Automated Identification of River Hydromorphological Features Using UAV High Resolution Aerial Imagery

Abstract

European legislation is driving the development of methods for river ecosystem protection in light of concerns over water quality and ecology. Key to their success is the accurate and rapid characterisation of physical features (i.e., hydromorphology) along the river. Image pattern recognition techniques have been successfully used for this purpose. The reliability of the methodology depends on both the quality of the aerial imagery and the pattern recognition technique used. Recent studies have proved the potential of Unmanned Aerial Vehicles (UAVs) to increase the quality of the imagery by capturing high resolution photography. Similarly, Artificial Neural Networks (ANN) have been shown to be a high precision tool for automated recognition of environmental patterns. This paper presents a UAV based framework for the identification of hydromorphological features from high resolution RGB aerial imagery using a novel classification technique based on ANNs. The framework is developed for a 1.4 km river reach along the river Dee in Wales, United Kingdom. For this purpose, a Falcon 8 octocopter was used to gather 2.5 cm resolution imagery. The results show that the accuracy of the framework is above 81%, performing particularly well at recognising vegetation. These results leverage the use of UAVs for environmental policy implementation and demonstrate the potential of ANNs and RGB imagery for high precision river monitoring and river management.

Keywords: Artificial Neural Network; Unmanned Aerial Vehicle; feature recognition; hydromorphology; photogrammetry.

Figure 1

(a) Location of the study site along the river Dee near Bala, Wales, UK; (b) Detailed view of the study area.

Figure 2

Workflow summarising the steps followed in the photogrammetry using Photoscan Pro and the image classification using the Leaf Area Index Calculation (LAIC) software, based on the workflows presented by [21,37], respectively. GDS, GCP and XP stand for Ground Sampling Distance, Ground Control Point (red points) and Check Point (yellow points), respectively.

Figure 3

Detailed diagram of the workflow for the Leaf Area Index Calculation (LAIC) image classification and validation based on [18] (a–d). (a) 300 m section within the reach showing the ADCP measurements obtained along with a detailed image of the radio control boat and ADCP sensor used; (b) Map showing the hydromorphological features obtained from visual identification on a 2 m × 2 m regular grid; (c) Examples of sections selected for and outputs obtained from the Artificial Neural Network (ANN) training; (d) Map showing the hydromorphological feature classification obtained with ANN on a 2 m × 2 m regular grid.

Figure 4

Example of trained outputs for (a) Vegetation in bars; (b) Side bars with no vegetation; (c) Trees; (d) Erosion and (e) Riffle. The outputs portray the portion of the imagery selected for analysis and the pixels selected (pink) by the cluster algorithm.

Figure 5

Example of Artificial Neural Network (ANN) classification outputs obtained with the Leaf Area Index Calculation (LAIC) for a selected portion of the orthoimage. Pixels elected within each class are shown in pink. (a) Original image; (b) Visual classification for the points defined by a 2 m × 2 m regular grid; (c) Erosion; (d) Side bars; (e) Deep water; (f) Vegetation (all classes); (g) Riffles. The image is not to scale.

Figure 6

Classification outputs at each of the points defined by a 2 m × 2 m regular grid obtained with (Left) The Leaf Area Index Calculation (LAIC) Artificial Neural Network (ANN) and (Right) The visual identification for two sections within the study reach.