Caesium‐137‐derived erosion rates in an agricultural setting: the effects of model assumptions and management practices

Abstract Soil erosion rates at the National Soil Tilth Laboratory Deep Loess Research Station in southwest Iowa (USA) were determined from 137 Cs inventories using a multi‐box numerical model that incorporates time‐dependent atmospheric 137 Cs deposition. Nine sites along a downslope transect where 137 Cs inventories were measured in 1974 by Spomer et al . ( Transactions of the American Society of Agricultural Engineers 1985, vol. 28, pp. 767–772) were revisited in 1998, allowing estimation of erosion over three time periods: 1954–1974, 1954–1998 and 1974–1998. Modelled 1954–1974 erosion rates (mean = 1·01 ± 0·41 g cm −2 a −1 ) were significantly higher ( p < 0·001, t ‐test) than those reported by Spomer et al. (0·36 ± 0·10 g cm −2 a −1 ). A key reason for the significant difference is that our model used a monthly time step whereas Spomer et al. used a single 20‐year time step. As the model time step was increased from 1 month to 20 years, the erosion rate decreased by a factor of three to rates in general agreement with those reported by Spomer et al. This suggests that failing to account for temporal variations in atmospheric deposition and/or cultivation practices will result in substantially miscalculated erosion rates. Erosion rates over 1974–1998 were calculated by comparing our 1998 measured 137 Cs inventory to the decay‐corrected 1974 inventories reported by Spomer et al . This comparison yielded erosion rates of −0·10 to 0·48 g cm −2 a −1 (mean = 0·19 ± 0·17 g cm −2 a −1 ), indicating that soil loss was reduced by 81 ± 52 per cent over 1974–1998 compared to the period 1954–1974. This estimate is in reasonable agreement with measurements of soil flux at the weir which indicate that soil loss has been reduced by 43 per cent since grass buffer strips were added to the field in 1991. Copyright © 2005 John Wiley & Sons, Ltd.