Monitoring toxic cyanobacteria Lyngbya majuscula (Gomont) in Moreton Bay, Australia by integrating satellite image data and field mapping

Abstract

Large-scale blooms of Lyngbya majuscula (Gomont) have occurred throughout Moreton Bay (south-east Queensland) and have been documented since 1997. L. majuscula is a toxic cyanobacteria which fixes nitrogen and is found attached to: seagrass, algae and coral. The toxic and smothering nature of L. majuscula has affected human and environmental health in sensitive coastal ecosystems. To reduce these impacts, monitoring is an essential component of studying the origins and development of L. majuscula blooms. An accurate and cost effective means to map the extent of a bloom and its biophysical properties is needed. This study presents an operational approach for mapping the extent of L. majuscula blooms in the clear and shallow water regions of Moreton Bay, eastern Australia, from a combination of field and remotely sensed data sets. The ability to discriminate L. majuscula from other substrate types over a range of depths was first examined using detailed field reflectance spectra, measured optical properties of Moreton Bay waters and a radiative transfer model (Hydrolight 4.1). A two-stage process was then used to map L. majuscula. The spatial extent of L. majuscula and other major substrate types was first recorded from a boat-based survey by marine park authorities using point-based GPS measurements. This sampling was timed to coincide with an overpass of the Landsat 7 ETM+ sensor. When the results of the boat-based mapping detected more than 25% L. majuscula cover in the study area, a cloud free Landsat 7 ETM+ image was acquired for that date. In the second stage of mapping, selected field survey data provided the basis for a supervised classification of the ETM+ image data to map L. majuscula. Effort and accuracy assessment of both field and image mapping methods indicated a trade-off between areal coverage and mapping accuracy. The Landsat 7 ETM+ based mapping procedure provided 100% areal coverage with 58% accuracy. In contrast, the boat-based field survey method covered only 0.5% of the study area, but with almost 100% mapping accuracy. The approach outlined in this work has been adopted as a standard operating procedure in Moreton Bay. This study illustrates how remote sensing can be combined with field monitoring, to provide marine park authorities with useful information to understand and manage blooms.