Dr Leigh Tait1, Dr Hamish Biggs1, Dr Monique Ladds2, Mr Francois Thoral3, Dr Mads Thomsen3, Dr Matt Pinkerton4, Prof David Schiel3
1NIWA, Christchurch, New Zealand, 2Department of Conservation, Wellington, New Zealand, 3University of Canterbury, Christchurch, New Zealand, 4NIWA, Wellington, New Zealand
Long-term ecological datasets underpin robust assessments of the impacts caused by climatic and anthropogenic stressors. Yet, such datasets are distressingly rare and un-evenly distributed. Remote sensing provides a solution to these limitations by passively providing imagery on a global scale which can be analysed to detect and delimit macroalgae. However, remote datasets are frequently applied to a very small subset of macroalgal species or habitats and currently provide little or no information on biodiversity metrics or shifts in species composition. Here we combine in situ sampling with low-altitude hyper- and multi-spectral sensing to generate realistic spectral profiles for multiple macroalgal species and functional groups. We then apply spectral un-mixing algorithms to satellite imagery and assess how closely coverage estimates resemble classification outcomes from low-altitude imagery. We show that low altitude multispectral imagery can produce classified maps with up to 96% agreement with spatially referenced in situ samples of more than 10 macroalgal species or functional groups. We will assess the potential for spectral unmixing to enable passive macroalgal biodiversity monitoring and discuss the challenges and potential solutions for applying satellite products to coastal rocky reef habitats.
Biography:
Leigh Tait is a marine ecologist based at New Zealand’s National Institute of Water and Atmospheric Research, in Christchurch New Zealand. With a key interest in the biogeochemical functioning of macroalgal assemblages, Leigh has been working on providing scalable tools to help estimate the impacts of key stressors on primary productivity of macroalgae, particularly the influence of coastal sediments and climate change.
