Data from space can provide near real-time answers to the questions forecasters are asking. For example, the graph above is from Wuhan, China, where the pandemic began. Note how activity increases for the holiday season, then drops off as coronavirus impacts the area. As the lockdown is lifted, there is an initial surge, followed by sustained high levels of activity in the first week of June.
At SpaceKnow, we have been producing independent near real-time indicators of economic activity since our inception in 2014. We extract and aggregate changes from satellite images with the help of machine learning, advanced statistics, and industry expertise.
To break it down, we monitor tens of thousands of locations globally using a variety of space-based sensors. Our cutting-edge artificial intelligence technology analyzes billions of pixels to detect changes at these locations. We focus on detecting changes related to output, product movement, construction, factories, oil wells, ports, mines, and more.
Creating activity indicators from this massive database is not an easy task. The quality of satellite imagery is dependent on environmental and imaging conditions. Different satellites have different sensors (optical, multispectral, SAR), revisit rates, and orbital patterns. Our intellectual property is the ability to process all this information and translate it into indices. The process requires close collaboration between our team of data annotators, scientists, data scientists, and economists. Over the past 6 years, via experience and experimentation, we and our machines have learned and improved substantially.