The trickle-down effect of enhancing satellite technology
April 17, 2019
New satellites can measure the earth’s surface on the scale of centimetres, making remote sensing more powerful
Share
Like millions of people around the world, I have an abiding passion for space. Fortunately, my job brings me closer to that famous final frontier than most. As a remote sensing specialist, I work to analyze images captured far above Earth’s surface to give our clients a deeper understanding of our planet and how it changes over time.
Remote sensing helps us see the big picture and make more informed decisions about almost anything related to the surface of the earth, from the depths of the ocean to the heights of mountains.
Much of the imagery used in remote sensing is captured by satellites launched by both governments and private businesses. Drones and planes are frequently used as well. No matter how many times I do it, looking at images taken from orbit gives me a thrill. Sometimes it takes me back to my time studying the rain forests of Borneo, a massive island set off the northwest coast of Indonesia.
From the forests of Borneo to the Canadian plains—and all places between—remote sensing helps us see the big picture and make more informed decisions about almost anything related to the surface of the earth.
Tracking rattan in Southeast Asia
My graduate research project location was in Indonesia, where much of the farmland is of relatively poor quality. Cut from the rain forest, the farmland there can only support crops for a few years before it needs to be returned to the jungle. One of the earliest signs of the jungle reclaiming a barren field is the growth of rattan—a type of palm reed extensively used for wicker baskets and furniture.
For planning purposes, the Indonesian government wanted to accurately—and affordably—track rattan growth in their territory. This was easier said than done as Indonesia is made up of more than 17,000 islands. In terms of area, it’s about the same size as Alberta, Saskatchewan, and Manitoba combined. Taking rattan inventories on foot was out of the question. Doing it by plane was possible, but it would cost a lot and still leave them at the mercy of cloud cover.
Enter radar satellite imagery, with its unique ability to penetrate cloud cover.
Remote sensing in near infrared imaging as seen on the Manitoba Hydro, Bipole III project.
The advancements in remote sensing
Thanks to radar satellite imagery, we were able to accurately track the growth of rattan across the island chain by analyzing data taken from space. But that was in the early 2000s—before the advent of smartphones.
Newer satellites are equipped with finer resolution cameras that can, in some cases, measure the earth’s surface on the scale of centimetres.
In the years since, remote sensing has changed as much as your cellphone or your laptop—maybe even more. The most obvious change is the number of satellites taking pictures. Plus, newer satellites are equipped with finer resolution cameras that can, in some cases, measure the earth’s surface on the scale of centimetres.
This growth in the number of satellites has resulted in an explosion in the number of images available for any given project. Now, we can pull so many images together that finding an efficient way to analyze all of them is a major challenge—a far cry from the earliest days of remote sensing in the mid-20th Century.
An introduction to Object-Based Image Analysis
The latest and greatest solution to this problem of abundance is a method called Object-Based Image Analysis (OBIA). OBIA applies machine learning to high-resolution satellite images. Why? To find patterns of pixels that represent object classes—for instance, rattan.
Once the system has been taught what patterns to look for, it can scope out and analyze huge image sets almost instantly. This makes discovering what our clients need to know faster, cheaper, and safer—whether they are assessing the integrity of a dam, scouring a pipeline for leaks, measuring drainage in urban areas, or inspecting a wetland for ecological change.
OBIA has been a game-changing tool for remote sensing. But the biggest changes in the field might yet be ahead.
OBIA applies machine learning to high-resolution satellite images. Why? To find patterns of pixels that represent object classes, like on the Barrow Island Vegetation Recognition project in Australia.
The future of remote sensing
In recent years, private companies have been implementing a new technology called dove satellites. Much smaller than conventional satellites, dove satellites are launched in flocks and positioned very close to each other in orbit. This approach significantly boosts the rate at which satellite images can be updated.
Already, these dove satellites can take a complete picture of the earth once per day at a resolution of 3 metres per pixel. As that interval continues to fall and the resolution increases, our ability to track changes on the surface of the earth will improve. And when our remote sensing capabilities improve, our ability to make informed decisions for our clients will only grow.