科研之友
微信
新浪微博
Facebook
分享链接
摘要
Ephemeral gully erosion is responsible for large losses of soil on cropland and causes serious off-site damages. Understanding and predicting the occurrence of ephemeral gully erosion are therefore major concerns for land users, decision makers or alike. In order to explain and predict the initiation of gully erosion, numerous studies have focused on the concept of topographic threshold which relies on the slope and contributing area at the gully head. Different approaches have been used so far for defining this threshold. However, these approaches may be questioned because they are partly subjective, not always statistically-based or based on the statistical weight of all data points rather than on the data points at the threshold. To cope with these deficiencies, quantile regression is proposed as an alternative for determining the threshold line. It is applied to assess the regional and temporal variability of gully initiation in Wallonia (Belgium) and compared to previous thresholding approaches. A database of gullies was created from aerial photographs for three agro-pedological areas. The areas differed considerably in terms of number (102-282), mean length (84-151 m) and type of gullies. Most gullies were located on land with summer crops and more than 70% were restricted to a single plot. Significant differences in the topographical threshold were observed across areas, but these regional differences were not consistent across the various thresholding methods. Only 12-18% of gullies were recurrent over time, yet the topographic threshold determined by quantile regression seemed to be stable in spite of annual differences in land use and climate. The results reveal the need for greater standardization of thresholding methods. Quantile regression should be preferred over other previous approaches as it is more consistent with the concept of threshold and appears more robust.
- 出版日期2014-2-1
