We developed a script that collects and plots the raw data of the "nightlights" indicator over the Bay Area and then compares the same months of 2020 and 2021 (January, February, March) in order to show trends in population migration. It works in a similar manner to the sliders present on the dashboard; however, it is colour coded to clearly highlight areas where the number of lights increased (red) and those where they decreased (blue), making it easier to spot changes.

We then compared the changes shown by the overlapped maps to data about the trends in the Bay Area real estate market, in order to understand how the pandemic affected the relocation of people. The changes observed are important because they clearly show that given the possibility and wealth, a significant number of people would rather live away from city centres and densely populated areas and would prefer to buy a larger house somewhere calmer, where commute, noise and space are not as much of a problem.  

This is in stark contrast to the way mankind tends to build urban areas and commercial hubs, which are commonly associated with stress, noise, pollution and overcrowded areas.  As of now, only wealthy people can afford to relocate within a few weeks/months into a suburban mansion that costs upwards of $3 million, while poorer people are forced to stay in small flats even during a pandemic/lockdown. As more office-based companies adopt a WFH approach, more people will be able to avoid large cities, as they will not have to commute to work regularly. This way, they will be able to buy significantly larger properties for more affordable prices and better value-for-money.

The purpose of this project was to show how the future development of humanity revolves around the correct implementation of WFH (where possible), to improve the quality of life of those who would rather live a calmer, more isolated life while still being part of the working society. 

We chose this challenge because of its direct relevance to our team’s common background: as international university students currently studying in London, we are concerned about the habitation in urban areas such as the one our university campus is located in. Furthermore, the COVID-19 pandemic affected many changes in the way education was provided around the world, namely in the implementation of remote learning policies which resulted in many lecturers and students experiencing this past academic year from their homes or accommodations, without necessarily stepping foot on campus to begin with. Given the novelty of this approach, the team is interested in studying trends which are not only brought about by remote learning but by the WFH philosophy in general with the hopes of gaining insight into useful ways to help its implementation during and after the pandemic.

We began by brainstorming potential topic ideas which were encompassed under the chosen challenge. We identified several interesting avenues such as studying how social media usage or specific markets like commercial air travel, live service online videogames or the oil industry were impacted by the pandemic. In the end, we held a vote and opted for studying housing and population migration.

Having decided on a topic, our initial approach was to determine which EO Dashboard data and indicators would be most relevant. We looked at ones which demonstrated human habitation: urban indicators such as nightlights, car/containers and Facebook population density were studied, but environmental indicators such as CO2 and NOx emissions were also considered. We then analysed for which regions these data sets were available for us in order to choose the location for our case study. Our first option was London given its more local and immediate relevance to the team, but we settled on San Francisco and the bay area due to having data more readily available, our reasoning being that, if the data was provided, a similar analysis and process can be carried out for any metropolitan region in the world.

To develop our project, we used Jupyter Notebooks in the Google Collaboratory environment in order to have multiple team members working on the code simultaneously. The European Data Cube was accessed in order to obtain the Earth Observation APIs necessary to collect the data. Python libraries used include Requests, NumPy, Matplotlib, Xarray and the Python Imaging Library. An emphasis was placed on using commonly used libraries and modules which are supported and well documented in order to facilitate implementation of the code.

Our team ran into some challenges, namely in the development phase: none of the members had ever coded using APIs before, so learning on the go and figuring out was quite an adventure! Time constraints and issues with JavaScript coding (for Vue.js) meant that we did not run a local version of the EO Dashboard. That being said, notable accomplishments include very effective teamwork and clear communication which expedited the design phases of the project. 

The raw data from the nightlights indicator was used to show the trend of population displacement due to the pandemic, and it was compared to real estate market reports (see references). Seeing the areas that clearly showed variations in number of inhabitants allowed us to focus more on specific locations around the Bay Area in our market study, improving the quality and efficiency of our analysis.

The population density indicator was also used to check if the initial hypothesis was correct, as it helped us ensure that increases and decreases in population effectively corresponded to areas of lower and higher population density, respectively. This indicator would have been used as the base of our analysis; however, it only shows a single date so it could not be used to analyze a trend over time. The nightlights indicator on the other hand allowed us to compare the first 3 months of 2020 to those of 2021, meaning that we could observe the same period of the year before and during the pandemic, without having to consider changes due to seasons, holidays etc.

Click here to see a powerpoint presentation of our project.

Our solution could be implemented as an additional indicator. This would display a heat map, showing where the density of night lights has increased or decreased over the duration of the pandemic. Next to this, a population map could be shown, showing the trend of urban exodus.

Scrolling down, graphs showing the trend of house prices in the selected area over the duration of the pandemic could be shown for key areas, allowing for more quantitative comparisons to be made. 

As the code for generating the difference in night lights is relatively simple, integration would also be very simple. On top of this, night light data is readily and easily available for much of the world, and thus it is also very easily automated. Our code that we used to generate the images can be found here.

With enough data available, this could even be made into a global indicator, pulling data on the fly from the API and running the difference scheme on it. This could however prove too expensive and slow of an operation, and thus further testing would be required for this to be implemented.

The most difficult part of upscaling will be sourcing the house price data for additional cities. As there is no centralized worldwide database for this data, the dashboard would have to rely on more local sources, and in the worst case, analyses would have to be found for each individual city or region. However, for certain countries, this would not necessarily be required. For example, the United Kingdom has a nation-wide database of housing prices and trends with enough detail that they could be used. They also feature an API, making accessing them much simpler. This would allow for this data to be created for entire countries at a time, making upscaling substantially easier. 

Agency data:

Suomi NPP VIIRS

UN WPP NASA SEDAC

Non-agency data:

Bay Area Market Reports

HM Land Registry Open Data

#housing, #urban, #populationdensity, #economicimpact, #WorkingFromHome, #WFH, #urbandevelopment

This project has been submitted for consideration during the Judging process.