Fallout Radionuclides - Application of Geochemical Tracers to Fluvial Sediment - page 69

soils. However, Mabit et al. ( 2013 ) cogently argue that the 137 Cs (FRN) method
is not based on the absolute amounts in the soil, but on a comparison between the
reference and sample sites. Considering that the processes affecting the sample
sites are also likely to affect the reference sites, lateral redistribution in soils is
likely to be negligible in comparison to the movement of 137 Cs associated with
physical or anthropogenic erosion processes.
3. Estimates of soil erosion or deposition can be accurately estimated from mea-
surements of the FRN inventories
Differences in FRN inventories between an undisturbed (reference) and disturbed
site must be converted into an absolute quantity of eroded or deposited sediment.
This conversion is no easy task given that the character of the soil (e.g., its organic
matter content, grain size, and bulk density) will undoubtedly vary across the study
area, and the FRN activity changes as a function of depth (Fig. 3.3 ). A wide range
of models have been developed to tackle the problem over the past three decades
(e.g., Elliott et al. 1990 ; Ritchie and McHenry 1990 ; Kachanoski 1987 ; Zhang
et al. 1990 ; Fredericks and Perrens 1988 ; Walling and He 1999 ; Fornes et al.
2005 ; Soto and Navas 2008 ; Walling et al. 2002 ; Wallbrink and Murray 1996 ;
Walling et al. 2011 ). These models differ in (1) their degree of complexity and
sophistication, (2) whether they can be applied to cultivated or uncultivated areas,
and (3) the FRN to which they are applicable.
Selection of a conversion model is an important aspect of any FRN tracer study
and is discussed in more detail in Walling et al. ( 2011 ), and Porto and Walling
( 2012 ). The selection process will necessarily need to consider the intended use
of the resulting estimates of soil erosion and deposition, the land-use/land-cover
at the study site, and the data that are available or can be obtained for use in the
models. The point to be made here is that different models are likely to gener-
ate different results. Fornes et al. ( 2005 ), for example, found an approximately
threefold difference between the results generated from two different conversion
models when applied to 137 Cs data that had been collected in 1974 and again
in 1998 at the National Soil Tilth Laboratory Deep Loess Research Station in
Southwest Iowa. Part of the difference in modeling results in the Fornes et al.
( 2005 ) study is that one model incorporated variations in 137 Cs fallout through
time, whereas the other did not. The comparison of model results show that in
general the more complex models yield more reliable estimates of the amount of
soil erosion and/or deposition that occurs for the time period of interest, but they
require data that are often difficulty to obtain (Mabit et al. 2008 ). In addition,
many of the more refined models are highly sensitive to variations in the input
parameters requiring that the additional data be accurately collected.
3.3.2.3 Summary of the Limitation on FRN Inventory Derived Estimates
It is hard to argue that the assumptions required to estimate the magnitude of sed-
iment redistribution on hillslopes or within a catchment using FRNs are strictly
upheld. The question, however, is whether the departures from the methods inherent