Print PageEmail Page

    In the Heat of the Night

    AstroEcology slide

    What do you call a venture between ecologists and astronomers to combine their expertise to tackle the global challenge of species conservation? Astro-Ecology, reflecting a recent cross-discipline effort to offset species extinction.

    On Wednesday at SPIE Astronomical Telescopes + Instrumentation, Ph.D. student Maisie Rashman, of Liverpool John Moores University, introduced her team's groundbreaking work in a presentation titled, "Addressing environmental and atmospheric challenges for capturing high-precision thermal infrared data in the field of Astro-Ecology."

    "We use unmanned aerial vehicles and thermal-infrared cameras to survey animals to monitor population densities more effectively," Rashman said. "Monitoring population densities of animals has been flagged as a key strategy to reducing the mass extinction rate that's happening right now.

    Astro-Ecology was born, Rashman explained, from the similarities between the two fields.

    In Astronomy imaging, she pointed out:
    • Sources are bright objects on a dark background
    • Background temperature is lower than that of sources
    • Sources are mostly static, point sources of standard or fixed shape
    • Background actually foreground
    • Observing platform is stable
    • Sources have a predictable path

    Animals at night

    Animals that use metabolic heat to maintain a stable internal temperature, generally appear to be bright objects on a mostly unclear, dark background. Based on that premise, Astro-Ecology works with these key simple assumptions:
    • In the Thermal-IR regime, animals are bright objects on a dark background
    • Background temperature is lower than that of sources
    • From a height, animals are point sources of standard or fix shape
    • Animals can be tracked using adjacent image comparison.

    Before the Astro-Ecology approach, conservation groups would spend millions of dollars to hire a helicopter or light aircraft, place a bunch of ecologists on board, and have them manually count animals from the air. This method was inefficient and expensive - prohibitively so, in some cases.

    The current method being developed to survey animal species is straightforward and low-cost: buy a Thermal camera, attach it to a drone with a 25-minute flight time, collect high volumes of data, and condense that data to .jpg's with the goal of seeing the contrast between the animal and the environment.

    While less expensive and easier, this approach still has some challenges to overcome. Combing through 45,000 images manually is a long and inefficient process; poaching and habitat destruction - two other areas this data-collection is used for -- are very much time-sensitive issues. The longer time it takes you to analyze your data, the less impact you're going to have.

    Using the collective expertise and experience of astronomers dealing with large data sets and identifying objects within them Rashman's team felt confident they could join forces with ecologists and help them improve on their method to detect and identify different animals - and poachers - autonomously, anywhere in the world by using:
    1. Data collection (UAV and camera)
    2. Source detection and identification
    3. Source tracking
    4. Data analysis

    An initial test of this approach used a static drone in a zoo in northern England: ambient temperatures were cold, so it was easy to detect the two animals and their heat. A second test in Tanzania utilized a drone flying at 100 meters in the morning, afternoon, and evening. Environmental challenges including rocks, tree branches, high albedo surfaces, and warm ground worked to reduce the efficacy of this system.

    The assumptions vs the reality, as Rashman noted, include:
    • In the Thermal-IR regime, animals are not the only bright objects on a dark background
    • Background temperature is often comparable to animal temperatures
    • The shape and temperature of animals depends on the viewing angle
    • Moving sources coupled with a moving observing platform make tracking an unsolved computer vision problem.

    Animals as viewed with infrared imaging

    Ultimately, she said, "We need an intelligent system that can tell a rock from an animal. We started by applying simple physics and astronomy data reduction pipelines to reduce the data bottleneck occurring in ecology. "We are taking inspiration from CCD and CMOS characterization to reduce systematics in ‘off the shelf' systems."

    While they have identified that many astronomy techniques are not as analogous as initially believed and assumptions were too simple, it's inspiring to see two distinct scientific fields, coupling their technologies and techniques in an active attempt to address global challenges. The interdisciplinary nature of optics and photonics are once again solving problems too complicated for one field on its own.