Key points are not available for this paper at this time.
The development of remote sensing technologies and image classification methods has facilitated research on changes in river channels and floodplains by automating mapping of land cover and geomorphic units. In often highly heterogeneous river environments, object-based approaches proved sensible, since they are based on objects produced by image segmentation rather than on individual pixels. This study uses object-based image analysis to investigate land cover and river channel dynamics in the managed corridor of the Orljava River in the Pannonian basin (Croatia). In the last decade, the river has been affected by anthropogenic removal of riparian vegetation to increase channel capacity, which was followed by a big flood event. Five river corridor classes (water, bare soil, sparse vegetation, dense vegetation, and shadows) were classified based on RGB and near-infrared (NIR) aerial images in the period 2011-2021. A digital surface model generated from the images was used to separate bare river channel units (river sediments) from bare soil in the floodplain and to define high vegetation, while agricultural land was classified manually. The accuracy of the produced maps was between 85 and 93%, except for the year 2014 which lacked the NIR band. Based on classified river corridor units, changes in channel morphology were further analysed in GIS. The two main phases of river corridor changes were caused by the occurrence of a big flood in 2014. In 2011-2014, immediately after the flood, a significant increase in the area of water and river sediments was recorded, mostly at the expense of bare riverbanks and adjacent agricultural land. Large in-channel bars have formed due to sediment accumulation, as well as significant channel migration has been recorded. Contrarily, in 2014-2021 lower discharges allowed gradual channel narrowing and stabilisation, characterised by the spread and growth of vegetation in the river corridor. Observed changes in channel morphology and vegetation succession are natural processes related to actively meandering rivers. However, it has been shown that extensive bank erosion during the flood and subsequent land cover dynamics were primarily triggered by anthropogenic removal of riparian vegetation exemplifying how inadequate and isolated river management decisions may increase bank erosion and lead to a loss of agricultural land in floodplain areas.
Pavlek et al. (Fri,) studied this question.