Spatial variability of erosion and soil organic matter content estimated from 137Cs measurements and geostatistics

Abstract

The spatial variability of soil erosion was studied through the use of fallout radionuclides (FRNs) and geostatistics. The spatial correlation structures of radiocesium ( 137Cs), soil redistribution and organic matter (OM) content have been established in a 2.16 ha agricultural field located 30 km east of Quebec City, Canada. A significant relationship was found between 137Cs (Bq m − 2 ) and OM (%) in the 0–20 cm soil depth ( n = 42; r 2 = 0.63, p < 0.001), highlighting the relation between these two parameters. The conversion of the areal activities of 137Cs into soil redistribution (t ha − 1 yr − 1 ) was done using the Mass Balance Model 2 (MBM 2). The magnitude of soil redistribution, at the sampled points, ranged from an erosion rate of 62 t ha − 1 yr − 1 to a deposition of 17 t ha − 1 yr − 1 . Geostatistics coupled with a geographic information system (GIS) were used to create a map of soil redistribution, based on the spatial variability of FRNs, and to establish a sediment budget. Prior to mapping, semivariograms were produced, taking into account the autocorrelation present in the data. A significant autocorrelation and reliable variograms were obtained for the three tested parameters ( 137Cs, OM content and soil redistribution) (0.87 ≤ r 2 ≤ 0.95; 0.7 ≤ Scale/Sill ≤ 0.96 and 4% ≤ ‘nugget-to-sill’ < 20%). Using the Kriging interpolation and ‘area weighted mean’ of the soil redistribution map, a sediment budget was estimated for the whole field. A net sediment output was estimated as 16.6 t ha − 1 yr − 1 , for a sediment delivery ratio (SDR) of 99%. This high SDR is believed to reflect the joint impact of tillage, water and snowmelt erosion on the net sediment production. Approximately 85% of the agricultural field surface was estimated to be affected by erosion rates approaching or exceeding the suggested tolerance level of 6 t ha − 1 yr − 1 for most Canadian soils. The geostatistics concept is a powerful tool in soil science and especially for FRNs use in order to characterize the spatial variability of erosion and sedimentation processes.