4.1 Complexities in Transient Storage Flow Paths
We assume, based on our results, that the flume is more sensitive to surface flow paths and the numerical model to subsurface flow paths. For example, a mean arrival time of only minutes in the flume is likely consistent with a combination of slow flow in surface transient storage zones plus some of the fastest flow through subsurface zones. However, it is too short for the slowest flow through the subsurface (see groundwater ages in Figure 3). We therefore use the numerical model to quantify the behavior of the longest solute travel times in the subsurface. Both physical experiments and numerical models represent simplifications of real conditions, but the lack of fine sediment accumulation in the logjam backwaters, the rigid lateral boundaries, and the impermeable boundary underlying the sediment fill in the flume may all have influenced subsurface flow paths. In the models, it is unclear whether more skew (which we interpret as more transient storage) equates to a greater hyporheic exchange flux or longer residence times (Supplemental Table 5). As has been shown in past studies (Cardenas et al., 2004; Cardenas et al., 2008; Sawyer and Cardenas; 2009; Pryshlak et al., 2015), the two tend to work against each other (greater exchange rates tend to occur with shorter residence times).
We observed complex dynamics in the relationship between logjam distribution density, permeability, discharge, and transient storage. As logjams accumulate in a stream channel and accumulate more fine material, upstream pooling increases, driving surface and hyporheic flow paths and heightened complexity of the stream system (Abbe and Montgomery, 1996; Sear et al., 2010). Evolution in jam complexity (both in terms of permeability and longitudinal distribution density) can result in the activation of the floodplain and other portions of the hyporheic zone that might otherwise be dormant (Gooseff et al., 2006; Wondzell et al., 2009; Doughty et al., 2020). Changes in inundation area have been shown to have a profound influence on hyporheic connectivity at field scales on the order of tens of square kilometers (Helton et al., 2014). We see this in our results where increasing both longitudinal distribution density of logjams and the amount of coarse particulate organic material in a jam leads to an increase in wetted area and exchange fluxes between the surface water and groundwater (Supplemental Table 5).
Existing work provides evidence for both increased transient storage during low discharge conditions, when the water table near the stream is low (Harvey and Bencala, 1993; Harvey et al., 1996; Wroblicky et al., 1998) and during high discharge conditions, when the stream experiences greater channel wetted area and floodplain inundation (Nyssen et al., 2011; Doughty et al., 2020; Wilhelmsen et al., 2021). Comparisons of mean arrival times and skew in the surface and subsurface show consistencies across the flume and model. Our results show less skew and shorter mean arrival times when discharge increases (Figures 2, 5, 7). In the flume, flow paths are better imaged at low discharge when we can load more of the flow paths with a short salt pulse. In the models, all the mean arrival times are already limited to the subsurface (a transient storage zone), so skew is only telling us about the longest transient storage times in that case, and a larger skew indicates a wider, heavier-tailed distribution of residence times. We suspect understanding the role of discharge in these environments is dependent on sensitivity to detecting the smallest fraction of flows that stick around the longest, which becomes increasingly challenging to do at higher discharge. Both the flume and model results reflect simplification of the river corridor in which flow paths cannot access a floodplain. We recognize that longer and deeper flow paths might be constrained by the base of the sediment box in the flume and impermeable walls of the model. The extent of these flows in natural streams could depend strongly on the depth of alluvial cover (Tonina and Buffington, 2009), which is in turn influenced by the presence of large wood, because logjams store sediments (Massong and Montgomery, 2000; Montgomery et al., 2003). The interacting effect of jams, sediment storage, and longer, deeper hyporheic exchange flows cannot be tested in these experiments but is an area for future research.