SimplySed: how to scale reach sediment input as a function of both flow from the land and flow from the reach?

[LeahJB]

Reach sediment input in the original version of SimplyP was a function of reach flow. However, this meant a linear increase in sediment input to the reach down a network of reaches as Q increases. In SimplyP v0.3 we therefore switched it so that the ReachSedimentInput equation used just the water inputs from land to that reach (equation ReachFlowInputFromLand).

However, this solution does not take into account that reaches further downstream will have greater stream power/sediment mobilisation capacity. E.g. if you have two subcatchments/reaches which are identical in size, reach slope, erodibility and flow input from land, but one is downstream of a large headwater subcatchment and the other is not (so that the discharge is larger in one), then we would expect greater sediment input in the one with larger discharge.

To do: find study areas with good sediment data from various points down the river network and from different sub-catchment types, to test out whether it is better to use flow from the land or reach flow (or some combination of the two) in the ReachSedimentInput equation.

Conceptually, the simplest solution is to split the sediment input equations into two:

Delivery of sediment from land: M_land = a_land * S_sc * C_cover * C_measures * (Q_quick+soil ^k_land)
Delivery of sediment from bank erosion & instream entrainment: M_instream = a_instream * S_reach * C_measures * (Q_reach ^ k_instream)

where a_land, a_instream, k_land, k_instream are calibrated parameters.

A benefit of this is that it also includes a more detailed sediment source apportionment, which would help in targeting measures to reduce sediment inputs (e.g. field-based or bank protection). However, I think this equation is too complex to be robust in reality. We know that for a single reach, [SS] = aQ^k works well. Splitting a and k into two is going to lead to non-identifiability, unless we can come up with clever ways of constraining these parameters. This needs thinking about more, and suggestions are welcome!

[Maginor]

Note also that the downstream reach will still have higher concentrations than if it was just on its own since it gets inputs from upstream, and this is captured in the current model.

What we don't capture is if there is a reach in a dry (no rainfall) lowland area that gets clean water from an upstream mountain area, which should still get some instream mobilization in the lowland area.