Process realism: flow paths and storage - Runoff Prediction in Ungauged Basins: Synthesis across Processes, Places and Scales - page 62

Figure 4.9. By comparing isotopic
concentrations, δ

18 O, of inputs and
outputs, information on flow paths
and storage can be inferred. From
McGuire and McDonnell ( 2006 ).

Precipitation (input) Stream (output)

-4
-8
-12
-16

-4
-8
-12
-16

Time [y] Time [y]

gradients of the hydraulic potentials can be estimated from
bedrock topography (or surface topography as a surrogate),
mean transit times (MTT) can be related to average trans-
missivities along the flow paths. The fast component of the
TTD usually stems from vertical and lateral preferential
flow in the unsaturated zone, so contains proxy informa-
tion about the vertical extent of the unsaturated soil store.
The slower component of the TTD usually reflects time
scales for deep percolation into the saturated zone and
travel times through the groundwater system, so contains
information about the vertical extent of the aquifer system,
tortuosity of flow paths and the pattern of transmissivity
along the flow paths. Both environmental tracers (isotopic
and/or chemical characteristics of the waters, e.g., Tetzlaff
et al., 2009a , b) and artificial tracers (Sánchez-Vila and
Carrera, 2004 ; Blöschl and Zehe, 2005 ) can be used to
infer the TTD of the activated flow paths by deconvolution
of the tracer signals of the input (e.g., rainfall) and the
output ( Figure 4.9 ). Transit time analysis can also be
combined with mixing analysis (e.g., Katsuyama et al.,
2009 ). Transit time distributions are often assumed to be
constant during the year, but there may be strong seasonal
variations due to variations in precipitation, evaporation,
and in catchment water storage and the associated acti-
vation of dominant flow paths (Hrachowitz et al., 2010 ;
Heidbüchel et al., 2012 ).
Learning from decay of tracers: age of the water in the
catchment: Some chemicals and isotopes in the atmosphere
show long-term trends and these can be exploited to iden-
tify the age of the input waters (i.e., precipitation) to
catchments. The traditional tracer used to determine the
age of precipitation is the radioactive hydrogen isotope
tritium ( 3 H). The high levels of atmospheric nuclear
weapons testing that took place prior to the enactment of
the Partial Test Ban Treaty in 1963 resulted in high atmos-
pheric tritium concentrations. The subsequent decay of
tritium then allowed the determination of the age of pre-
cipitation. However, due to radioactive decay, the atmos-
pheric tritium concentrations in many parts of the world

have approached their detection limit. Because of this,
other tracers, such as chlorofluorocarbons (CFC), are being
increasingly used (see Kalbus et al.( 2006 ) for a review).

Learning from spatial patterns of tracers in many
catchments
In most instances of catchments without runoff data, there
are no tracer data available. It is then necessary to relate
information from tracer data that are more widely available
to climate and catchment characteristics. This can be
done either through hydrological models or through regres-
sion analyses. A number of studies have identified the
dominant controls of transit times (and therefore the most
useful climate/catchment characteristics for prediction in
ungauged catchments), which differ between different
hydroclimatic regions.
In the peat-dominated catchments and the wet climate of
Scotland, soil properties rather than catchment structure
and organisation are the first-order controls on MTT
(e.g., Tetzlaff et al., 2009a ). Specifically, the proportion
of fast-responding soils taken from the Hydrology of Soil
Types (HOST) classification (Boorman et al., 1995 )
explains most of the spatial variability in MTT ( Table 4.1 ).
If additional variables are taken into account (precipitation
intensity, drainage density and topographic wetness index)
the explanatory power can be further improved (R² 0.88,

Hrachowitz et al., 2009 ). In contrast, soil types are of less
importance than catchment structure in regions such as
the Pacific North-West of the USA, and topographic
indices, such as the ratio of median flow path length over
median flow path gradient, can explain MTT (R 2

¼

0.91)
(McGuire et al., 2005 ). In the Maimai catchment in New
Zealand, a similarly significant relationship between MTT
and flow path distance was found (Stewart and McDon-
nell, 1991 ). The low importance of soil types in explaining
MTT suggests that, in these wet forested areas, a well-
organised network of preferential pathways integrates the
drainage process, and most drainage water bypasses the
soil water store. In some semi-arid, snowmelt-dominated

¼