| Tidal network, as the irreplaceable geomorphologic component rooted in tidal flat, links the supratidal salt marsh zones with the sea, and functions as the water path transferring the nutrients, sediment, planktonic organisms and pollutants. In many cases, the geomorphic processes of tidal networks are very sensitive and changeable under the compound effect of the tides, waves, storm surges, human activities and other factors. These complex geomorphologic processes, including lateral migration, headward, sideward and downward erosion, pose a great threat on the adjacent dams, ports and other coastal projects. Tidal networks are well developed along the middle section of Jiangsu coast. The study of tidal networks is essential in order to understand the geo-morphologic dynamics of intertidal zones, and is crucial for the sustainable development of regional economy and society under the background of the rapid changes in sedimentary dynamic environment and the intensified impact of human activities. Mainly focused on key issues concerning about technic difficulties in extraction, delineation and analysis for tidal networks, this thesis makes progresses in the following fields:(1) The automated method for extracting channel-like feature (APMECF) was proposed based on the geometric property of the tidal network profiles documented in airborne LiDAR DEMs. Specifically, a multi-window median neighborhood analysis was designed to enhance depressions both in mudflat and marsh environments; a multi- scale and multi-directional Gaussian-matched filtering method was incorporated to enhance width-variant tidal networks; and a two-stage adaptive thresholding algorithm was implemented to segment low-contrast tidal networks. APMECF was tested on two large LiDAR datasets of the Jiangsu coast with different resolutions. The quantitative assessments show that APMECF not only successfully extracted both small and large scale tidal networks, but also maintained integrity of tidal networks. Additionally, the extended experiments including terrestrial channel extraction in mountainous areas and wadi extraction in arid west Asia, demonstrated the transportationality of APMECF for channel-like feature extraction, especially in information carriers with homogenous background.(2) Regular monitoring of the long-term evolution of tidal flat is difficult until the advent of remote sensing, either based on DEMs obtained by airborne LiDAR directly, or indirectly based on high-resolution satellite imagery. It is acknowledged that, satellite imagery with high-frequent revisiting period is comparatively more cost- and temporal effective, and guarantees the sampling of the coastal surface in a more efficient way. Nevertheless, challenges arise from the spectral carrier, instead of documenting elevation directly, as environmental heterogeneity, spectral non-uniformity and geometric variability. This study presented a systematic workflow for extraction of shape-varying tidal networks with high automation. It is termed the multiscale local differential structure induced (MLDSI) method: ① the contrast disparity was greatly mediated by a modified Fuzzy C-means algorithm (MFCM), making tidal network outstand the uniform background; ② the scale-varying tidal networks were delineated by a geometric linear feature indicator (LFI) decomposed from the local differential structure; ③ pseudo-responses stimulated by heterogeneous environment was suppressed through a joint multi-band refinement strategy;④ final result was acquired by applying region-based level set method incorporating spatial coherency. Using ZY-3 high-resolution images, MLDSI was validated over different four test sites over the littoral tidal flat (3 sites over 90 km2 in total) and offshore sandbank (1 site over 520 km2), both in Jiangsu Middle coast, China. Quantitative evaluation demonstrated that the proposed MLSDI method outperformed five classic methods on maintaining feature completeness, and exhibiting more robustness under heterogeneous disturbance.(3) Based on the abovementioned tidal network extraction methods, coupling measures were proposed to quantitatively explore the corresponding morphometric patterns by analyzing geometric characteristics and hydrologic characteristics at the same time, namely quantifying anisotropy and delineating unchanneled density of tidal networks. Accordingly, within inshore areas, predominant directions took place in relatively low tidal regions, i.e., mudflat zone, and were approximately perpendicular to the shorelines facilitated by the persisting tidal currents. But in the supratidal regions like salt marsh zones, such oriented pattern disorganized with the growing ramifications. In short, anisotropy rapidly increased landwards in littoral tidal flats. Global statistics of offshore sandband show that creeks of east-west direction were nearly two times than those of north-south direction. Skewed distribution was more obvious of embedded creeks than inserted creeks. Generally, anisotropy variation of tidal network is a regional adaptive evolution and a morphological adjustment to spatial gradients, including zonal, bio-geographic and topographic patterns. Otherwise, It is clearly observed that strong linear trend on the semi-log plot suggests the exponentially decaying distribution of unchanneled flow length for watershed development at distinguishable aggregation property. Meanwhile, the mean overland flow length seems to greatly fluctuate with an absence of scale-free distribution for the adjacent watersheds in the same site, even for secondary basins in the same watershed. Consequently, network features can be discriminated mutually by this distinctive distribution, which indicates geometric branching and meandering characteristics of tidal network. |
PDF Full Text Request