This project was made possible with a Contribution Agreement from Natural Resources Canada through the GeoConnections program. Deliverables of the project included the development of a document detailing best practices guide for the collection of high quality, positionally accurate, and interoperable geospatial data using Unmanned Aerial Vehicles (UAVs). Another deliverable is to produce a document reviewing existing metadata creation tools as part of the interoperability component. The focus was the collection of geospatial data in northern communities. However, case studies and testing were conducted in Ottawa (Ontario) and in Nicolet (Quebec). More information about the project can be found in this GeoConnections announcement.
Three sites in Ottawa and one site in Nicolet were selected for the case studies. One of the sites in Ottawa was located in the Marlborough Forest and the two others in St. Isidore. These sites were selected to enable data collection over a mixed forest and over agricultural fields, respectively. The site in Nicolet was selected because it allowed for data collection over a wetland. Furthermore, this site was already used in a previous project by one of the collaborators which will allow the comparison of results from both projects.
Lessons learned and experience gained from the case studies were used to guide UAV data collection in Kuujjuaq, an aboriginal community in northern Quebec. Positional accuracy was determined by laying photo targets on the ground at appropriate locations within the sites. Photo target locations were determined using a Trimble ProXRT GPS with realtime differential corrections provided by OmniStar. This setup gives horizontal accuracy of about 10-cm and vertical accuracy of about 30-cm. Photo targets were divided into Ground Control Points (GCPs) which were used in performing bundle adjustment and Check Points used in evaluating the positional accuracy of the resulting orthomosaics.
This project was for the City of Peterborough Public Works Department which also has responsibility for Parks and Trees. The project site is a woodlot approximately 6 hectares in size that is situated on the southern shores of the Otonabee River and directly beside a snow dump and the city’s Waste Water Treatment Plant (WWTP). The project is part of an on-going investigation into the apparent poor health of some trees in the woodlot.
ASG Mapping Ltd collected high resolution aerial imagery using a multirotor UAV system on three occasions during the growth season to produce a series of orthomosaic imagery of the woodlot foliage so that tree foliage can be compared over the summer months. In order to allow for overlaying of other geospatial/GIS data over the orthomosaic imagery the data collection was designed to meet a horizontal accuracy of better than 1 metre.
A multirotor UAV with its Vertical and Takeoff Landing (VTOL) capability was used to collect the aerial imagery due to the site’s proximity to WWTP buildings and Highway 7. A Sony Nex 5 mirrorless digital camera was used for the first two data collections while a MicaSense RedEdge 5-band multispectral camera was used for the third data collection. Photo targets were laid out within the site to serve as Ground Control Points (GCPs) and as Check Points.
The resulting imagery clearly shows the distribution of dead and unhealthy trees within the woodlot. It appears that they follow the path of the water outflow from the snow dump and down towards the river bank. This is critical and important data that biologists, foresters, and other specialist can use in their ongoing investigation of the factors contributing to the unhealthy trees in the woodlot and ultimately finding solutions to mitigate the effects.
This project was done in collaboration with Morrison-Hershfield Limited and used Landsat satellite multispectral data to conduct a change detection analysis of vegetative cover in an area in Manitoba. A 12 category landcover classification scheme was used in the project. Unsupervised classification (ISOCLUST algorithm) were done on 2015 and 2014 Landsat imagery using TerrSet software.
Pansharpened and Composite (natural colour and false colour) images were derived from the Landsat imagery to help build the training set necessary to run the unsupervised classification. The results were then compared between each other (2014 vs 2015) and with previous unsupervised classification results that were done by other organizations using Landsat data taken in 2006 and 2011. The Change Detection analyses were performed using ArcGIS Image Difference algorithm together with tools (Map Algebra, Local and Zonal tools) available within the Spatial Analyst extension.
Results of the change detection analysis seem to show that areas covered by Grass has increased over the years. Tree cover appears to be decreasing from 2006 to 2014 with an increase recorded in 2015. Fluctuations in the amount of Burned and Built/Bareground areas as well as Water and Water-Dependent areas are observed.
A limitation of the the change detection analysis is that no ground truth data was collected and used to check and validate results. In the report, a recommendation was made to collect ground truth data to have better confidence in the results of the analysis. The use of Unmanned Aerial Vehicles (UAVs) may be an appropriate method of collecting ground truth data efficiently.