Evaluation of cesium-137 conversion models and parameter sensitivity for erosion estimation

The Cs technique has been widely used to provide soil redistribution estimates since the 1970s. However, most Cs-conversion models remain theoretical and largely unvalidated. Our objectives were to validate the four widely used conversion models, examine model parameter sensitivity, and evaluate the potential of using kriging to improve soil redistribution estimation. Soil loss was measured from a 1.6-ha plot since 1978. Winter wheat ( L.) was grown primarily under conventional tillage. Soil samples in a 10-m grid were taken from the plot to estimate the Cs inventory. Soil redistribution rates were estimated using four models and were further interpolated using ordinary kriging. The parameter sensitivity analyses at the 95% confidence limits showed that reference inventory had the most impact on estimated water erosion, followed by particle size correction for erosion and tillage depth, with minimal impacts from mass depth, bulk density, and particle size correction for deposition. Compared with the measured water erosion, the relative errors of the mean net water erosion estimates across the entire plot without and with kriging were 28 and -17% for the proportional model (PM), 141 and 106% for the simplified mass balance model, 133 and 100% for the improved mass balance model (MBM2), and 109% for the extended MBM2 with tillage erosion (MBM3). Results indicated that the PM performed better than the mass balance models under the study conditions and that kriging improved mean soil redistribution estimates. However, the full potential of the MBM2 and MBM3 needs to be further evaluated under conditions where loss of newly deposited Cs exists.