The Glow Getters| Light the Path
Awards & Nominations
The Glow Getters has received the following awards and nominations. Way to go!
Light the Path
Light Graph
Summary
Many countries around the word have established restrictions on movement in response to the COVID-19 pandemic. Using night light data, Light Graph monitors how human networks across large geographical regions have changed during the pandemic, towards understand the effectiveness of measures used to slow the spread of the virus and to understand for the future how to effectively disrupt the propagation of the virus through these networks.
How We Addressed This Challenge
One key challenge for decision-makers is to understand how human activity in geographical networks has changed during the pandemic. Images are an incredibly useful tool, however they can often be open to interpretation.
Starting from nighttime data from the NASA Black Marble [1,2] suite of products (VNP46A1), Light Graph is a layer of quantitative analysis of the luminance networks using graph theory. This provides an objective simplification of the Black Marble data to a reduced set of numbers that can be more effectively monitor over time. Decision makers can then have a clearer understanding of what has happened and makes it easier to make evidence based future policies.
With the Black Marble data, these luminance networks can be monitored on a daily basis (allowing for interruption due to cloud cover) without the need to deploy any addition infrastructure.
Light Graph provides a graph representation of these networks and so allows the application of graph theory to monitor and manipulate these networks. For example the minimum number of edges that are needed to disrupt such a network to achieve a desired level of connectedness can be calculated analytically to make the most effective use of deployed government resources.
How We Developed This Project
We considered what expertise we have as a team. With a combination of data visualisation, full stack development, and mathematical modelling this challenge was particularly appealing. We found that this challenge had similar objectives to other challenges such as Quiet Planet, however we felt that Light The Path most aligned with our skill set. We had also considered the challenge SDGs and COVID-19 however after our initial research we felt that as a team of three with no expertise in international relations or economics, that challenge could not be effectively tackled by our team in 2 days.
Having identified the challenge we performed, we began an initial period of research into the available datasets such as NASA Black Marble [1,2] and what technologies we would want to use with this data. The scope of this challenge is big, and is reflected in our project presentation slides and on how we would like to develop this project further. However for the 2 day hackathon we decided to focus on a few major cities, London, Wuhan, and New York.
We each identified different areas that drew from our own backgrounds.
Using his experience of Data Science Alain identified ways in which data complementary to that available from NASA and NASA's partner agencies for the Space Apps COVID-19 Challenge could be visualised alongside processed Black Marble data.
Using his background in Physics Ed related the nighttime light data to the filamentary structure of galaxy cluster. Using existing Astrophysics tools, the filamentary structure in Black Marble data is extracted using the FilFinder Python package [3,4].
Using his experience of Full Stack Development, Ryan created a computational environment on Amazon Web Services to encapsulate the Black Marble data processing and made this accessible through a Rest API.
In the two day hackathon, we have been able to create a proof of concept for network analysis of Black Marble data. We have been able to prepare an interactive application that requests offline processed Black Marble data through the Rest API for the regions surrounding our three major cities of choice.
To be useful for decision-makers the quantity of offline datasets needs to be increased by orders of magnitude. The computational infrastructure also needs to be able to run the network analysis of the Black Marble luminance data in near real time. However the currently workflow runs on the order of minutes and not seconds. We believe that by developing filament detection algorithms tailored to Black Marble data this is achievable. With these changes areal-time web interfacecan be created that allows decision-makers to freely scrub a timeline while focused on a particular region and see the change in the graph representation of the luminance network of the region and its derived properties.
Following those steps, Deep Learning techniques can be considered on those graph representation . Image segmentation for instance,could be perform to identify the previous networks using U-Net variations . Also, if those images were used in combination with others datasets (other satellite data such as NO2, epidemiological , socio-economic) others domainscould be tackle using this set of images.
Project Demo
Link to slides presentation : https://github.com/Ricool06/TheGlowGetters/blob/master/light-graph-technical-presentation.pdf
Link to webApp : https://light-graph-story.glitch.me/
Project Code
Data & Resources
[1] Román, M.O., Wang, Z., Sun, Q., Kalb, V., Miller, S.D., Molthan, A., Schultz, L., Bell, J., Stokes, E.C., Pandey, B. and Seto, K.C., et al. (2018). NASA's Black Marble nighttime lights product suite. Remote Sensing of Environment 210, 113-143. doi:10.1016/j.rse.2018.03.017.
[2] Román, M.O., Wang, Z., Sun, Q., Kalb, V., Miller, S.D., Molthan, A., Schultz, L., Bell, J., Stokes, E.C., Pandey, B. and Seto, K.C., et al. (2018). VNP46A1 - VIIRS/NPP Daily Gridded Day Night Band 500m Linear Lat Lon Grid Night. ladsweb.modaps.eosdis.nasa.gov.VNP46A1.
[3] Koch, E. and Rosolowsky, E. (2015). Filament identification through mathematical morphology. Monthly Notices of the Royal Astronomical Society, 452, 3435-3450. doi:10.1093/mnras/stv1521.
[4] Koch, E. and Ginsburg, A. (2015, June 10). FilFinder: Mostly minor bug fixes (Version v1.1). Zenodo. doi:10.5281/zenodo.18463.
[5] Public Health England. Coronavirus (COVID-19) Cases
https://data.london.gov.uk/dataset/coronavirus--covid-19--cases
[6] NYC Open Data. COVID-19 Daily Counts of Cases, Hospitalizations, and Deaths
https://data.cityofnewyork.us/Health/COVID-19-Daily-Counts-of-Cases-Hospitalizations-an/rc75-m7u3