When the sun goes down and the lights come on – or not – a wealth of data can be collected by satellite from the night sky, giving insight into the dynamic human activities taking place on the surface.
With remote sensing, things like land use change, urban development and forest management can be reliably and accurately measured in daylight. At night, we can collect different types of data. One way to do this is to use NASA’s Black Marble, a product suite that scans the sky every night and is powerful enough to detect all kinds of lights on Earth’s surface, from Christmas lights to a single 12,000 lumen flashlight, from space.
However, data collected at night can be difficult to analyze, says Zhe Zhu, an assistant professor in the Department of Natural Resources and Environment and director of the Global Environmental Remote Sensing Laboratory (GERS). He explains that nighttime satellite data can be influenced by many factors, resulting in a high degree of temporal variation, even for well-calibrated NASA black marble data.
“The first time I looked at a time series of data, I felt like it was almost impossible to use,” Zhu says.
A team of GERS researchers including Zhu and NRE Ph.D. student Tian Li, in collaboration with researchers from NASA and the University of Maryland’s Interdisciplinary Earth System Science Center, developed a method capable of provide daily maps of night light changes at medium resolution. The results are published in the journal Remote Sensing of Environment.
Although Black Marble already reduces outside noise a little, Zhu and Li looked for a way to further reduce fluctuations and further improve the data.
Many factors cause the signal that reaches the sensors to vary over time. Zhu gives the example of moonlight, which impacts the amount of illumination sensors receive throughout the phases of the moon and movement in the night sky. Some objects can distort the moonlight reflection, such as buildings, trees, or other structures. Atmospheric conditions can cause other signal deviations, along with things like aerosols or haze.
Zhu says they realized daily fluctuations could also be caused by the direction of view and the source of artificial lighting; for example, if the satellite is looking directly over a tall building, it may not see the light emission through the windows, or in more rural areas there may be trees obstructing the view to the sides .
To overcome some of these hurdles, Zhu says three-dimensional numerical models could be incorporated to account for the height of buildings or if there are trees around to predict the influence of these lighting differences, but it would be almost impossible to take everything into account, such as the height of each tree or building. Instead, Li found a different method, incorporating multiple angles of view for individual pixels, organizing them into groups, and then basing observations somewhere in between.
“Tian spent a few years on this, and she found a way to reduce this variation in the data significantly,” Zhu says. “We created an algorithm called VZA-COLD which stands for Viewing Zenith Angle stratified CONtinuous monitoring of Land Disturbance.”
“We applied different sets of models to describe the entire time series of night light data and subsets of data from different viewing zenith angle intervals. In this way, we smoothed the direction viewing angle and local geometry variations without the need for three-dimensional data.A change will be captured if it is predicted by one of the viewing angle interval models, says Li.
With VZA-COLD, researchers were able to use Black Marble data to continuously track human activity and behavioral changes, on a global scale. Zhu says this greatly reduced signal variation and within each group they were able to detect changes with high accuracy and in a timely manner.
Researchers can get time of change, magnitude of change, and direction of change. Zhu says this is important because at one viewing angle the change may not be visible, but is visible at another range of viewing angles.
“This makes the information useful, solves the temporal variation problem, and provides additional insight into angular changes for artificial lights.”
In the past, researchers typically analyzed a month’s worth of data to create composite data sets. This analysis is limited and cannot provide accurate detection of short-term nighttime light changes. For example, with recent hurricanes, Black Marble can now monitor nighttime light changes as they occur, quickly identifying power outages shortly after the storm has passed.
Other applications range from monitoring wars, such as those in Ukraine and Syria, where near real-time information could aid humanitarian efforts.
Tracking urbanization processes, migration, power grid variations, gas flares, illegal fishing operations, wildfires, and even the transition to LED lights are other applications this technology has. can be used to monitor. The possibilities are vast, as Zhu points out, because human activities are so intertwined with how we use energy, and it will inevitably show up in nighttime lights measured by satellite.
Zhu says the change in nighttime light can provide different information about people’s activity and behavior, such as public gatherings and holidays, than daytime observations. One example he cites is the Olympic Stadium in Seoul, South Korea.
“We can see when the building started being built and when the roof was finished, and then after the Olympics it’s used less,” he says. “This is an example of transitions in the stages of build activity and use activity that we are not able to monitor during the day. I think it’s going to be a game-changer for a lot of things.
“We are excited to see the patterns of all these different types of man-made light changes at night over the past decade, it gives us unique opportunities to understand the human-environment system across the world,” Li said.
“We are currently working with NASA to create a global product using this algorithm which will be released in 2023. It will be updated annually for every location on Earth’s land surface. There are also plans to integrate the near real-time capability into the global product,” says Zhu.
This work was supported by the NASA Terra, Aqua, Suomi-NPP, and NOAA-20 Program Grant 80NSSC22K0199NASA Earth Science Remote Sensing Theory Program Grant 80NSSC20K1748and the Office of the Director of National Intelligence (ODNI), Intelligence Advanced Research Projects Activity (IARPA), via 2021-2011000005.