Abandoned channel fill sequences in the tidal estuary of a small mountainous,dry‐summer river

This study proposes a modification of the current model for abandoned channel fill stratigraphy produced in unidirectional flow river reaches to incorporate seasonal tidal deposition. Evidence supporting this concept came from a study of two consecutive channel abandonment sequences in Ropers Slough of the lower Eel River Estuary in northern California. Aerial photographs showed that Ropers Slough was abandoned around 1943, reoccupied after the 1964 flood, and abandoned again in 1974 with fill continuing to the present. Planform geomorphic characteristics derived from these images were used in conjunction with sub‐centimetre resolution stratigraphic analyses to describe depositional processes and their resultant sedimentary deposits. Both abandonment sequences recorded quasi‐annual scale fluvial/tidal deposition couplets. In both cases, tidal deposits contained very little sand, were higher in organic and inorganic carbon content than the sandier, fluvially dominated deposits, and possessed millimetre‐scale horizontal laminations. The two abandonment fills differed significantly in terms of the temporal progression of channel narrowing and fluvial sediment deposition characteristics. Aerial photographic analysis showed that the first abandonment sequence led to a more rapid narrowing of Ropers Slough and produced deposits with a positive relationship between grain size/deposit thickness and discharge. The second abandonment resulted in a much slower narrowing of Ropers Slough and generally thinner fluvial deposits with no clear relationship between grain size/deposit thickness and discharge. The δ¹³C values and organic nitrogen to organic carbon ratios of deposits from the first phase overlapped with Eel River suspended sediment characteristics found for low flows (one to five times mean discharge), while those of the second phase were consistent with suspended sediment from higher flows (seven to ten times mean discharge). When considered together, the results indicate that the early fill sequence recorded a reach experiencing regular fluvial deposition through flow conditions during the wet season, while the latter fill sequence records a reach more disconnected from the main stem in terms of flow and sediment. The major factor affecting the difference in sedimentation between the two fill periods appears to have been the morphology of the upstream river bend in relation to the position of the bifurcation node. During the first fill period, the upstream entrance to Ropers Slough seems to have remained open, in part due to the placement of its entrance on the outside of the mainstem river bend, and despite stronger tidal effects caused by a larger tidal prism and closer proximity to the tidal inlet. By the second fill sequence, the upstream bend morphology had altered, placing the entrance to Ropers Slough on the inner bank of the mainstem bend, which resulted in more rapid plug bar formation. The role of tidal effects in the geomorphic trajectory of the two abandonment sequences is unclear, but appears to have been less important than local bifurcation geometry.     

基于不确定性迭代优化的山地植被遥感制图

山地因其较高的异质性和特殊的环境特征给遥感科学及其应用带来了诸多问题和挑战。为实现山地植被信息的精准提取,本研究选择部分滇西北山地区域作为研究区开展方法实验,利用高分辨率遥感影像数据和数字高程模型,结合分区分层感知思想,提出一种基于不确定性理论的山地植被型组分类制图方法。首先结合地形对研究区影像进行多尺度分割制作图斑;然后根据图斑特征使用随机森林方法进行分类,将分类结果与对应类别样本间的相似性作为优化目标,并构建混合熵模型定量计算图斑推测类型的不确定性,据此进行针对性的样本补充和分类模型的迭代优化。实验总体分类精度达90.8%,较迭代前提升了29.4%,Kappa系数达到0.875。在高不确定性区域,该方法相比使用一次性补样和随机补样方法的分类结果,精度分别提高了17%和13%。研究结果表明,通过人机交互的方式,基于不确定性理论为样本库融入增量信息的迭代优化方法能够有效提高植被型组分类的精度,相较于传统的样本选择方法具有更高的效率和更低的不确定性。