Dr Tim Brown1,2, Ruaraidh Mills1,2, Kirsty Yeates2, Dr Stuart Ramsden2,3, Ming Dao-Chia2, Dr Justin Borevitz2
1Australian Plant Phenomics Facility, ANU node, Acton, Australia, 2Research School of Biology, Plant Science, ANU, Acton, Australia, 3NCI VisLab, Acton, Australia
He is currently the Director of the ANU node of the Australian Plant Phenomics Facility. His work focuses on developing open source hardware and software pipelines for enabling high throughput phenotyping, high resolution monitoring and visualization of environmental and research data. He is a co-founder of the Australian Phenocam Network and creator of EcoVR, a software package for modelling research sites and visualizing time-series environmental data in VR using game engine software.
From 10-min interval satellite data to mm-resolution 3D hyperspectral timelapse scans of crops, new technologies are providing researchers with ultra-high resolution datasets, but we often lack the tools to make best use of these new data types.
The visual effects industry has created powerful software such as Houdini for building photo-realistic 3D worlds. As these tools increasingly integrate real-world physical processes to improve realism, they have begun to represent a viable, scientifically robust platform for modeling ecosytems. Houdini has implemented advanced tools for procedurally generating realistic landscapes that grow mixed-species ecosystems using competition models for water and light resources. Landscape models can integrate geological processes such as stratification, effects of slope on erosion, and weathering (see vimeo.com/273986776) and such models can be easily created from real DEMs and topographic maps.
Procedural modelling is relatively computationally light but can scale to areas far larger than the earth, for example the game No Man’s Sky features a procedurally generated universe with 18 quintillion planets (10^19) with unique geologies and planetary ecosystems.
We are exploring using these methods to model real-world farms and ecosystems using scientifically valid growth models parameterized by satellite and ground-based measurements for modelling regenerative agriculture and carbon drawdown scenarios. We propose to operationalise these workflows so ecosystems can be rapidly modelled from any data source that is available via API. This approach has the potential to transform the availability of modern high precision datasets and presents a significant opportunity to improve the modelling tools available to researchers and managers.
Tim Brown has a PhD from the University of Utah (USA), in complexity theory and modeling of self-organized swarming behavior in New World army ants. From 2006 to 2012 he founded and ran an environmental consulting business developing phenocams and billion-pixel resolution timelapse cameras. In 2012, Tim moved to the Australian National University as a postdoc, to develop high throughput plant phenotyping systems for the the Borevitz Phenomics Lab at the Research School of Biology.