Enormous effort is invested in monitoring the Great Barrier Reef (GBR), but data collection is currently fragmented over dozens of publicly and privately funded organisations, with data collected using different methods and for different purposes. As a result, the data are rarely analysed together.
This work led by Associate Professor Erin Peterson developed a new weighted Bayesian spatio-temporal modelling framework used to integrate data from professional and non-professional data sources, with various levels of quality. The methodology also accounts for multiple influential factors such as (e.g. cyclones, coral bleaching, and crown of thorns starfish) and allows us to generate predictive maps of coral cover across the whole-of-the GBR and over time.
We found that integrating data from different sources including data elicited by citizen scientists significantly increases the quality of the model predictions and produces valuable measures of uncertainty.
This work is part of the Virtual Reef Diver project (https://www.virtualreef.org.au/), which uses the contribution of thousands of volunteers to monitor the GBR and has so far produced more than 3 million classification points.
- The article was published by the journal Environmental Modelling & Software and it can be found at https://www.sciencedirect.com/science/article/pii/S1364815219309582.