Multispectral Image Analysis for Vegetation Monitoring

South West, Western Australia

Vegetation health has traditionally been monitored using field-based observations at discrete sampling locations. In contrast, remotely sensed data provides an ability to detect changes in vegetative health across the entire area.

Situated on the edge of the Gnangara Groundwater Mound, the project area is in close proximity to wetlands with high biodiversity value. Groundwater abstraction in and around the project area has occurred for many years and the impact of mining related drawdown on the native vegetation was previously monitored using ground-based observations.

In vegetated areas, remote sensing indices like the Normalized Difference Vegetation Index (NDVI) measure changes in photosynthetically active biomass. Substantial negative changes in NDVI can be indicative of areas with declining vegetation health and substantial positive changes often indicate increasingly active and healthy vegetation. For this analysis NDVI was calculated for 2014 and 2015. Changes in vegetation health were categorised using inter annual differences in NDVI. A cluster analysis identified areas of negative NDVI change, most of which were associated with infestations of Phytophthora dieback mapped in previous on-ground work. Importantly, the cluster analysis indicated there were likely new areas, previously unmapped, that may be infested with Phytophthora dieback.

There was little evidence of groundwater abstraction impacts on vegetation within the borefield. With a baseline dataset for change now established, future analysis will have increased statistical power to assess negative impacts associated with groundwater abstraction. This provides an evidence-based framework to support decision making.

The detection of new dieback infested areas and the non-significant response of vegetation to groundwater abstraction for 2014-2015 demonstrates how remote sensing enables the timely detection of changes in vegetation health. Critically, timely detection equates to timely intervention for managers, if required.

Multispectral Image Analysis for Vegetation Monitoring
Multispectral Image Analysis for Vegetation Monitoring
Multispectral Image Analysis for Vegetation Monitoring  : Image 1
Multispectral Image Analysis for Vegetation Monitoring  : Image 2