Mapping vegetation-induced obstruction in agricultural ditches: A low-cost and flexible approach by UAV-SfM

Abstract Drainage network efficiency plays a key role in terms of irrigation and control of floods generation in cultivated areas. A crucial aspect to consider in water resource and flood management risk is the assessment of the network storage capacity. The latter can be reduced by the uncontrolled growth of vegetation that can modify the ditch cross-section altering the hydrological response and network functionality. Therefore, the continuous monitoring of the potentially critical areas, where vegetation could obstruct the section, become a fundamental point that can be solved by the advanced in the high-resolution topographic technologies. In the last few years, the exploration of Structure from Motion (SfM) photogrammetry technique in parallel with Uncrewed Aerial Vehicles (UAVs) have increased the possibilities for the realisation of rapid, low-cost and very detailed Digital Elevation Models (DEM). The accurate and high-resolution DEM generated by UAV-SfM survey is fundamental to derive geomorphometric features of the agrarian landscape. This research aims to present a flexible and low-cost workflow to generate an accurate and high-resolution UAV-SfM DEM, in a large agrarian area of Taiwan, that can be used to automatically detect the drainage network and to map the vegetation into the ditches through the roughness index. The high roughness index values due to the presence of vegetation in the ditches were identified and compared with corresponding measurements in the field to validate and assess the methodology. The results confirm the effectiveness of the approach used and underline how the developed workflow could provide, on a farm scale and with an unusually high level of detail, useful, rapid and low-cost information to map vegetation obstruction in channel network. This data could become essential for stakeholders to precisely intervene on obstruction points, planning the ditches maintenance, and identify some numerical benchmarks to be included in flood risk models and in the computations of the storage size of the drainage network.