Conor Jones, BMT

Conor Jones is a Principal Scientist at BMT, where he has worked for the last 10 years on a range of environmental consulting projects.  His prior research at the University of Qld examined interactions between parasites and fishes on the Great Barrier Reef, and the recruitment of coral reef fishes. He enjoys a range of technical tasks including electrofishing, commercial diving, acoustic mapping, remote sensing, underwater photogrammetry and drones.  He also loves data analysis, GIS and working with the numerical modellers at BMT. His role at BMT includes managing projects for Local Governments, mining and ports, the State Governments and Non-Government Organisations. He is the monitoring program director for remediation processes at Douglas Shoal for the Federal Government, which is challenging but extremely rewarding.

Abstract:

WB has conducted annual seagrass surveys in the vicinity of sand sourcing works for the Sunshine Coast Council in the northern Pumicetone Passage since 2017.  Community composition, percent cover, and extent have been mapped based on ground-truthing of recent high-resolution aerial imagery.

Community changes have largely been driven by the presence and absence of colonising species amongst relatively unchanging Zostera muelleri meadows.  After a low point in extent following cyclone Debbie in 2017, total extent increased through to 2019 and has remained steady through to 2020.  These increases in total extent are consistent with increasing extent elsewhere in Moreton Bay.

To this point, meadow extent, cover, and community composition have been manually digitised using ground-truthing data and high-resolution RGB imagery.  While greater spectral information is freely available from Sentinel2, pixel resolution is too coarse for the required mapping resolution of this project.  The presence of wrack and benthic microalgal mats have been difficult to discern from seagrass; requiring extensive ground-truthing effort.  Computer vision machine learning offers great potential to improve the speed and objectivity of digitisation.  Initial results of model-training using data from the northern Pumicestone Passage are presented.

Supporting Documents