Development of systematic errors in the East Asian summer monsoon

Mean-state errors in the simulation of the Asian summer monsoon are common to many CMIP6 climate models, with some biases persisting over several generations of models.  In this work, we use the Met Office Unified Model (MetUM) as an example to understand the physical processes involved in the emerged circulation patterns.  We focus on the west Pacific subtropical high (WPSH), an important feature of the East Asian summer monsoon, that modulates the distribution of summer rainfall in the region and influences the tropical cyclone activity in the Western North Pacific.    MetUM exhibits robust systematic biases in its representation of the WPSH, including a weakening of the anticyclone and a location too far east, which leads to an underestimation of the southwesterly monsoon flow over East Asia and contributes to seasonal precipitation errors in the area.

We present results from a combination of various techniques and sensitivity experiments, used to understand the sources of the circulation errors.    We benefit from MetUM’s seamless approach of employing the same dynamical core and physical parameterisations in configurations used for various forecast systems, from NWP to climate projections.  Previous studies have shown that many systematic errors in the climate model develop within the first few days of simulation and persist to climate timescales.   Using an ensemble of NWP hindcasts, we examine the error development after initialisation. This allows to reduce the impact of circulation-physics feedbacks and separate the roles played by local physical processes and remote teleconnections.  We apply the nudging methodology, where velocities and temperatures are relaxed towards analysis in chosen regions, to examine the remote effect that biases developing in one region produce in other regions.  The information from these experiments highlights a key role for physical deficiencies over the Maritime Continent in the development of the WPSH biases in MetUM.  The use of an idealised model (semi-geotriptic balance), which allows to study the effect of individual physics tendencies, shows that diabatic heating errors associated with convection are the main source of MetUM WPSH circulation bias.  Further analysis with moisture tendencies reveals that in the model’s parameterised convection, a large drying of the lower boundary layer occurs, balanced only by surface fluxes.  In places with low exchange coefficient (low surface wind), surface fluxes are not able to sustain convection over a long period and the bias is established. 

We examine the persistence of the circulation bias in various models, by evaluating   a perturbed parameter ensemble (PPE) of MetUM climate simulations, which samples climate uncertainties arising from differences in parameter values in physics schemes.   We compare the circulation biases in the ensemble members with model simulations with a new convection parameterisation scheme, CoMorph, and with a convection-permitting simulation. We also show preliminary results of circulation bias development in a machine-learning model for weather forecast.