About a year ago, I had the pleasure of visiting one of our Trimble eCognition customers in Brest, France. The team at the National Botanical Conservatory of Brest (CBN) has one of the largest eCognition software constellations in France and has taken on the task of meticulously mapping the sensitive vegetation throughout the region of Brittany in Northwestern France.
The article “Mapping the Big Vegetation Picture”, recently published in the May/June 2020 edition of GeoConnextion, reviews the work being done by CBN. Like many rural areas throughout the world, Brittany is experiencing the strain of population growth – the saturated transportation infrastructure and increased demand for housing are threatening farmland and natural environments that make up 65% of the region. In response, “planners , authorities and biodiversity specialists in Brittany have been pressured to balance this urban sprawl with environmental protection policies,but existing vegetation maps of the region have made that difficult”. Vanessa Selling, a project manager at CBN, indicated that “vegetation maps are an essential tool for stakeholders”.
As is so often the case, the method for creating such maps had been primarily manual – painstakingly transferring notes recorded in the field into a GIS. This time consuming approach made it extremely difficult to generate an “accurate picture of the vegetation landscape as a whole”. And this is where Sellin and her team come in, she has “established a viable semi-automatic mapping method combining orthophotos, GIS data and object-based image analysis software (OBIA)”. It is no surprise that CBN has chosen Trimble eCognition as it provides the unique data fusion technology to seamlessly and flexibly work with various geospatial data sets.
In 2016, eCognition was put to the test at CBN as the team selected the Armorique Regional Natural Park (PNRA) as a trial area for their OBIA approach. Over 1,700 aerial images (RGB+NIR) from France’s national mapping agency (IGN) were acquired. An orthomoasaic was generated from the 0,5 meter resolution images and combined with two texture-specific images to enhance the discrimination between “similar looking vegetation varietals”. In addition, 10 different GIS shapefiles were integrated into the eCognition project. These layers included various land cover information such as the location of bogs, sand dunes and diverse forest types. Adding “this vector information enables eCognition to better discriminate among vegetation types with homogeneous spectral signatures such as wet and dry heathlands”.
A hierarchical classification approach was chosen for the benefits it provides in processing and transferability. “Predominantly using the texture images, brightness and a normalized difference vegetation index (NDVI), eCognition moved from one orthomosaic tile to the next”. In an initial classification, the “easiest” classes (i.e. roads, buildings and water-bodies) were addressed. Subsequent classification steps took a deep dive and broke up the initial classes into more detailed specific vegetation sub-classes – here the broad parent class of forest of dune was split into more difficult classes such as shrub or heathland. In total 27 different classes were extracted and then exported to ArcGIS where they were used as input for the final map with a 1:25,000 scale.
Needless to say, Sellin and the CBN team were happy with the results and eCognition experience – “One of the powers of eCognition is its ability to take in any spatial-based data and classify whatever you tell it to,”says Sellin.“It’s smarter and more efficient and definitely more accurate than pixel-based image processing software. And it proved that it’s capable of producing large-area coverage vegetation maps at a large scale with high precision.”
As a result of the successful pilot project, in 2018, CBN with the financial and administrative support of 10 regional partners hired 4 additional staff members and launched a 2-year €600,000 project to map Brittany’s vegetation wall-to-wall.
In March 2018, the team began with the department of Finistère then moved on to Ille-et-Vilaine in September 2019 and Côtes-d’Armor’s map was released in January 2020. The individual vegetation maps are available online via the CBN website. The team is currently working on the map for Morbihan and expects to publish this in June 2020.
The maps being produced by Sellin’s team provide a “a holistic view planner and managers have never had before, and they have been rushing to see and use the data”. Sellin concluded in the interview that “there was no way we could’ve produced this regional map this quickly without eCognition… The OBIA approach is a great complement to our small-scale traditional mapping methods. It’ll now be our standard tool for classifying and mapping major vegetation types at larger scales”.
I would like to thank Vanessa Sellin and here team for taking the time to speak with us and for graciously hosting me in Brest last year where we had a chance to sit down and talk shop and examine their workflows. In addition, I would like to thank our regional Trimble eCognition channel partner, GeoSystems France, for assisting with the interview and logistics of my visit.