It’s a point of view that has brought astronauts to tears. Earth seen from far above is a perspective that gives things real perspective.
Images taken from this angle have also proven an incredibly powerful, especially when considered over time. Satellite imagery – ‘the world’s selfie’ as some in the field call it – has been used to track deforestation, dying lakes, forest fires and cracks in the ice caps.
Only last month, satellite pictures of the Abbot Point Coal Terminal in Queensland before and after Cyclone Debbie appeared to show waterways contaminated by coal dust, sparking an investigation by the state’s Department of Environment and Heritage Protection.
Such rich data has long only been accessible to governments and science organisations – mainly due to its sheer size and complexity – but there are growing efforts to make the resource more readily available the general public. Leading the charge is an Australian developer whose weekend pet project has opened the data to anyone with an iPhone.
Every 16 days, a satellite the size of a minibus called Landsat 8 passes over you at 705 kilometres above ground level and takes a photo, capturing visible, infrared, near-infrared, and thermal-infrared light bands. The NASA and US Geological Survey satellite splits all of the earth’s land mass into ‘scenes’, captures 700 scenes per day and shoots the whole world in little over a fortnight.
“It’s a bit like having one of those Wi-Fi connected security cameras in your backyard except you’re catching the entire planet,” explains Lachlan Hurst, who launched his app ObservedEarth last year and introduced it at the AWS Summit in Sydney earlier this month.
As anyone who has spent time with Landsat data will admit, the views up here are stunning.
ObservedEarth shows the changing water level at Eildon Dam, Victoria.
Source of imagery: US Geological Survey
“Many 'development' hours were spent viewing interesting parts of the world; and really when viewed from space most parts of the world are interesting,” Hurst says.
“Part of my motivation towards making this app, and making it accessible, was so people could download and play with the data themselves. We’ve got this humungous dataset there that’s begging to be used.”
Landsat satellites have been whizzing above the atmosphere since 1972. Since late 2008, Landsat data has been available free of charge. However, the data really opened up at the beginning of 2015, when Amazon Web Services committed to make up to a petabyte of Landsat earth imagery data widely available as an AWS Public Data Set, available for anyone to access via Amazon S3.
Each light band of each Landsat scene (of which there are more than a million as of February) is available as a stand-alone GeoTIFF (a TIFF file with geo-referencing information like coordinates embedded) and the scene’s metadata is hosted as a text file. But picture of the earth, even a small slice of it, is big.
“While a desktop user may be happy to launch an hour long background process to download multiple gigabytes worth of Landsat data, mobile users would likely not,” Hurst, a senior engineer at geospatial computing firm VPAC Innovations, says.
The trick to making this ‘little app for a LOT of data’ Hurst says: “Download just the data that the user needs”.
Within the free app, a user can scroll around the world to pick a scene of interest. Within a scene, if a user zooms to see the high resolution, 30m view of a region, only the GeoTIFF data for that region is downloaded.
“What this means is that for a small region of interest only a few hundred kilobytes will be downloaded instead of the entire 50MB GeoTIFF per band,” the Melbourne-based developer adds.
Spin the globe
“When I was much much younger my grandparents used to have one of those big globes. And I used to like spinning it round and you’d stick your finger down on it and it would land somewhere – ‘hey that’s Uzbekistan’ – that’s great. That’s the sort of user experience I was going for here. Something that did justice to the data I guess,” says Hurst.
Part of that UX comes from the use of Instagram-style filters. The filters combine data from different light bands to indicate vegetation, urban areas or water. There are other features like a cloud-coverage threshold filter.
Coal mining in and around Leard State Forest, NSW.
Source of imagery: U.S. Geological Survey
“Filters are a nice way of embedding domain expertise into your application in a nice, user-friendly way. A filter on Instagram is actually quite a sophisticated image processing algorithm, which you can apply to your photos by just pressing that button. ObservedEarth has filters too, but we’ve also got better data to play with,” Hurst says.
The app is built on a number of Apple specific technologies and written in the Swift programming language, with some C code. The user interface makes use of SceneKit – Apple’s 3D graphics framework – in combination with Metal, a lower-level API aimed at squeezing maximum performance from the GPU.
Since its release in August last year, Hurst’s app – built over six months’ worth of weekends – has been used to view 200,000 scenes, download 400GB of Landsat data, and was recently given the nod of approval by NASA.
He hopes more people will dip into Landsat’s rich data, and plans to add the 15 years’ worth of image data from Landsat 8’s predecessor, Landsat 7.
“We think of Australia as this big dry country with a big desert in the middle. Maybe it’s a bit boring in there. Not when you’re looking from space,” Hurst says. “Landsat data for Australia is actually really good because we’re a largely cloud free country. You can find lots of interesting things all over.”