Assessing the Potential of Remote Sensing in Support of Phytophthora Work

Phytophthora are a large group of pathogens that plant disease and many species of tree. At this moment in time there is some concern over the threat they pose to trees and forests in the UK and other plants and shrubs such as Rhododendron and Bilberry.

In a recent project Environment Systems working with Callen-Lenz Associates were commissioned by FERA (Food and Environment Research Agency) to look into how remote sensing using automated classification techniques could be used to map the precise locations of species particularly prone to Phytophthora infection. This focused on Larch (Larix), Bilberry (Vaccinium myrtillus) and Heather (Calluna vulgaris). The project then looked at the possibilities of mapping the advance of any disease over time.

This project involved the use of satellite imagery and data plus aerial photography from 2006 and 2009 surveys including Near Infra Red. For the Wales study area, plans of forestry planting, including tree species and proportions of tree species within each ‘coupe’ or area of forest were used. There was also a remit in this project to look at the potential for the use of Unmanned Aerial Systems (UAS) for accurate Larch crown identification and health monitoring in targeted areas.

To effectively map Larch we first needed to differentiate between coniferous forests and all other woodland, something we could do effectively using satellite data and our own established techniques. Further analysis could then identify the Larch within the coniferous plantations. Other techniques using ‘fuzzy logic’ made it possible to produce a map whose objects matched the criteria for the heathland species i.e. the Bilberry and Heather. This necessitated the use of contextual data such as slope and short wave infrared satellite data.

The higher resolution aerial imagery increased the spatial delineation and accuracy of the identification of Larch species within coniferous woodland which relies on comparison between on leaf and off leaf season surveys. Based on these findings the UAS alternative, for specific areas of risk starts to look attractive especially given that the Forestry Commission has just released a very detailed inventory of woodland in the UK.

The UAS data capture process, undertaken by Callen-Lenz, sought to capitalise on the amount of useful data gathered by maximising the operational radius and altitude plus the flight duration. Using a standard RGB camera and Visible Near Infra Red (VNIR) sensor multiple flights were undertaken over one test area. The RGB imagery was mosaiced together for detailed analysis and used to generate a high resolution digital surface model capable of showing tree heights. Suffice to say the results have been very promising.
This pilot study has proved the benefit of looking at a multi-scale hierarchical approach, utilising as much existing data as possible. It also concluded that there was a coherent and cost effective staged approach that could be recommended for rolling out across a region or country.

• Identification of woodland, including coniferous
• Through satellite interpretation (broad scale only)
• Through summer photography with NIR band
• Through use of the new National Forest Inventory
• Identification of individual Larch from either leaf on / leaf off air photography (where available)and or supplemented with the use of UAS imagery – particularly on areas on the front line of disease spread.

The project report has now been published and can be downloaded from here.