Process-based diagnostics using atmospheric budget analysis and nudging technique to identify sources of model systematic errors

Abstract. Identifying sources of model systematic errors is a fundamental step to successfully reduce them in general circulation models by improving the representation of relevant physical processes. In this study, we examine model error sources in the Met Office Unified Model at numerical weather prediction timescale by the combined use of two diagnostics: (1) the relaxation or “nudging” in which wind and/or temperature fields are relaxed back towards analyses throughout the simulations, and (2) atmospheric zonal-mean zonal momentum and thermal budgets. The budget analysis quantifies resolved processes and subsequently estimates unresolved processes as a residual, corresponding to model dynamics and physics, respectively. This correspondence is demonstrated by a direct comparison between the budgets and the model tendencies. A systematic error addressed in this paper is the Northern Hemisphere mid-latitude zonal wind bias in the lower stratosphere in boreal winter, characterized by an initial easterly bias that subsequently develops as a westerly bias. The momentum and thermal budget analysis for control and nudging experiments indicates that a mechanical forcing predominantly from parametrized gravity wave drag causes the easterly error and an overly strong temperature gradient around the tropopause is one of the main sources of the westerly error through the Coriolis forcing. The relevant warm bias over the tropical tropopause is mainly attributed to the budget residual term that corresponds to a thermal forcing dominated by radiative processes. This is consistent with the experimental result that temperature nudging over the tropical tropopause significantly reduces the westerly wind bias.