Measuring Gravel Transport in an Active Natural System: An Analytical Framework

In order to measure sediment flux in Porter Creek, a small tributary to the North Fork of the Siuslaw River near Florence Oregon, we have deployed ~600 pieces of tracer gravel embedded with passive integrated transponder (PIT) tags, 8 fixed antennas, and 9 logging pressure transducers spaced along 130 m of channel comprising 3 wood jams and substrates of sand, gravel, cobble, and bedrock. Tracer deployment is uniform along the instrumented reach, analogous to constant-source solute or dye injection, so that sediment flux [L3/T] for the ith grain size class is Qi = niVpiFi/fTi, where ni is count rate, [T-1], Vpi is particle volume, and Fi and fTi are fractional coverage of the ith size class of grains and tracers, respectively. Tracer concentrations, fTi, must be large enough for accurate estimation of ni = 1/TAi where TAi is the mean inter-arrival time of tracers at an antenna, during a period of nearly constant discharge. A square wave or constant sediment injection is undertaken by placing a concentration of tracers dispersed upstream of the study reach, such that it will add to the concentration within the study reach as gravels migrate downstream, replacing the gravels within the antenna network. Preliminary results show dispersion values ranging from ~7 m2/month for 8-16mm size fraction, to ~0.2 m2/month for 32-64mm size fraction, with travel distances of 60 meters for the 8-16mm, 16 meters for the 16-32mm, 8 meters for the 32-64mm, and 4 meters for the >64mm for 1 water year. Since there is a high level of variability in dispersion within the antennae array given the heterogeneity of substrates and wood placed within the system, it is likely that some tracers will need to be added within the regions between antennae after high water events. The tracer concentration within the regions occupied between antennae must remain at such a level as to provide viable statistical relationships between tracer and non-tracer gravels, and percent mobile versus percent stationary.