Catchment flows are often treated as continuums of hydrological processes from low flow to
peak flow and back to low flow again, when analysing streamflow by methods like recession
analysis or flow duration curve analysis. Such a conception of catchment response could not
be further from the truth, catchment drainage instead is the sum of various flow components
deriving from different parts of the catchment, as is well-understood by modellers.
Why then have we traditionally applied the analysis methods to streamflow rather
than to the separated components? (Apart, that is, for practical reasons.) Applying
recession analyses to separated components turns out to give surprising results, and
removes confusion arising from the mixture of components in streamflow (Stewart,
2014).
The simplest separation of components is into quickflow and baseflow, which have very
different sources and behaviours as shown in particular by tracer measurements. Quickflow is
direct runoff from runoff events and often drops to zero between events, while baseflow is
sourced from groundwater aquifers and continues as long as the stream flows. As an example,
recession analysis using recession plots (i.e. plots of Q (flow) versus -dQ-dt) for quickflow
and baseflow was applied to data from Glendhu GH1, New Zealand, a schist catchment of
2.2 km2. Whereas the streamflow points were fitted by power law slopes of up
to 4 (i.e. dQ-dt = -0.09Q4) which proved to be artifacts due to the mixing of
components noted above, the quickflow and baseflow points fitted power law slopes of
1.5 revealing the actual quadratic nature of storage reservoirs in the catchment.
Other catchments have given similar results although a wider selection may show
differences, nevertheless the message remains: In order to understand catchment
and hillslope responses we need to be analysing separated components, not just
streamflow.
Stewart, M.K. 2014: New base flow separation and recession analysis methods for
streamflow. Submitted to Water Resources Research, 8 Oct. 2013. |