However, it should be noted that the plume thickness is very sensitive to the chosen tracer threshold value, and our plume thickness could fall into the same range as Fer and Ådlandsvik (2008) if we used a different threshold. We therefore do not overemphasise the detailed comparison of the modelled plume height with actual observations of the Storfjorden plume as many aspects of our model setup are idealised and not designed

to replicate observed conditions. The absolute plume thickness hFhF is normalised by the Ekman depth HeHe defined here as He=2ν/fcosθ for a given slope angle θ   and the vertical viscosity ν   (calculated here by the GLS turbulence closure Selleck GSK J4 scheme) which is averaged over the core of the plume. The vertical diffusivity κκ is also shown to assess the vertical Prandtl number Prv=ν/κPrv=ν/κ which is ≈O(1)≈O(1). The Entrainment ratio is calculated as E=we/uFE=we/uF, where wewe is the entrainment velocity dhF/dtdhF/dt (Turner, 1986) and uF=dL/dtuF=dL/dt is the downslope speed

(L is the distance of the plume edge from the inflow) of the flow. E is calculated over the time taken by the flow until it has reached 1400 m HCS assay depth (or until the end of the experiment if this depth is not reached). The results for both subsets of experiments are summarised in Table 1. Values for vertical viscosity ν   and Ekman depth HeHe are typical for oceanic scales (e.g. Cushman-Roisin and Beckers, 2011) and they are similar in both regimes. However, the plume height hFhF differs considerably between both sets of experiments. A piercing plume is on average 44 m thick towards the bottom end Palmatine of the flow compared to 166 m in experiments where the plume is arrested. An explanation is found in the entrainment ratio E which changes with the depth level of the plume head and thus varies through time. The value of E is larger while the plume head is at the depth level of a density interface in the ambient

waters (which is a considerable portion of the total experiment time in arrested runs). Its value is smaller during the plume’s descent through a homogenous layer of ambient water (as it does for the majority of the experiment time in piercing runs). Based on buoyancy considerations alone one could expect that the incoming plume with a density greater than the density of the bottom layer (in our case for S > 34.85) should always penetrate into that layer. However, our results show that this is not the case because of mixing processes that result in density changes of the plume as it progresses downslope over time. In this section, we examine the downslope propagation of the plume. Fig. 6 shows the depth of the plume edge over time calculated from the deepest appearance of a concentration PTRC⩾0.05PTRC⩾0.05 in the bottom model level.