Abstract/Summary

Shallow cumulus clouds are precursors to deep convection, which can lead to extreme weather events that are difficult to forecast. Therefore, to more accurately predict these events, the preceding environment must first be accurately forecast. Advances in computational power have enabled numerical weather prediction models to run with kilometre and sub-kilometer scale grids. Such models can only partially resolve the dominant turbulent structures within the flow, and therefore the unresolved turbulence still needs to be parametrized. Traditional parametrization schemes for shallow cumulus convection are not valid at these grid scales, as they rely on the assumption that all turbulent motions are entirely subgrid scale. This gives rise to the grey-zone regime, which occurs when the scales of the dominant turbulent motions are comparable to the grid length of the model. The aim of this work is to improve the model’s ability to capture the effects, at coarse resolution, of the turbulent structures which transport heat, moisture, and momentum to cumulus clouds, thereby mitigating the impact of the grey-zone. The primary objective of this research is to develop a parametrization that enables the extension of large eddy simulation (LES) abilities to coarser grid scales, traditionally categorised as grey-zone scales, while maintaining accuracy and low computational costs. The Met Office/NERC Cloud (MONC) LES model was used to produce high-resolution fields for three case studies: an idealised dry convective boundary layer case, the Barbados Oceanographic and Meteorological EXperiment (BOMEX) case, and the Atmospheric Radiation Measurement (ARM) case. The dynamic Smagorinsky equations were applied to the resulting fields to produce flow-dependent Smagorinsky parameters. These parameters define the mixing length in the model. By analysing their behaviour, significant variations in turbulent mixing have been identified between the mixed layer, cloud-free environment, and in-cloud regions of the cloud-topped boundary layer (CTBL). The findings reveal that the Smagorinsky parameters for momentum, heat, and moisture are significantly influenced by both the flow regime and filter scale. These dependencies are not accounted for in the standard model. A scale-adaptive relationship between height and the Smagorinsky parameters could then be derived for each variable. A novel parametrization scheme has been developed using these relationships to serve as a grey-zone adaption, enabling the model to capture the key flow dependencies without incurring high computational expense. This aims to deliver the benefits of a dynamic Smagorinsky method but negates the need to compute the parameters at each grid point and time step. The MONC model was modified to include this parametrization in the subgrid scheme, and the resulting grey-zone simulations demonstrated substantial improvements, particularly in terms of cloud initiation time and cloud layer growth. Furthermore, the research underscores the importance of recognizing the variations in turbulent mixing lengths, as the dynamics of the turbulent flow in the CTBL are inherently linked to the grid scale, stability, and flow regimes. This parametrization addresses the limitations of using fixed parameter values when modelling convective turbulence in a CTBL. This research presents a significant step forward in addressing the challenges posed by the grey-zone in modelling shallow cumulus convection. Valuable insights into the dynamics of turbulent mixing in the CTBL offer a promising framework for advancing the capabilities of LES models in the grey-zone regime.