The longer human activities are changed due to the COVID-19 pandemic, the more visible, wider, and longer lasting the scope of environmental impacts may be. Your challenge is to tell a visual story about the impacts of the pandemic using Earth observation data and other complementary information.
Humanity is in the middle of a global health crisis. COVID has impacted countries globally that are already striving to improve their health care systems. Pandemics like these have shown the need and importance of preparedness. Handling a Pandemic requires not only an individual to take appropriate measures, but whole governments and entire societies to participate and help eradicate the spread. In great adversities like these, we should find the strength, tools, and systems to rise above and help humanity.In this project, I present the impact of Pandemic using Earth observation data and propose new indicators to be integrated with EODashboard as part of COVID impact as seen by Satellites.
I analyzed the EO Dashboard for these indicators for Los Angeles counties in the USA: COVID 19 data, Air quality, and car/container activity to understand the impact of COVID on the other factors. It is noticeable that as COVID cases increased and a lockdown was imposed, Satellite data shows significant drops in NO2 levels. As the lockdown restrictions were lifted, the NO2 levels started to increase in an upward trend again. For the same period, the car activity is reduced during the lockdown period. The Satellite imagery shows an interesting pattern wherein rental cars are filled in different stadium parking lots. The car activity indicators show the difference in the car densities before the start of restrictions and during lockdown periods. It also shows parking lots full in other areas like airports etc. The data shows how pandemic affected human travel, unused rental cars filling up vacant spots impacted rental car business as well as reduced NO2 levels. The COVID numbers in the LA area have a direct correlation to car activity and NO2 levels.
My project observes the relation between COVID, air quality, and car/container activity by presenting the difference in the numbers for each of these indicators with trends in COVID. I downloaded the time series data for these indicators from different satellite data sources like Sentinel hub. A machine learning model needs to be implemented to understand the correlation between the data. EO dashboard requires a combined indicator (Groups COVID data with mobility and environment indicators ) - to show how COVID affects human patterns and the environment. For the study, I developed time-lapse videos for the above indicators in the LA area between pre-lockdown and during lockdown in 2020.
My project also determines the need to have a hospitals detection mechanism be built that will be a part of the mobility indicator or a new indicator called Hospital fullness. This is specifically to analyze the parking lot traffic for hospitals. It has been observed that the rise in hospital parking lot fullness indicates an approaching high number of hospitalizations in upcoming months(Please see reference links). The new indicator will present a region/countries hospital capacities during major outbreaks. This will help public and health care officials plan for intake as well as prepare for growth in incidents such as a lockdown. The number of visits approximately can be calculated based on the car densities in the hospital parking lot. Along with hospital patient data from local sources, we can build the dashboard for the relation between hospital visits, COVID cases, and car activity during an outbreak. With Satellite data, we can see hospital parking lot trends over some time. This gives insights into COVID's impact on human activities change that will occur after a few weeks or months. Because it is observed that lockdown goes in to effect after increasing hospital activity in any region/country. My project aims to present a need for this new indicator by showing data around activity vs outcome.
My inspiration was the Satellite data that provides invaluable data and dimensions to look at, that were unprocessed before. Satellite data can cover remote areas and observe/record images that people can’t always detect on the ground in real-time. Mobility data study helps in people and government efforts to fight a pandemic. The data and dashboards could help the normal public to view the trends and possibly make them cautious and help to prevent the spread. Government can plan for lockdowns by observing trends in regional hospital fullness. I think satellite imagery analysis can provide the relation between COVID and human activity and we can build tools to predict an impending lockdown. Lockdown indicators also help view the trends in region-based activity according to the COVID numbers.
First I learned how to access Sentinel and GeoDB data in the Jupyter notebook for visualizing Satellite data. I picked the box/coordinates for LA to create a data cube filled with Satellite imagery. I was able to generate time series plots for different data types and export the data cube to GeoTiff. I tried to create time-lapse videos through Jupyter notebook, but I couldn't find appropriate variable/data and there were environmental issues(didn't get proper graphs). This will require some more time to figure out. So I built time-lapse using Quicktime player by using existing indicators in the EO Dashboard and EO Browser.
I used Sentinel hub data for exploration of images and building timelapse video using Jupyter notebook. I checked out https://eurodatacube.com/marketplace/data-products/public-open-data and used Sentinel hub in EO Browser and selected different themes to view the LA area at different times to build the presentation data for the project.
The two new indicators proposed in this project can be integrated into EO Dashboard. I couldn't build source code for integrating within EO Dashboard but have identified requirements for the indicators to build the algorithm.
1. Combined indicator: Users can select multiple indicators in the dashboard to view a resulting dimension. Let's take an example there is a new lockdown indicator, users should be able to add other indicators like mobility data or air quality or covid data to see the impact of lockdown in any region due to the related indicators. Currently, indicators in the dashboard show individual views, and in the graphs, COVID data is added from other data sources at run time. Every indicator has a timelapse where a period is selected, instead if there were covid numbers in place of timeline, then we can see as the covid numbers fluctuate the indicator pattern (mobility/air quality). A combined indicator like lockdown can show an aggregated or proportioned view of selected sub-indicators.
2. Hospital fullness indicator: Need to implement a hospital detection mechanism to identify hospital buildings from Satellite images. In the indicators, the user should be able to select dimensions like :
The number of hospitals by size/type - will show hospitals recognized by satellite.
Hospital capacity - will show color intensity for the hospitals according to the fullness of capacity. Hospital capacity is directly proportional to the parking lot fullness. There exists a car/container activity indicator in the dashboards. Utilizing this along with Covid data, we can present changes in hospital fullness over a period of time and during a pandemic.
These two indicators follow similar inputs like the NO2 air quality indicator where the user can draw a polygon or rectangle area to select an area of interest.
The 2 indicators should be stored in GeoDB. The indicator data could be cached as JSON into AWS S3. The hospital fullness indicator is calculated by aggregating average Hospital parking lot car/container number time series data. For the #1 requirement, unsure of how to build a different dimension like covid numbers instead of timelapse (selecting period data). I haven't started building this yet given the time constraints, but I should be able to explain my idea and work or help the NASA/ESA/JAXA team build it to completion.
https://eurodatacube.com/marketplace/data-products/public-open-data
https://www.gislounge.com/analysis-of-satellite-imagery-and-search-data-suggests-the-coronavirus-outbreak-may-have-started-earlier-in-2019/
https://www.un.org/sustainabledevelopment/health/
https://www.devex.com/news/data-mapping-key-to-track-disease-spread-and-plug-health-gaps-experts-say-96772
https://medium.com/euro-data-cube/exploring-time-and-space-a-guide-to-accessing-analysing-and-visualising-data-in-the-euro-data-e4a46f2bb55b
https://people.cs.vt.edu/~ramakris/papers/06798554-parkinglots.pdf
https://blog.mapbox.com/notable-maps-visualizing-covid-19-and-surrounding-impacts-951724cc4bd8
https://www.cdc.gov/mmwr/volumes/69/wr/mm6933e2.htm
https://www.devex.com/news/data-mapping-key-to-track-disease-spread-and-plug-health-gaps-experts-say-96772
https://www.un.org/sustainabledevelopment/health/
https://www.un.org/sustainabledevelopment/
https://www.gislounge.com/analysis-of-satellite-imagery-and-search-data-suggests-the-coronavirus-outbreak-may-have-started-earlier-in-2019/
https://www.gislounge.com/using-location-data-to-map-peoples-movements-social-distancing-efforts-and-the-spread-of-covid-19/
#air quality #pandemic #hospitals #new indicator #car activity #parking lot detection #COVID preparedness
This project has been submitted for consideration during the Judging process.