Detecting forest structure from space

Australia’s sheer size poses challenges for those conducting ecosystem monitoring programs – it is simply not feasible to have teams of people on the ground making observations at the temporal and spatial scales required. This means that satellite-based remote sensing of a wide range of ecosystem characteristics, which includes stocks of carbon and changes in those stocks, becomes essential if we are to observe not only the current status but also the effects of different land use and management policies and practices across our continent.

However, many of the internationally accepted methods and algorithms for interpreting satellite data perform poorly when applied to Australia. Our ecosystems, atmosphere and soils are very different in character, and the natural and human-mediated events and processes at work across our landscapes are unusual, compared to the systems for which these methods and algorithms were developed. For this reason, TERN’s AusCover facility is working closely with state, national and international stakeholders to improve the usefulness and reliability of satellite data, and the products derived from them, for Australian conditions.

One area of concern has been the amount of carbon stored in vegetation and how this is changing. For some time, remote-sensing estimates of the area of land covered by vegetation have been undertaken using green vegetation indices such as the Normalised Difference Vegetation Index (NDVI). However, these measures cannot tell us about the three-dimensional structure of woody vegetation, although this information is essential if we want to know whether the vegetation cover observed from space is, say, grass, woody regrowth or the canopy of a mature tall-eucalypt forest. We also need it to quantify carbon stocks and how they change over time.

AusCover researcher Dr Peter Scarth from the Joint Remote Sensing Research Program has just returned from a nine-week fellowship at the University of Aberystwyth in Wales, where he has been working with Professor Richard Lucas and other international colleagues to find a possible solution to this problem. Peter said that, starting with a number of remotely sensed datasets of Australia provided by the Japanese Aerospace Exploration Agency (JAXA), he’s come up with an integrative approach for monitoring vegetation height and structure that’s likely to have applications overseas as well as at home.

‘It turns out that if we combine satellite data from three sources – the Geoscience Laser Altimeter System ICESat instrument, the Phased Array type L-band Synthetic Aperture Radar (PALSAR) aboard JAXA’s Advanced Land Observation Satellite, and Landsat-derived Foliage Projective Cover – we get a more comprehensive view of the three-dimensional structure of the vegetation,’ Peter says.

‘This is a significant step forward from vegetation indices, as it will permit us to make a more accurate estimation of carbon dynamics and biodiversity, and lead to a better understanding of the response of ecosystems to change. We’re now looking at exactly how we can use this newly developed method to observe state and changes in forest structure and biomass, and refine these further so that it can provide a globally robust benchmark that can be applied across a wide range of forest and woodland types.’

The plan is to use this new method to collaborate with parallel programs at the CSIRO, and other Commonwealth and state-based monitoring, mapping and management agencies in a continent-wide effort to improve, validate and ground-truth estimates of standing biomass.

‘We’ve already conducted a pilot study near Injune in Queensland, and demonstrated that we could reliably and remotely detect forest structure in a complex landscape shaped by burning, regrowth and clearing,’ Peter says. ‘We hope that further collaboration with JAXA’s Kyoto & Carbon Initiative, which has supported much of our research in this area, will enable us to expand the trial to other parts of Queensland, New South Wales, Victoria and the Northern Territory.’

Internationally, these results are beginning to interest government agencies responsible for meeting national reporting obligations under the Kyoto Protocol and the UN Reducing Emissions from Deforestation and Degradation (REDD+) program. Agencies may incorporate future versions of the methodology into their standard systems for the estimation of routine carbon stocks and greenhouse gases.

At home, this research will contribute to the development of combined remote-sensing approaches to map and monitor carbon stocks across the continent. The results will also help us understand the impacts of natural events such as bushfires, or of policies and practices for managing biomass, biodiversity, and ecosystem goods and services. The better we can observe what’s happening, the better we can understand the underlying ecosystem processes – and the more effective we will become at managing the land sustainably. 

Published in TERN e-Newsletter July 2012

Share Article