Mapping three-dimensional morphological characteristics of tidal salt-marsh channels using UAV structure-from-motion photogrammetry

Chen, Chunpeng and Zhang, Ce and Schwarz, Christian and Tian, Bo and Jiang, Wenhao and Wu, Wenting and Garg, Rahul and Garg, Pradeep and Aleksandr, Chusov and Mikhail, Shilin and Zhou, Yunxuan (2022) Mapping three-dimensional morphological characteristics of tidal salt-marsh channels using UAV structure-from-motion photogrammetry. Geomorphology, 407: 108235. pp. 1-13. ISSN 0169-555X

Text (GEOMOR_S_21_00881_accepted)
GEOMOR_S_21_00881_accepted.pdf - Accepted Version
Restricted to Repository staff only until 5 April 2025.
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (2MB)

Abstract

Tidal channels (TCs) are geomorphological features of coastal and tidal landscapes. They provide a pathway for the exchange of material and energy between marshes and adjacent water bodies and thereby control the hydrodynamic, morphological, and ecological processes on marsh platforms. Due to difficulties in terms of accessibility, limitations on the duration of exposure during low-water stages, and variations in morphology over time, rapid and accurate mapping of such intertidal morphological features at a high frequency is extremely challenging. Here, we present an efficient method integrated unmanned aerial vehicles (UAVs) structure-from-motion (SfM) photogrammetry, and spatial morphological fitting and delineation for accurately quantifying channel three-dimensional (3D) morphological features in terms width, depth, width-to-depth ratio, and cross-sectional area. We also relate these measured proxies to salt marsh species distributions. A two-step thresholding approach combining elevation and slope is developed in order to determinate TC boundaries from salt marsh and tidal flat area, and a Gaussian fit is used to estimate water-bearing channel depth and cross-sectional area. Salt marsh species are identified from fine-resolution multispectral satellite data and a field training dataset using a Random Forest classifier. Our results indicate that (1) UAV-based SfM photogrammetry can achieve centimeter-level accuracy in mapping the topography of TCs, with a root mean square error (RMSE) of 5.7 cm — mainly from the strong reflection of light from smooth TC water surfaces and the presence of water-bearing layers; (2) the morphological features of TCs, ranging from tidal flats to salt marsh areas, demonstrate a similar tendency, which increases at first and then decreases. The maximum depth and cross-sectional area of TCs is in sparse salt-marsh area, up to 4 m and 150 m2, respectively; and (3) TC morphology has a major impact on the distribution of salt marsh plants and such effects vary across different plant species. These results greatly contribute deep understanding of feedbacks between TCs and salt marsh plant species distribution and have significant implications for developing ecological and morphological salt marsh restoration guidelines.