<sup> 137 </ Sup> Cs Soil Erosion Tracer Of Spatial Variability,

The severe soil erosion problem and related problems: such as nutrients loss, land degradation, water pollution and so on has affected the sustainable development of agricultural system. Investigating the magintude and clarifying the law of spatial distribution of soil loss is premise to slove those problems. Using ¹³⁷Cs as a tracer to estimate the soil erosion rate is the most effective and economical one among the existed approaches. It has been widely employed and achieves great success in soil erosion study since it was introduced into China.However, there are still some problems about theory of foundations and fundermental principles of this approach, such as the principle and method in ascertaining the reference inventory, accuracy comparision between different models, spacing of the sampling design and the numbers of sampling required. The study area is located near the NingZhen Mountain, Jiang Su Province. The sampling sites include one dry farmland with at least 100 years cultivated history on the genle slope and 5 paddy fields which were used to choose the most appropriate reference site.We totally collected 80 samples on the gentle slope including 8 stratified samples; in the 5 possible reference locations, we collected 10 stratified samples. In order to make more detailed study on reference inventory, we chose one paddy field of the 5 possible reference locations and one dry farmland, and collected 38 corings, with the sampling pattern using 5m×5m grid-sampling and 5m spacing method.The study show that the random variation of the space is the main reason to influence the spatial distribution of ¹³⁷Cs, so it is essential to collect samples as many as possible in order to ascertain the reference inventory. We need to collect 21 samples to ascertain the reference inventory in the study area at least. The soil erosion rates estimated through the three mass balance models are similar and the results are superior to that obtained by the Proportional model and the gravimetric Approach, so it is reasonable to use any one of the three mass balance models in theory. Sampling method is different according to the requirement for ensuring the accuracy of the soil erosion rate and the spatial distribution of soil loss, hence, we adopt the unequal intervals method to sample, considering the change of landform and the longitudinal variation of soil loss in sampling process can reduce the amount of samples and improve the calculated accuracy. Through the study, we concluded: for the small area, the proper sample design is 3 transects with the proper spacings between them and 12 sampling points which can control the errors within 10% and reflect the spatial distribution of soil loss. This investigation is significant to study the soil erosion on the small scale area in China. However, it needs to further explore the sample design and the amount of samples in different study areas in order to enrich the content in study of reference inventory and the numbers of the samplings required.