Forest Research is working with Rezatec, a leading UK geospatial data analysis company, to help develop new approaches to identifying and monitoring plant health. Now in its second phase, the SAPPHIRE project is exploring the extent to which tree ill health can be detected using remote sensing data.
The potential of Earth Observation data
The massively increased availability of data from satellite operators is stimulating the development and uptake of new products and services that use Earth Observation (EO) data to monitor environmental change. A huge range of new data is freely available from the Copernicus programme, which collects the data through its Sentinel satellites, using various technologies. For example, Sentinel-1 uses radar, with the all-weather advantage of being able to see through cloud, and has twin satellites in the same orbit, resulting in a revisit time of just six days.
To explore whether this data could be turned into an effective tree-health mapping and assessment tool, Forest Research’s Business Development Manager, Alison Melvin, turned to Rezatec and formed a collaborative team to secure funding from the ‘Space for Smarter Government Programme’ for project SAPPHIRE (‘Space Applications for Precision Plant Health Information, Response and Evaluation’). Rezatec, headquartered within the UK Space Gateway in Oxfordshire, has expertise in aggregating large amounts of diverse data from satellite, airborne and ground instruments to produce geospatial data that meets users’ specific needs.
The SAPPHIRE project has focused on working with tree health experts to achieve the precision mapping required. In the first phase, the team aimed to differentiate between oak and ash trees in selected study sites using their typical annual phenological pattern. Rezatec used data from various EO sensors (Sentinel 1 C-band SAR, Sentinel 2 multi-spectral sensor and Landsat 8) to map the distribution and abundance of these species. These EO data sets were complemented with existing ground-based and accessory data (LiDAR and National Forest Inventory) provided by Forest Research, and integrated using data mining approaches.
The results were very encouraging, with tree species maps achieving a minimum classification accuracy of 80%. Rezatec also developed a method to differentiate broadleaved trees that may be exhibiting stress associated with pathogens.
In September, the project team received funding for a follow-on project and immediately started the Phase II project, working on an expanded range of tree species and sites, and validating identified areas of tree stress against known tree health information.
Tim Vallings, Head of Forestry at Rezatec explains: “With Phase I of this project we demonstrated the feasibility and commercial potential of using satellite data analytics to provide real insight to tree health issues. We have also applied this to the conifer sector and launched first products at the APF 2016, the UKs largest Forestry, Woodland and arboricultural exhibition. By Spring 2017 we will be in a position to expand on these tools within the broadleaf sector, through Rezatec’s geospatial data web portal, which will mean that users can start to view tree health information alongside other inventory information, such as species distribution, tree height and windblow damage for many of the species in the UK; an exciting development for the sector.”