What inspired our team to choose this challenge?

We live in large towns or cities it was only natural to think about phenomenons occurring in urban environments. Thinking how COVID and lockdown changed our usual daily routines, we have noticed that on one hand there are abandoned brick and concrete desserts in the city centre where we used to go to work, school, do our shopping and on the other hand there are now full houses and people going out to local parks for their daily exercises.

Urban Heat Island is a phenomenon observed in built-up areas. It affects the everyday lives of residents and increases heat-related mortality. There are observed correlations of growing UHI with increased usage of air conditioning. As we are approaching Summer time, with lockdown measures in place, we can expect increased use of private air conditioning appliances in residential areas. It is important to understand the possible increased risk of heat-related heart problems, especially for elderly residents locked down in residential areas. Being able to assess the magnitude of the change local authorities will be able to deploy life-saving resources more effectively to residential areas.

Quantification of the occurring UHI change may also serve as an useful data source for researchers working on pollution, greenhouse gases emission, energy efficiency use whether as an input or control dataset.

What was our approach to developing this project?

Having time limitations in mind we have focused on London and Paris and their surroundings to study the UHI phenomena, formulating following research questions:

• Is there a shift in Urban Heat Island pattern due to limited human activity?
• Has the drop in day time population due to lack of commuter impacted Urban Heat Island?
• What is the magnitude of the change?
• Are there any local shifts in urban heat island pattern?
• How quickly did the temperature change after lockdown measures were put in place?
• What will be the delay in UHI response to easing the lockdown measures?

We have identified potential locations that may have been affected the most and studied them on a case by case basis.

For places that have experienced reduced human traffic we looked at office areas, main commuter stations, shopping areas and tourist landmarks. For places that have experienced an increase in day time population we looked at residential areas and local parks adjacent to them.

To make our analysis insusceptible to seasonal meteorological changes we have used relative Urban Thermal Field Variance Index (UTFVI) calculated as the difference of Land Surface Temperature (LST) in any given point to mean Land Surface Temperature (LST mean) for the study area.

 UTFVI = (LST - LST mean) / LST mean

Next step was to create a difference map between the two snapshots in time to compare the data.

We have also used the same principles for comparing meteorological data from Meteomatics. We identified locations within London and outside London, calculated the difference between them for two points in time. Comparisons were made for day-time temperatures for the period of March and April for both 2019 and 2020.

How did we use space agency data in your project?

We did that for both London and Paris. We used Landsat-8 OLI/TIRS bands 4 and 5 for NDVI calculation and band 10 to assess Land Surface Temperature and further to derive relative Urban Heat Index using the above equation.

What tools, coding languages, hardware, software did we use to develop your project?

We downloaded LANDSAT data using the Earth Explorer engine and explored it in QGIS software, using Python to speed up the process. Time series analysis of temperature data was conducted in Python. For chart plotting, we used Matplotlib library. Meteorological data provided by Meteomatics was accessed using REST API.

What problems and achievements did our team have?

The major problems that we encountered were:

• Cloud interferences found in LANDSAT images and the acquisition frequency proved it difficult to find suitable imagery for LST and UHI calculation. To mitigate these problems in future research, we would advise using data from other data sources such as ASTER. We also considered using MODIS data, as it has greater acquisition frequency, however, lack of detail due to smaller resolution was unacceptable for the study focused on local changes within a city.
• The difference in meteorological conditions on the days LANDSAT images were captured. Even though we used relative metric for Urban Heat Index calculations, different meteorological conditions can make data incomparable. There are studies showing that the Urban Heat Island is more prominent during heat waves and less prominent under other conditions. Atmospheric fronts can make data unusable.
• Limited access to meteorological data (number of queries per day, historical data ) from Meteomatics to use for time series analysis.

The major achievements of our team were:

• Development of python scripts speeding up our work with data.
• Calculation of Urban Thermal Field Variance Index for London and then for Paris and the difference. The result from this part of analysis shows that local UHI index variations in London ranged from -0.5 to 1. Variance for Paris ranged from -1 to 1.
• Identification of residential regions with increased UHI like Walthamstow, London.
• Identification of decreased UHI in Paris around Eiffel Tower and other parts of the city centre.
• Additional access to meteorological data provided by Meteomatics, gave us the opportunity to look at the temperature time series for the period of March and April 2020 (pre-lockdown and lockdown) for London and surrounding rural areas. Also we accessed historical data for the same period in 2019 to use in further comparisons.
• Comparison of the difference in day time temperature for London and its surroundings in 2020, observing the decrease in the UHI after 2 weeks of lockdown measures being in place.
• Confirmation of our observation using data for 2019. The average difference in UHI magnitude between 2019 and 2020 for the period of lockdown (23 March - 30 April) was -0.4°C, which supports our thesis that with reduction in human activity and number of commuters in London, UHI has flattened.
• Setting up our website and Twitter account to share our journey with the wider SpaceApps Community.

https://github.com/xmichele/DataRoamers/blob/master/Presentation - Quiet Planet Challenge .pdf

Data:

• EarthExplorer - Home: LandSat 8 OLI/TIRS data obtained from Collection-1 Level-1 Dataset. Image Capture dates: London:15.04.2020 & 21.05.2019, Paris: 01.04.2020 & 02.06.2019
• Meteomatics: REST-style API to retrieve historic, current, and forecast data globally.
• Built-up Areas (December 2011) Boundaries V2: the digital vector boundaries for built-up areas in England and Wales as at 27 March 2011 (Census day). Last updated: 07 July 2018.
• Daytime Population of London 2014: this data presents the average daytime population on a weekday during term-time.
• DIVA-GIS: Data for Paris background layer, France Administrative Area.

References:

• United States Environmental Protection Agency: Heat Island Impacts.
• United States Environmental Protection Agency: Measuring Heat Islands.
• Coronavirus (COVID-19) Mobility Report: provide evidence of how COVID-19 restrictions have affected movement in London.
• Heaviside, Clare & Macintyre, Helen & Vardoulakis, Sotiris. (2017). The Urban Heat Island: Implications for Health in a Changing Environment. Current environmental health reports. DOI: 10.1007/s40572-017-0150-3.
• Jovanovska Kaplan, Gordana & Avdan, Ugur & Yigit Avdan, Zehra. (2018). Urban Heat Island Analysis Using the Landsat 8 Satellite Data: A Case Study in Skopje, Macedonia. DOI: 10.3390/ecrs-2-05171.
• Géo Tech. “Calculating Land Surface Temperature Landsat8 by ArcGIS”. YouTube, May 5, 2018, https://www.youtube.com/watch?v=G09MOTHFikM&feature=youtu.be.