2021-04-22, 20:30–21:00, Track 1
In this talk, we will first discuss the process of analysing GeoTIFF images of surface lights on Earth from space using multicore processing tools on AWS. Second, we will discuss how the data can then be used to predict GDP and other economic metrics, especially during supply-demand shocks like COVID-19.
How can we tell how COVID-19 has affected our economies ? Can we rely on official estimates ? Can we estimate the impact ourselves - with data from space - using satellite images available anyone in real-time using just a computer and a connection to the internet ?
With offices closed, data scarce and reliability of published results in question, the usual tools of obtaining such information is limited. It is in this backdrop that we explore using images of Earth taken at night-time from space. These images are available as GeoTIFF files with precise lon/lat co-ordinates at every pixel. Each pixel in turn contains radiance values that can reflect the economic activity on the surface. Such datasets have been shown to have correlation as high as 99% with actual GDP. Combined with data from power grids, the predictive power can be unparalleled. In this talk, I'll walk through examples of a) how such GeoTIFF files can be processed at scale using standard tools like GNU Parallel, Python Multiprocessing and how the resulting data can be then analysed by assigning radiance values with state/national boundaries and creating models that track changes over time.