How drone technology is helping Ireland’s River Moy retain its status as the country’s premier salmon river
Straddling the counties of Mayo and Sligo, the Moy offers a wide variety of angling and has long been famous as Ireland’s premier salmon river. Approximately one hundred kilometres long, the Moy attracts visiting anglers from around the world who consider the west of Ireland a salmon fishing paradise.
The Ballina ‘canal stretch’ of this productive river is owned by The Ballina Salmon Anglers Association which works hard to ensure that the area retains its angling value and offers those fishing it a rewarding experience. However, this part of the river is particularly susceptible to native pond weeds such as Curly Pondweed (Potamogeton crispus) and Perfoliate Pondweed (Potamogeton Perfoliatus) which have become problematic due to the settlement of fine sediment on the river bed from the upper catchment. Consequently, the open water has become choked by the mostly submerged leaves of these weeds which have also created conditions that support the proliferation of the invasive fish species, roach. Curly and Perfoliate pond weed are therefore considered noxious and, in this case, sufficiently troublesome for the Association to commission a full intensive survey of the canal stretch. This survey would determine the extent of the problem by providing data from which the surface of raised sediment can be calculated and a 3D profile of the river banks created.
Aerial Agri Tech, a company that specialises in providing high resolution spatial and temporal data across a range of industries, was commissioned to do the work under the guidance of its GIS and remote sensing technician, Cian Gallagher. The company aimed to meet the Association’s objectives by supplying the following services:
- A drone survey of the Moy River
Deliverables – geoTIFF orthomosaic and KML tiles (Google Earth)
- A calculation of surface area cover of weed and raised beds of sediment
Deliverables – NDVI*, land classification map, land class shapefiles and surface area calculation
- An elevation model displaying river bank heights and river profile
Deliverables – digital surface model, 3D point cloud (X,Y,Z), volumetric calculation and contour lines
For this project, Aerial Agri Tech used a senseFly eBee Ag that it had purchased from Irish distributor, KOREC. In seeking a surveying drone that could provide high resolution multiband images, Cian selected the eBee Ag rather than a multirotor drone, finding it more suitable for the larger areas he would be surveying in agricultural environments and also for its flexible lightweight frame making it easy to carry – a small but important factor.
Surveying the River Moy and dealing with water reflection
A flight plan was created utilising the default difficult terrain setting with an 85% lateral and 75% longitudinal overlap and the camera set to take images at a 15° angle. These settings ensured all relevant features were captured.
The survey was conducted at 5:30am as conditions were calm with little or no wind and Cian believed that an early survey would minimise any reflection issues that may have been experienced on the surface of the water if the sun was directly overhead. The total survey took Cian approximately 80mins (4x 20minute flights), 2 with the RGB camera and 2 with the NIR camera along the same flight plan. Using both cameras allowed Cian to record as much data as possible and to stack the layers in any array he wished to create a multispectral image.
A period of ground truthing was carried out once the imagery had successfully been captured. The collection of ground-truth data enabled calibration of remote-sensing data and aided in the interpretation and analysis of what was being sensed. This was done by walking the river backs and noting characteristics along the water surface. Photographs were also stored as a reference.
Cian then spent 20 minutes carrying out an initial rapid processing of the imagery to ensure that the images were suitable and the entire survey area had been captured to avoid revisits.
However, despite Aerial Agri Tech’s best efforts to minimise the reflection of the water surface, there were issues during the classification of the imagery. Therefore, in order to reduce the reflection, each image was individually analysed and a selection of the best images chosen and processed.
The river was then divided up into 200m sections and the NIR camera recalibrated, based on each section. This gave more accurate results for detecting weed, especially when running an NVDI process.
Deliverables and using the imagery
Once analysed and processed, the images were passed on to the client in a map format. Statistics on weed area and distribution were also calculated and passed to the client.
The imagery was divided into specific land classes with each land class division based on its spectral properties. Using the specialised NIR camera, weed was detected and classified which was particularly effective along the river given the stark contrast between the spectral properties of water and vegetation. Similarly, the raised beds of sediment displayed contrasting spectral values compared to their surroundings.
The Normalized Difference Vegetation Index (NDVI) was used to detect plant “greenness” or photosynthetic activity within each land class. This index is based on the observation that different surfaces reflect different types of light. The NDVI is measured on a scale of -1 to 1, where -1 represents areas with no vegetation and 1 represents areas of dense vegetation.
Once the weed and sediment classes had been detected and analysed, the imagery was imported into a GIS environment and the land classes were converted into a polygon format. The GIS allowed the areas of these classes to be accurately calculated providing the client with exactly the information it needed to plan its eradication of both weed types.
Cian concludes, “The eBee Ag was the perfect solution for this project enabling me to capture images in a range of formats in an efficient and user friendly manner. Flight plan, flying and post processing can all be done in less than a working day leaving plenty of time for post processing. The technology has so much potential and in KOREC I have a partner that’s as enthusiastic as I am to explore the eBee’s potential.”
* Normalized Difference Vegetation Index
All information was kindly supplied by Cian Gallagher of Aerial Agri Tech who would like to thank Ross Macklin, Senior Ecologist & Fisheries Scientist at Triturus Environmental Services and Gerhard Seck, Club Secretary of Moy Anglers, for their assistance with this project.