Image: Pixabay
Since 2017, there have been media reports that China has been sending Muslims in large numbers to re-education camps in Xinjiang. These were largely based on interviews with former inmates. As international pressure began to build, China went about deleting related documents and electronic traces such as social media posts. In a country as closely guarded as China, the story would have been easily suppressed, if it were not for satellite imagery.
Two investigative stories broke simultaneously at the end of October 2018 — by the BBC and the Australian network ABC — giving not only the exact location of the camps, but also satellite images detailing their physical existence. In both cases, the investigative teams painstakingly worked with citizen rights activists and imagery analysts to pour through years-worth of images to tell the world about the horror unfolding in Xinjiang.
Let’s take the BBC investigation. The BBC sought help of GMV — a company specializing in earth observation data — to look into a list of 101 suspected facilities in Xinjiang. They measured the growth of new sites and the expansion of the existing ones, and identified and compared common features such as watchtowers and security fencing. GMV placed 44 of them in high or very high likelihood category of being prisons.
The BBC took one of GMV’s findings — the increase in building size at the facility in Dabancheng — and consulted an Australian company with experience in prison design. Using the measurements from the satellite images, they calculated that at an absolute minimum, the facility could provide space for about 11,000 detainees. Even that minimum estimate would place it alongside some of the biggest prisons in the world. And that was just one site. There was a very high probability of 44 sites being prison camps.
A set of imagery from 2015 to 2018 shows how China's re-education camps in Xinjiang have grown over the years. Courtesy: BBC/Google Earth/ESA
A set of imagery from 2015 to 2018 shows how China's re-education camps in Xinjiang have grown over the years. Courtesy: BBC/Google Earth/ESA
A set of imagery from 2015 to 2018 shows how China's re-education camps in Xinjiang have grown over the years. Courtesy: BBC/Google Earth/ESA
That was in late 2018. This year, The Xinjiang Data Project by the Australian Strategic Policy Institute (ASPI), a Canberra-based think tank, again used satellite imagery as evidence to show detention facility construction across 2019 and 2020. The researchers identified and mapped over 380 sites across Xinjiang that were newly built or expanded since 2017, with at least 61 sites seeing construction activity between July 2019 and July 2020. This includes at least 14 facilities still under construction, according to latest imagery available.
The Xinjiang Data Project presents satellite imagery evidence to show newly built detention facilities between 2019 and 2020.
ASPI also used the imagery to reconstruct 3D models of each tier of detention facilities — from lowest security (tier 1) to highest security (tier 4) — showing the key structural features.
Making Headlines
Satellite imagery is certainly grabbing headlines as it becomes an indispensable tool for journalists. Be it for fact-finding or gauging the impact of a significant development, reporting on climate events or conflict zones, because of the unbiased and continuous insights it provides, satellite imagery is being extensively used in newsrooms today.
September 14, 2019 image almost immediately after the coordinated drone attacks on multiple oil facilities in Saudi Arabia. Screenshot
For instance, Planet came out with an image almost immediately after the coordinated drone attacks on multiple oil facilities in Saudi Arabia on September 14 last year. This is possibly the first time a satellite photo was used as the lead image on page 1 of The New York Times.
But then the imagery is not just for stunning visual effects — it is also for giving great insights. In the Saudi attack case, the imagery provided immediate and exact details of the damage caused to the oil facilities. Likewise, it can provide insights into almost all spheres of life on Earth, especially the things that the human eye can’t see.
Monitoring Disasters
California, Oregon, and Washington state in the United States have endured a fire season of historic proportions. Wildfires this year have left dozens of people dead, destroyed thousands of structures, and scorched millions of acres, causing extreme air pollution that has threatened the health of millions of residents and sparking fears of lasting damage on species and ecosystems. Satellite imagery dated September 9 shows the damage caused by the wildfire in Oregon.
Close-up of neighborhoods in Phoenix, Oregon, shows widespread devastation from the wildfires. Courtesy: Maxar Technologies
The August Complex Fire in California, which began on August 17, was still burning months later. One of the instruments on NASA’s Terra satellite is the Multi-angle Imaging Spectroradiometer, which has nine different cameras pointing towards Earth at different angles. As Terra passed over the August Complex Fire on August 31, the sensor collected snapshots of the smoke plume from different angles. The instrument also collected information about the amount, size, and brightness of the particles within the smoke plume based on how the particles scatter light at different angles and wavelengths.
This data gives researchers information about the characteristics of the wildfire smoke to predict how it will move and affect air quality. For example, the southern part of the smoke plume emitted by the August Complex Fire on August 31 was made of mostly small, dark particles usually released when a fire is burning intensely. But as the plume moved downwind, the particles became larger and brighter, possibly because water or other gases emitted by the fires condensed on the smoke particles.
On August 31, a satellite detected several hotspots in the August Complex Fire in California, as well as several other actively burning areas to the north, west, and south. Courtesy: NASA
Fires around the Arctic Circle have been burning nonstop all summer this year. Siberia’s fire season, which began in April, has broken records with some places in the region recording over 30 degrees Celsius. Again, NASA’s Terra satellite captured this image of fires wreaking havoc across Siberia on July 21, 2020.
Satellite image of fires wreaking havoc across Siberia on July 21, 2020. Courtesy: NASA
Change Detection
In the past three decades, several of Iraq’s largest lakes have experienced fluctuating water levels due to droughts, dam management, and continuing unrest and civil wars. Two lakes in particular — Milh and Habbaniyah — experienced substantial declines in water levels. Now, satellite imagery shows the two lakes are refilling.
These natural-color images show Lake Milh on August 16, 2018 (red/orange) compared to August 21, 2020 (green). The red color is most likely due to the presence of algae and bacteria with red pigments. The images were captured by NASA’s Landsat 8. Older imagery suggests that the lake has not appeared this full since 2009.
The images were captured by the Operational Land Imager (OLI) on Landsat 8. Courtesy: NASA
Breaking News
On June 22, 2019, a volcano on one of the uninhabited islands that connects Japan to Russia suddenly sprang to life. It was dormant and inactive for nearly a century. The unexpected volcanic activity was first discovered by astronauts on the International Space Station.
Raikoke Volcano in eastern Asia on June 22, 2019. Image captured from International Space Station
Access to the Inaccessible
In disaster zones where immediate human visit is impossible, satellite imagery can give insight into the damages or warn for worse to come. Satellite imagery showed the immediate aftermath of August 4 deadly explosion at the port of Beirut which reportedly left around 300,000 people homeless.
In one best case example of access to the inaccessible, Maxar (then DigitalGlobe) took the picture of Fukushima nuclear plant complex, three minutes after the explosion at the third reactor, on March 11, 2011. This is a classic example of how fast satellites can give information from inaccessible places.
Steam bellows out of Fukushima reactor complex, roughly three minutes after the explosion on March 11, 2011. Courtesy: DigitalGlobe (Maxar Technologies).
Conflict Zone Reporting
Satellite imagery has become an effective tool for reporting in conflict zones. A recent investigation by the Guardian used imagery with flight data to discover that Turkey and the United Arab Emirates are using large-scale military cargo planes to funnel in goods and fighters to forces inside Libya, in blatant violation of the UN arms embargo, fueling a proxy war and making political solutions difficult.
In addition to broad investigations, one can focus on particular incidents for daily reporting. For instance, India and China were engaged in a violent conflict on the Himalayan border in June. Satellite images by Planet taken less than 24 hours after the face-off on the night of June 15 show a massive build-up by the Chinese side despite a de-escalation agreement on June 6.
HawkEye 360 shows major build-up of Chinese military in the Galwan Valley, on the Indo-China border. Courtesy: HawkEye 360/Planet
While these were the first visual details of the location, a few days later, a unique technology from HawkEye 360 combining space-based radio frequency analytics on top of optical imagery substantiated the Indian claim of Chinese build-up on May 22. By overlaying its analysis on Planet’s SkySat imagery, HawkEye 360 was able to reveal Chinese build-up, including trucks, tents, and equipment along the valley.
A set of images dated June 22 from Maxar further shows freshly built structures — not visible in earlier photographs — by China overlooking the Galwan.
Imagery dated June 22, 2020, shows freshly built structures by China. Courtesy: Maxar Technologies
Crime and Smuggling
The Royal United Services Institute (RUSI) — the UK’s leading defense and security think tank — exposed North Korean illicit shipping networks using satellite data, open source data-mining, and data-fusion techniques. In its latest report, published on March 5, 2020, RUSI used imagery collected over North Korean ports and islands around Zhoushan to confirm that the vessels were all DPRK flagged and were loading coal and other resources, before sailing into Chinese waters, in violation of the UN resolutions.
This appears to be a large-scale, coordinated effort with China to evade sanctions imposed on North Korea over its nuclear weapons and ballistic missile programs.
Planet imagery shows cluster of North Korean ships in Taean with some vessels loading coal, October 3, 2019. Courtesy: RUSI/Planet
Interestingly, in March 2020, RUSI reported an unprecedented number of vessels at North Korea’s main ports, in what could be interpreted as Pyongyang ordering its ships back home in the aftermath of the COVID-19 outbreak.
North Korean ships heading back home in March 2020 with the onset of the pandemic. Courtesy: ESA/RUSI
Tracking COVID-19
As COVID-19 brought the entire world to a standstill early this year, the role of satellite data in assessing the industrial and environmental impacts of the pandemic has been paramount. It has become particularly handy for journalists in the past few months, as travel restrictions and safety issues are making on-the-ground reporting difficult.
Pollution
Satellite data of multiple imagery providers across geographies showed significant reduction in pollution during the lockdown period in March-April compared to the same period last year.
These satellite maps, produced using data from the Copernicus Sentinel-5P satellite, show average nitrogen dioxide concentrations over India from January 1 to March 24, 2020 and March 25 (the first day of the lockdown) to April 20, 2020 — compared to the same time-frame as last year. Screenshot
Supply Chain and Manufacturing
Images from Planet show the difference in production capacity at the Volkswagen plant in Tianjin, China. Manufactured cars fill the space on the left in May 2019, while the March 2020 image is only half full, as production was stopped. The pandemic impacted manufacturing worldwide, given the globalized state of modern supply chains. One can get into visual comparison of imagery taken over different time periods to get an indication of the level of the impact expected in terms of the drop in production.
At this Volkswagen plant in Tianjin, China, manufactured cars fill up the space on the left on May 1, 2019. It is only half full in the March 3, 2020, image. Courtesy: Planet/PwC
Air and Marine Traffic
Aviation has been one of the most impacted industries due to border restrictions and stringent lockdown measures imposed at the onset of the pandemic. Marine traffic, too, took a major beating as operations at most major ports around the world came to a standstill. With both business and leisure travel impacted, satellite data showed the busiest ports and airports around the world looking like ghost towns at the height of the lockdowns. As more than 80% of global trade by volume is transported by sea, one can imagine the impact on the global supply chains.
Los Angeles International Airport, the third busiest in the world by passenger count, on March 17, 2020. Courtesy: Maxar Technologies
The Port of Long Beach, California — the second-busiest container port in the US — on February 8, 2020. Courtesy: Maxar Technologies
Impact on Economy
Data from satellites can support in quantifying the impact this could have not only on the shipping industry, but also by extension on the global economy as a whole. In this satellite data analysis by MarineTraffic in April, red dots represent fully loaded oil tanker ships stranded at sea on April 22 due to the lack of global demand. With ports out of operation, ships were stranded on the high seas. And with production plants shut across the world, there were no takers for oil. On April 20, 2020, prices of US crude oil futures turned negative for the first time ever, at one point plunging roughly 300%.
The red markers are fully loaded oil tankers stranded at sea in April due to the lack of global demand for oil. Courtesy: MarineTraffic
Investigative Journalism
Other than the imagery becoming the story, often the imagery can help establish correlations to arrive at a larger conclusion. You may have some information, but you need confirmation, tying up loose ends, and clues for what, where, and when.
A Harvard research article combined satellite imagery of hospital parking lots and Baidu search queries of disease-related terms to find that the coronavirus may have been spreading in China as early as August 2019, and not December. The research is not yet peer-reviewed, and Beijing has, of course, denied the allegations.
The team analyzed commercial satellite imagery and “observed a dramatic increase in hospital traffic outside five major Wuhan health facilities beginning late summer and early fall 2019,” according to Dr. John Brownstein, the Harvard professor who led the research.
The Harvard researchers used satellite imagery of hospital parking lots and Baidu search queries of disease-related terms to investigate. Using techniques similar to those employed by intelligence agencies, they observed an upward trend in hospital traffic and search volume in the latter half of 2019. Screenshot
The researchers collected 111 satellite images of Wuhan (multiple sites per image) from January 9, 2018, to April 30, 2020, resulting in 140 successful daily extractions of parking lot volume from hospitals, and 117 from the three high-volume control sites. They observed that between 2018 and 2020, there was a general upward trend of increased hospital occupancy as measured by the parking lot volume.
Individual hospitals had days of high relative volume in both fall and winter 2019. However, between September and October 2019, five of the six hospitals showed the highest relative daily volume of the analyzed series. This coincided with the elevated levels of Baidu search queries for the terms “diarrhea” and “cough.” Both search query terms show a large increase approximately three weeks preceding the spike in confirmed COVID-19 cases in early 2020.
“While we cannot confirm if the increased volume was directly related to the new virus, our evidence supports other recent work showing that emergence happened before identification at the Huanan Seafood market,” the study says.
Below is one of the most famous examples of use of satellites for investigative journalism. The Pulitzer Prize-winning story of the Associated Press which exposed a human trafficking ring in 2015 was tracked with DigitalGlobe (now Maxar) cameras. After months of going around in circles chasing a trail of slave trafficking in and around Indonesia and Thailand, AP requested DigitalGlobe train its best-in-class satellite WorldView 3 on a designated area in the deep sea. The satellite caught two trawlers loading slave-caught seafood onto a commercial cargo ship. More than 2,000 enslaved men were freed as a result of this investigation.
DigitalGlobe provided this image to the Associated Press. It showed two trawlers loading slave-caught seafood onto a commercial cargo ship. Courtesy: Maxar Technologies
Free Sources of Satellite Data
Despite the huge costs associated with putting each satellite in place and operating them, there are several sources of data that don’t cost anything.
- NASA’s Landsat imagery is freely available on the USGS website, and has the longest continuous global record of the Earth’s surface.
- The European Space Agency’s Copernicus Data and Information Services are available through the DIAS or the Conventional Data Hubs. It is currently the largest space data provider in the world on a free, full, and open access basis.
- Google Earth gives seamless availability of data, but not all of it great resolution, and there can be stitching errors.
- Commercial satellite imagery providers, like Maxar, Planet, Airbus, etc., give some data for free, especially at times of disasters or for humanitarian causes.
Many journalists still think that imagery is for just visualization — the last stage while presenting a story — when it could actually be the primary data. But then, it is always prudent to remember that imagery is just one form of data, and not the story itself. Every image needs to be backed up with traditional reporting methods and extensive research. Reporting on disasters or change detection stories is simpler. Investigative stories may take weeks or months to develop, and not all investigations are fruitful.
Additional Reading
Resources for Finding and Using Satellite Images
When a Picture Tells the Story: 9 Investigations That Used Satellite Imagery
This article was originally published on GW Prime, a subscription-based platform from Geospatial Media. It is republished here with permission.
Anusuya Datta is a writer, editor, and commentator who focuses on the intersection of technology innovations and their applications. She is the technology and innovation editor-at-large for Geospatial Media and Communications, where she previously worked as executive editor. She is based in Canada.
