the MJO in global and numerical weather prediction models. In this study, a regional, atmosphere-ocean coupled model is used to perform a series of high- and low- resolution experiments in coupled and uncoupled configurations to address the effects of moist physics, resolution, and atmosphere-ocean coupling on the simulation of MJO. As a case study, we use the second MJO event (MJO2) observed during the Dynamics of the MJO (DYNAMO) field experiment (October 2011 - March 2012). MJO2 is currently the best observed MJO event on record, and the copious amount of data collected by multiple observational platforms is used for model evaluation. We find that the MJO in the coupled model run at a relatively low resolution (12 km grid spacing) is very sensitive to the choice of convective parameterization, which does not only affect the amount and distribution of precipitation, but also influences the vertical structure of winds and relative humidity in the atmospheric boundary layer. Two convective parameterizations, with different triggering mechanisms for convection are considered, with one producing tropical cyclones, and the second one simulating an MJO, though not accurately representing individual features. When resolution is increased to a convection-permitting grid spacing (4 km resolution) in the coupled configuration, the representation of individual convective features improves regardless of the convective parameterization outside the high-resolution domain. A high precipitation bias is present in all experiments and can be linked to a high bias in the surface-layer air-sea flux parameterization, and a positive bias in the ocean mixed layer depth. Reducing the air-sea flux bias through a modification of the buoyancy-driven turbulence parameterization for air-sea fluxes succeeds at reducing the precipitation bias and improves (weakens) the surface winds, and enhances the representation of MJO’s eastward propagation. With weaker surface winds, the ocean cooling is reduced, which slightly offsets the introduced modification. The high-resolution uncoupled experiment (atmosphere only) does not produce an MJO - it produces excessive precipitation that is present throughout the 15-day simulation and extends eastward from the Indian Ocean, but never propagates out of it. The study concludes that high (cloud-permitting) resolution is integral in accurately representing the precipitation features associated with the MJO, and the MJO-induced upper-ocean cooling in atmosphere-ocean coupled experiments is essential for the MJO’s eastward propagation.


Madden-Julian Oscillation; MJO; intraseasonal variability; atmosphere-ocean interaction; high-resolution coupled modeling; DYNAMO