Hundreds of researchers use nighttime lights as a measure of economic activity. While NOAA and NASA release the data, there is a lack of open-source implementations and a lack of efficient computing frameworks through which this data can be utilised at modest levels of computational budgets. In our open-source Julia package, we have pushed the frontier on statistical methods for cleaning this data, making these techniques, alongside traditional methods, available to researchers.
The package, NighttimeLights.jl, was a foundation for a research paper, "But clouds got in my way: bias and bias correction of VIIRS nighttime lights data in the presence of clouds" by Ayush Patnaik, Ajay Shah, Anshul Tayal, Susan Thomas. This paper diagnoses a source of bias in the data and responds to this problem with a bias correction scheme. Along with other mainstream methods of data cleaning, this method is also implemented in the package.
While there are packages to do image processing in Julia and packages to handle geostationary data, the assumptions about the sensor producing the data, such as in Images.jl, make it incompatible with nighttime lights data. We built the package from scratch without making any assumptions about the sensor. Functions in the package take regular float 3D arrays as input, which makes it possible to extend the package to data from any sensor and not just VIIRS nighttime lights.
In summary, our work is at the global frontier of exploiting nighttime lights data, and the first open-source implementation of such work, and harnesses Julia to make the large scale computation uniquely accessible within modest computational budgets.