Abstract Characterisation of the ecosystem functioning of mudflats requires insight on the morphology and facies of these coastal features, but also on biological processes that influence mudflat geomorphology, such as crab bioturbation and the formation of benthic biofilms, as well as their heterogeneity at cm or less scales. Insight into this fine scale of ecosystem functioning is also important as far as minimizing errors in upscaling are concerned. The realisation of high-resolution ground surveys of these mudflats without perturbing their surface is a real challenge. Here, we address this challenge using UAV-supported photogrammetry based on the Structure-from-Motion (SfM) workflow. We produced a Digital Surface Model (DSM) and an orthophotograph at 1 cm and 0.5 cm pixel resolutions, respectively, of a mudflat in French Guiana, and mapped and classed into different size ranges intricate morphological features, including crab burrow apertures, tidal drainage creeks and depressions. We also determined subtle facies and elevation changes and slopes, and the footprint of different degrees of benthic biofilm development. The results generated at this scale of photogrammetric analysis also enabled us to relate macrofaunal crab burrowing activity to various parameters, including mudflat elevation, spatial distribution and sizes of creeks and depressions, benthic biofilm distribution, and flooding duration. SfM photogrammetry offers interesting new perspectives in fine-scale characterisation of the geomorphology, benthic activity and degree of biofilm development of dynamic muddy intertidal environments that are generally difficult of access. The main shortcomings highlighted in this study are a drift of accuracy of the DSM outside areas of ground control points and the deployment of which perturb the mudflat morphology and biology, the water-logged or very wet surfaces which generate reconstruction artefacts through the sun glint effect, and the time-consuming task of manual interpretation of extraction of features such as crab burrow apertures. On-going developments in UAV positioning integrating RTK/PPK GPS solutions for image-georeferencing and precise orientation with high-quality inertial measurement units will limit the difficulties inherent to ground control points, while conduction of surveys during homogeneous cloudy conditions could reduce the sun-glint effect. Manual extraction of image features could be automated in the future through the use of deep-learning algorithms.

中文翻译：

---

从基于无人机的结构-运动摄影测量评估大型动物挖洞活动与泥滩地貌之间的关系

摘要 泥滩生态系统功能的表征需要了解这些海岸特征的形态和相，还需要了解影响泥滩地貌的生物过程，如螃蟹的生物扰动和底栖生物膜的形成，以及它们在 cm 或更小的异质性秤。就最小化升级中的错误而言，深入了解生态系统功能的这种精细规模也很重要。在不干扰这些泥滩表面的情况下实现对这些泥滩的高分辨率地面调查是一个真正的挑战。在这里，我们使用基于运动结构 (SfM) 工作流程的无人机支持的摄影测量来解决这一挑战。我们分别制作了法属圭亚那泥滩的数字表面模型 (DSM) 和正射影像，像素分辨率分别为 1 cm 和 0.5 cm，并绘制并分类成不同大小范围的复杂形态特征，包括蟹洞孔、潮汐排水小溪和洼地。我们还确定了细微的相、海拔变化和坡度，以及不同程度的底栖生物膜发育的足迹。在这种摄影测量分析尺度上产生的结果还使我们能够将大型动物蟹穴居活动与各种参数联系起来，包括泥滩海拔、小溪和洼地的空间分布和大小、底栖生物膜分布和洪水持续时间。SfM 摄影测量在地貌、底栖活动和通常难以进入的动态泥泞潮间带环境的生物膜发育程度的精细尺度表征方面提供了有趣的新视角。本研究中强调的主要缺点是地面控制点区域外的 DSM 精度漂移，其部署扰乱了泥滩形态和生物学、积水或非常潮湿的表面，通过阳光效应产生重建伪影，以及人工解释提取特征（例如蟹洞孔径）的耗时任务。无人机定位的持续发展将 RTK/PPK GPS 解决方案用于图像地理参考和精确定向与高质量惯性测量单元相结合，将限制地面控制点固有的困难，而在均匀多云条件下进行调查可以减少太阳闪光效果。未来可以通过使用深度学习算法自动提取图像特征。