Director of the Australian Plant Phenomics Facility, ANU node
Research Fellow with the ANU Centre of Excellence in Plant Energy Biology
Background and Research
My interests are in using emerging camera technologies to facilitate next generation ecology research across scales from genomics studies in growth chambers to high-resolution phenological research at the landscape level. Growth chamber studies allow highly targeted dissection of the genetic basis of plant growth and development. At the ecosystem level, the Next Gen ecosystem monitoring tools we are developing enable landscape-scale phenology studies and targeted genetic analysis of thousands of individuals in wild plant populations.
Throughout my academic career, I have tackled high-complexity research questions requiring solutions that extend the boundaries of what is technically possible. My PhD research, modeling swarming behavior in Eciton burchellii army ants (1999-2006), combined challenging fieldwork with advanced computational modeling and numerical analysis. I filmed army ants in the field in Costa Rica, taught myself to program and then wrote software to track manually the position of 30,000 ants in the swarm videos. With these data I built the first individual-based model of army ant swarming behavior parameterized from field-collected data.
Ecosystems emerge out of a complex mix of environmental drivers affecting interconnected individuals and populations with unique genetic histories assembled by geologic history, interacting across scales from the microscopic to the landscape. Conservation ecology is challenged by the fact that the ecosystems we seek to understand are large and highly complex. Yet standard research tools typically provide low complexity, small spatial resolution data with little or no information on the genetic structure of the population.
After finishing my PhD, I co-founded my company TimeScience in 2006 to create better tools for scientists to record, analyze and visualize long‑timescale environmental change. Over the last 6 years I have worked on numerous projects that have improved the ability of researchers to collect ecological data, understand bio-control efforts and invasive species impacts, improve resource management and enhance public outreach and science education (see Recent Projects).
My scientific interests have also evolved over this time from an initial goal of visualizing long-term environmental change, to using emerging camera and internet technologies to enable very high resolution, high throughput phenotyping of plants across scales from highly controlled greenhouse studies to multi-hectare field sites. The Internet and emerging data standards projects can give us the ability to rapidly share these types of data nationally and globally. I strongly believe that these are the tools that will enable next-generation ecology, allowing us to understand both ecosystem function and crop science at the level required to solve the ecological challenges of this century.
I pursued a PhD focused on complexity theory in part to better understand how the emergence of the Internet and the massive increase in computing technology and human interconnectivity might impact the rate at which humans address challenging social and environmental problems. The problem-solving abilities of complex networks are directly tied to the rate and quality of information discovery, and the rate that this information can be transmitted through the network (Camezine et al. 2003. Self-organization in Biological Systems). When ecology is viewed from a network-science perspective, I believe it is clear that solving the pressing ecological challenges of our times requires (1) an exponential increase in the quantity and complexity of data that we are able collect and analyze, and (2) highly accessible, networked data sharing tools that provide researchers with the means to rapidly build off of current and previous work.
The rapid rate of technological change gives us the opportunity to develop low-cost data discovery tools that can be widely used by others to advance phenomics across scales. At the landscape and field site scale, the analysis tools and camera systems developed for greenhouse work can be adapted to provide high-resolution “near-remote sensing” data streams to improve novel landscape monitoring capabilities and enhance existing data streams. This will enable us to examine population structure and target outlier populations for genetic analysis. My work with the Borevitz Lab represents where I can best apply my unique research skills in new technology development, advanced numeric modeling and data analysis to addressing core scientific questions in phenology and plant developmental genetics.
TraitCapture (TraitCapture.org): High throughput phenotyping with dynamic environmental conditions and automated Time-Lapse
NextGen data visualization tools
- TraitCapture – is “seeds to traits” pipeline for high throughput phenotyping of plants in growth chambers and glasshouses with automated capture from DSLR’s and automated image segmentation and analysis of 2100 plants continuously
- In the growth chambers we have developed the “SpectralPhenoClimatron” (SPC) – retrofitted Conviron growth chambers with 7 or 10-band multispectral LED lights, multiple DSLR cameras and dynamic environmental and lighting conditions.
- The SPC system lets us dial in semi-realistic growing conditions for almost any region on the planet and grow plants under those lighting, temperature and humidity conditions.
- We can monitor up to 300 plants per chamber (300×300 px per plant resolution) at 5-minute intervals over 7 growth chambers (~2,000 plants).
- All data is organized and visualized via the TraitCapture.org website
- In addition to expanding the capabilities of our existing online and kiosk systems, we are actively testing the awesome new crop of novel interfaces like the LeapMotion controller, GearVR, Oculus Rift and Vive VR.
TraitCapture high throughput phenotyping software pipeline
- The TraitCapture phenomics pipeline takes incoming DSLR images, color corrects and segments them and delivers standard phenotypic data outputs (green area, growth curves, etc) to our GWAS pipeline for further analysis.
- G2P Project Page
- Following on the the very successful AusPheno conference in Canberra, Sept 2016, we have been very interested in the emerging use of shipping containers with LED’s for building highly efficient controlled plant growth environments. This is the next frontier in high efficiency plant growth systems for research and Vertical Agriculture.
- In June 2017 we were been funded through an ANU Major equipment grant and $1.2million in GRDC funding to install 4 “growtainers” at the Research School of Biology at ANU. This will add the capacity to grow to 2,000-6,000 plants in dynamic environmental conditions with real-time phenotyping.
- One container will be off the grid and run completely on solar power.
Glasshouse (greenhouse) timelapse camera systems
- I have equipped the ANU NCRIS-funded Eucalyptus glasshouse with 30 5-megapixel IP cameras (5 per room) to enable researchers to monitor their projects.
- There will be 5 cameras per room in 6 rooms. Users will be able to log in and watch live timelapse of their experiments from 5 different camera views.
National Arboretum Projects
National Arboretum Phenomic Sensor Array (PESA)
- We received an ANU Major Equipment Grant to instrument the ANU research forest at the National Arboretum with a “next-gen” phenomic sensor array.
- Monitoring systems include megapixel and gigapixel time-lapse cameras (see gigavision.org); a microclimate-mesh sensor network that measures temperature, humidity, and PAR, soil moisture and temperature below ground and minute-resolution tree growth at 20 locations.
- All data will be streamed live online to a data portal.
- Click here for a 3D point cloud of the ANU forest at the National Arboretum
- Main project page
Online visualization of point clouds
- A common output of UAV 3D reconstruction data is time-series point clouds. These type of data are hard to work with and even harder to visualize. We have implemented the open source Potree package to enable online visualization of point cloud data on the TraitCapture website
EcoVR – Virtual and Augmented Reality tools for visualizing high-resolution time-series data in 3D on the landscape
- Using data from the National Arboretum PESA array, we have developed Virtual Reality 3D model of the National Arboretum ANU research forest.
- The user can view time-series data from the sensor arrays displayed on the VR model of the arboretum.
- Tree size, color and position data are all set from UAV Quadcopter data.
- EcoVR project page
Multi-billion pixel (gigapixel) resolution timelapse
Gigavision Timelapse Camera for tracking plant-resolution phenology in the field
- Gigavision Project Page
- Current work:
- Through the Arboretum Phenomics Sensor Array funding, we are developing a new gigavision system with improved PTZ hardware and a new visualization and data collection interface that streamlines the Gigavision stitcher and visualizations tools. We anticipate deploying the first new Gigavision camera at National Arboretum in Canberra, ACT in Jan/Feb, 2015.
- The gigavision control system will be designed to support any PTZ camera system that can be controlled via URL. Initial hardware we will support is the J-Systems 2MP PTZ system and the AXIS Q-series PTZ cameras (or any other VAPIX compatible PTZ systems.
- Below is a recent 550 megapixel panorama from the Arboretum, assembled from 770 2MP images. The camera can shoot one of these every 15 minutes although we are only capturing one per hour.
Time-Series gigapixel imaging with Gigapans
- The Gigapan is a low cost sommercially available robotic pan/tilt head that makes it super-easy to capture huge panoramas
- I shoot a lot of Gigapans and I am working to set up time-series gigapixel capture locations at our field sites around NSW.
- My gigapans are all online here
Australian Phenocam Network — Phenocam.org.au
- I have been working with collaborators at the Australian Supersites network, (part of Australia’s TERN network) to develop data standards and a web portal for an Australian Phenocam Network.
- We have recently (mid-2015) completed an initial data ingest which required re-organizing, time-stamping and re-sizing 140,000 images from 22 cameras at 7 Supersites (105GB of images).
- To process the images we had to develop some pretty sophisticated code for renaming and downsizing thousands of folders of non-timestamped images via exif data.
- Our exif2TimeStream processing code is available on the Borevitz Lab Github site, please contact us if you are interested in using it or collaborating on further development.
- All camera data is now available to view online in our timelapse player at Phenocam.org.au.
- Towards long-term standardised carbon and greenhouse gas observations for monitoring Europe´s terrestrial ecosystems: a review. Franz, D. et al. 2018. International Agrophysics, 2018, 32(4), 439-455. doi: 10.1515/intag-2017-0039 (PDF)
- Deep Phenotyping: Deep Learning For Temporal Phenotype/Genotype Classification. Sarah Taghavi Namin, Mohammad Esmaeilzadeh, Mohammad Najafi, Tim B. Brown, Justin O. Borevitz. Plant Methods. 2018. 14:66 DOI: https://doi.org/10.1186/s13007-018-0333-4
- Variation in Leaf Respiration Rates at Night Correlate with Carbohydrate and Amino Acid Supply. Brendan M O’Leary, Chun Pong Lee, Owen K. Atkin, Riyan Cheng, Tim B Brown, A. Harvey Millar. Plant Physiology. June 2017. DOI: https://doi.org/10.1104/pp.17.00610
- Moore, Caitlin E.; Brown, Tim; Keenan, Trevor F. et al. Australian vegetation phenology: new insights from satellite remote sensing and digital repeat photography. 2016. Biogeosciences vol. 13(17) p. 5085-5102 (link)
- Rungrat T, Awlia M, Brown TB, Cheng R, Sirault X, Fajkus J, Trtilek M, Furbank B, Badger M, Tester M, Pogson B, Borevitz JO, Wilson P. Using phenomic analysis of photosynthetic function for abiotic stress response gene discovery. The Arabidopsis Book 2016 (PDF)
- Brown, Tim B, Hultine, KR, Steltzer, H, Denny, E, Denslow, MW, Granados, J, Henderson, S, Moore, D, Nagai, S, SanClements, M, Sánchez-Azofeifa, A, Sonnentag, O, Tazik, D, Richardson, AD.(2016) Using phenocams to monitor our changing Earth: towards a global phenocam network. Frontiers in Ecology and the Environment. Vol 14, Issue 2 (March 2016). (PDF)
- Brown, Tim B., et al. TraitCapture: genomic and environment modelling of plant phenomic data. (2014). Current opinion in plant biology 18 (2014): 73-79. https://doi.org/10.1016/j.pbi.2014.02.002. PDF
- Nagler, P, Pearlstein, S, Glenn, EP, Brown, TB, et al. 2014. Rapid dispersal of salt cedar (Tamarix spp.) bio-control beetles (Diorhabda carinulata) on a desert river detected by phenocams, MODIS imagery and ground observations. Remote Sensing of Environment 140. 206–219. (PDF)
- Nagler, P. L., Brown, TB et al. 2012. Regional scale impacts of Tamarix leaf beetles (Diorhabda carinulata) on the water availability of western U.S. rivers as determined by multi-scale remote sensing methods. Remote Sensing of Environment, 118(0), 227-240. (PDF)
- Brown, Tim B, C. Zimmermann, W. Panneton, N. Noah, J. Borevitz. 2012. High-resolution, time-lapse imaging for ecosystem-scale phenotyping in the field. in: High Throughput Phenotyping in Plants. Methods in molecular biology. J. Normanly, ed. New York: Springer. (PDF)
- PL Nagler, Brown, TB, Hultine, KR, van Riper, C, Bean, DW, Murray, S, Pearlstein, S. 2010. Monitoring impacts of Tamarix leaf beetles (Diorhabda elongata) on the leaf phenology and water use of Tamarix spp. using ground and remote sensing methods. AGU Fall Meeting Abstracts 1, 0320. (PDF)
- PL Nagler, T Brown, PE Dennison, KR Hultine, EP Glenn. 2009. Using Webcam Technology for Measuring and Scaling Phenology of Tamarisk (Tamarix ramosissima) Infested with the Biocontrol Beetle (Diorhabda carinulata) on the Dolores River, Utah. AGU Fall Meeting Abstracts 1, 0389. (PDF)
- Brown, Timothy B. 2006. Biology and Modeling of Self-Organization in the New World army and Eciton burchellii. PhD Thesis. Dept. of Biology, University of Utah. 116 pp. (PDF)
From 2006 to 2012 I ran a technology consulting business prior to returning to work in academia at the ANU in June 2012. Below are the major projects I led while at TimeScience.
1 Alta Bark Beetle Gigapixel Imaging Project. 2011-2012. Designed and implemented protocols and monitoring plan to incorporate the use of gigapixel resolution imagery for early detection of bark beetle outbreaks at Alta Ski area. Gigapixel imagery augments traditional aerial surveys, allowing the ski area a low-cost meant to survey every tree in a ~800ha area at a resolution of about 1 pixel/10cm2. Project Gigapans: http://bit.ly/AltaBeetle2011). Presentation and project details (PDF).
2 Utah Museum of Natural History Interactive Construction Time lapse. 2012. TimeScience filmed the 3-year construction of the UMNH building. We then designed and built an interactive “TimeWindow” kiosk system for installation in the museum. The TimeWindow permits users to spin a dial to view the entire time lapse in seconds then zoom in to time-regions of interest to watch changes unfold on a minute or hourly time-scale.
Website: View interactive timelapse online
3 Utah Sky Exhibit. 2011. Natural History Museum of Utah. Utah Sky is an interactive exhibit that combines high resolution time-lapse imagery with live weather data to teach visitors about weather cycles in the Salt Lake Valley. I was the lead project manager for the TimeScience components in this exhibit and provided back-end programing for weather data integration.
4 Gigavision – A billion pixel resolution solar powered time-lapse camera for monitoring every plant in the landscape. 2010 to present. Lead project manager and hardware and software engineer; responsible for design of all system hardware and control software and field installations.
Website: http://www.gigavision.org. See also Brown, 2012.
5 TimeCam.TV. 2010. TimeCam is an online system for providing automated timelapse and dynamic movie playback for any online camera. Co-designer of all front-end and back-end software and server tools.
6 TimeGraph. 2010. TimeGraph is an online data visualization system for integrating time-series image data and weather sensor data into an interactive interface to enhance scientific collaboration, promote data sharing and provide “virtual lab” educational tools. Lead project manager. Co-designer of software and online interface. Example: ufsn-data.chpc.utah.edu/crfs.htm
7 Virgin River phenocam tower and wireless network. 2011. The Virgin River phenocam tower is an 11-meter tower in a remote location on the Virgin River in SE Nevada, USA for monitoring plant responses to an introduced bio-control herbivorous beetle. Installed monitoring systems include 3 RGB and 2 NIR imaging systems, automated PC data collection with remote data syncing via 3G wireless and a 250W solar power system. The project also included installation of a webserver and high capacity hard drive systems to support storage and online visualization of phenological data from all our Utah and Nevada desert research projects. I designed, built and installed all monitoring, solar and wireless systems and sourced and oversaw the tower installation. I also installed and configured the webserver and built the website to host the data.
Website: http://phenocam.org. References: See also pubs.
8 Rio Mesa field station phenocam tower system and wireless network. 2008. The Rio Mesa phenocam tower system consists of two 10-meter monitoring towers installed in a remote field site in SE Utah, USA. Both towers house multiple visible and infrared camera for monitoring plant responses to an introduced bio-control herbivorous beetle. I also designed and installed a solar powered wireless mesh network that provides wireless connectivity over the 120ha area of the field station. Lead project manager for camera and solar systems and wireless network installation and maintenance.
References: See pubs.
9 MealReader 2.0. 2008. Custom Matlab package for analyzing rodent feeding behavior.
References: 1) Torregrossa, A.-M., Azzara, A. V. and Dearing, M. D. (2011) Differential regulation of plant secondary compounds by herbivorous rodents. Functional Ecology, 25: 1232–1240. doi: 10.1111/j.1365-2435.2011.01896.x (PDF). 2) Torregrossa, A-M; Azzara, AV and Dearing, MD. (2012) Testing the diet breadth trade-off hypothesis: differential regulations of novel plant secondary compounds by a specialist and generalist herbivore. Oecologia, 168:711-718. (PDF).
10 TimeSystem. 2007 – 2009. TimeSystem is an advanced visualization tool for recording and analyzing time-series image sets and numeric data. Lead project manager; software designer.
Website: http://www.time-science.com/timescience/products_timesystem.asp. References: 1) Adam Nelson, Colson, K.E., Harmon, S. and Potts, W.K. 2013. Rapid adaptation to mammalian sociality via sexually selected traits. BMC Evolutionary Biology 2013, 13:81 doi:10.1186/1471-2148-13-81 . (PDF). 2) Adam Nelson, AC & Potts, W.K. (In prep). Rapid adaptation to sociality involves increased MUP expression in house mice.
11 MouseCam. 2007. A 9-camera monitoring system with integrated RFID monitoring for field analysis of mouse behavior and Hantavirus disease dynamics. Lead project manager and hardware designer. Software programmer. References: 1) Dearing, MD and Dizney, L. (2010) Ecology of hantavirus in a changing world. Annals of the New York Academy of Sciences 1195:99-112. (PDF)
12 Virtual Great Salt Lake Exhibit, Living Planet Aquarium, Salt Lake City Utah. 2006. Edited new and existing content into a 20-minute video for an interactive educational exhibit about the Great Salt Lake, Utah. Created month-long video timelapse of Great Salt Lake and provided technology consultation