Sensitivity of Regional Cold-Air Outbreak Simulations to Ice-Nucleating Particle Concentrations

Cloud phase feedbacks remain a major source of uncertainty in climate projections, with shallow mixed phase clouds at mid- and high-latitudes contributing substantially to this uncertainty. Poor understanding of the microphysical processes governing these clouds, particularly their ice content, and limitations in model representations of ice formation, including the frequent neglect of ice nucleating particles (INPs), are key drivers of this problem.  

 In this presentation we show an extensive model analysis of many aircraft flight days during the M-Phase and ACAO projects. The two projects addressed key uncertainties related to mixed-phase clouds through extensive observations of present-day shallow mixed-phase cloud environments collected during two aircraft campaigns, one over the Labrador Sea and one over the Norwegian-Barents Sea. These campaigns sampled cold air outbreak (CAO) clouds in environments characterised by differing sea surface temperatures and INP concentrations (Clarke et al., submitted to GDJ). 

 High resolution (1.5 km) regional simulations are performed for each case using the UK Met Office Unified Model, enabling explicit representation of convection and aerosol cloud interactions over approximately 1000 km domains with 36 hour forecasts. Model output is evaluated against aircraft and satellite observations to assess how well CAO cloud properties are represented. To investigate the role of primary ice production, model INP concentrations are varied, including using an average representation of the INP observed for each flight campaign. 

 The CAO cloud properties show a clear sensitivity to the prescribed INP concentrations, with consistent responses in liquid and ice water path, albedo and cloud fraction across most cases. The magnitude of this sensitivity is larger in colder Norwegian-Barents Sea CAO cases than in the warmer Labrador Sea cases. INP variability explains a larger proportion of liquid water path bias variability in Norwegian-Barents Sea CAOs than in the Labrador Sea cases, suggesting that other processes dominate liquid water path variability in the latter. The warmer Labrador Sea environment indicates a potentially greater role for secondary ice production mechanisms. Variations in the slope of the INP temperature relationship, particularly at colder temperatures, strongly influence ice production in the colder Norwegian-Barents Sea cases. 

 INP parameterisations that are more representative of observed conditions generally reduce the model biases compared to satellite data, although INP sensitivity alone does not account for the full range of biases, which also vary substantially from day to day due to environmental and large-scale meteorological factors along with inherent biases in satellite observations. 

 These results demonstrate that including more-realistic INP concentrations in simulations can improve the representation of cloud properties within CAOs, offering a potential pathway to reducing cloud phase related uncertainty in climate projections.