Quantifying the spatial variation of 7Be depth distributions towards improved erosion rate estimations

There is growing interest in the application of the natural fallout radionuclide Be-7 as a soil erosion and sediment tracer. Development of robust datasets is, however, hampered by unquantified variability in its vertical distribution within surface soil. Models that convert Be-7 inventory measurements to soil erosion estimates are all based on the observed depth distribution of Be-7, described by the relaxation mass depth (110) parameter. Previous work, however, has not considered potential spatial variation in 110 linked to natural variability in soil physical properties, which could have major implications for the reliability of soil erosion estimates. Two complementary experiments were designed to study the variability in depth distribution within and between potential reference sites. First, a field sampling programme was carried out whereby two reference sites with variable degree of compaction were sampled using two different sectioning techniques, i.e. by use of a fine increment soil collector (FISC) and the scraping methodology. During a laboratory rainfall simulation experiment, water spiked with stable Be-9 was used to study the variability in 9Be soil depth distribution within and between the two reference sites. In the field experiment, variations in the Be-7 depth distribution, and thus in 110, were limited between both reference sites (13 to 16%). In contrast, the impact of the sectioning technique was remarkable, with scraping resulting in a higher 110 (up to 60%) compared to the estimates based on the use of a FISC. The rainfall simulation experiment offered the opportunity to study the variation in 9Be depth distribution in more detail. With an average 110 of 4.66 kg m(-2), Be-9 penetrated deeper in the non-compacted (NC) reference site cores, while the compacted (C) cores showed an average 110 of 2.42 kg m(-2). The reported 110 values at the former site were also characterized by a larger coefficient of variation (24%) than those at the latter site (11%). Lower bulk density, higher infiltration rates and a pore network characterized by a higher macroporosity and connectivity, as revealed by the X-ray Computed Tomography (a) scans, explained the deeper penetration of Be into the topsoil of reference site NC. The results indicate the importance of selecting appropriate reference sites and for ensuring an adequate sampling strategy to encompass local variability in soil physical properties. Hydraulic conductivity assessment could be a useful tool to properly assess suitable reference sites and the number of samples needed to assess the reference inventory. (C) 2016 Elsevier B.V. All rights reserved.