Understanding global monsoon (GM) variability and projecting its future changes rely heavily on climate models. However, the models often exhibit pronounced biases in GM simulations. Notably, the dry and wet biases found in the Northern (NH) and Southern Hemisphere (SH) monsoon regions in CMIP3 and CMIP5 persist in CMIP6 models, though the reasons for these biases remain unclear.

Here, we identify the sources of GM simulation biases from an energy transport perspective and develop a diagnostic framework to understand the performance of current climate models. This framework focuses on fundamental and observable processes of interhemispheric energy transport that are physically intuitive. Local summer precipitation in both NH and SH monsoon regions is closely linked to interhemispheric energy transport. During boreal summer, a pronounced interhemispheric thermal contrast (ITC) promotes stronger southward and northward moist static energy (MSE) transport in the upper and lower levels, respectively, leading to more vigorous monsoon circulation and increased precipitation in the NH. Conversely, during the austral summer, similar processes enhance monsoon activity in the SH. Interhemispheric energy transport is primarily driven by interhemispheric differences in net energy flux into the atmosphere, associated with downward longwave radiative flux from the top of the atmosphere and upward longwave radiative flux from the surface.

Evaluating the skill of climate models in CMIP5 and CMIP6 using this diagnostic framework reveals a multimodel mean improvement in CMIP6 compared to CMIP5, with skill scores for various GM metrics increasing from 0.20~0.79 to 0.48~0.83. This improvement in GM simulations in CMIP6 is specifically attributed to reduced dry biases in NH monsoon simulations. These improvements are linked to smaller negative biases in downward surface longwave radiation and northward energy transport in CMIP6 compared to CMIP5. As a result, CMIP6 models exhibit more realistic ITC and meridional MSE gradient, contributing to smaller dry biases in the GM simulations.

By demonstrating the connections between model biases in the monsoon and energy transport, this study shows that accurately reproducing the meridional global atmospheric energy transportation is necessary for skillful GM simulation.