Our team applied NASA’s OCO-2 data and ESRI’s tools and libraries to the UK city of Barnsley to create an application to examine and visualize carbon dioxide concentrations, and its monthly variation, on an urban scale. Anyone can access our CO2 visualization application here as well as our ESRI Storymap for our project.


More than half of the global population lives in urban areas, which are at the heart of industrial activity. Hence, cities are responsible for 70% of greenhouse gas emissions globally. More importantly, to reduce carbon emissions on a city by city basis, it is imperative to measure them as cities cannot manage what they cannot measure. Our aim was to create a useful application to enable measurement and visualization of carbon dioxide concentrations and its monthly variation on an urban scale. This can aid city officials to monitor emissions and strategize to reduce them.

What it does

Our application enables users to visualize carbon dioxide concentrations in parts per million and local monthly variations in fine detail. The concentrations are mapped over land use polygons derived from the Urban Atlas Service. Our application is one example of carbon dioxide concentration visualization over the city of Barnsley in the United Kingdom.

How we built it

NASA’s open source Orbiting Carbon Observatory-2 or OCO-2 data was used for obtaining carbon dioxide concentrations over the UK city of Barnsley. Owing to cloud cover, there was a high degree of missing data over the city of Barnsley.

To resolve this issue, the OCO-2 data was interpolated using a B-Spline interpolation to create a continuous surface of CO2 concentrations for each satellite overpass; 9 CO2 concentration time slices. The data from Mauna Loa Observatory in Hawaii was used to synthetically create more observations (on the local global minima and maxima) and recreating the seasonality in the data; this resulted in 22 gridded time slices. The city of Barnsley was then interacted with the land cover polygons from the Urban Atlas Service and the average CO2 concentrations for each time slice was found by using Zonal Statistics.

To create the monthly estimates for all the land cover polygons we used the Cubic-Hermite Spline Method to fit the global seasonality trend observed in the global CO2 data. After interpolation, the output is a feature layer, on which all polygons have the predicted CO2 concentration, adjusted for land cover type and mapped with monthly averages from 2015 - 2021.

The processed layer is hosted on the ArcGIS platform and made available in two applications (a location inspector and a time traveller) as well as a story map showing the difference between 2015 and 2021. The Application is a JS component that is used to show the change of CO2 over time for different land use classes around Barnsley; this application was made by using ESRI’s Javascript libraries and map layers. The legend on the maps are showing the CO2 estimates between 395 and 430 ppm for the obtained time period, in equal brakes to be consistent between the Application and the Story Map.

Challenges we ran into

This Hackathon was a good learning experience for us, as a team, and with any hackathon, time is always a challenge, but besides the obvious challenges (time, resources, and a lack of coffee), there were only three challenging moments as listed below.

  1. The first challenge was early in the Hackathon, while we were testing our idea, and wanted to use Space-Time Cubes. This turned out not to be possible because this technology is part of ArcGIS Pro, and it was not possible to download and start a trial version of the software from our location (France).

  2. The second challenge came at the end of the Hackathon, when we were building our story map. The account we created for the Hackathon is registered under; therefore the background maps in the web layer are also linked to this domain. Normally this is not a problem, except when you want to share the story map because a login is required to view the layers, and our audience might not be registered with ArcGIS. This is a big limitation, but luckily the same background layers are available through ArcGIS Online, so we switched, but lost much time investigating the problem (luckily, we managed to get the story map up and running).

  3. The third challenge is the steep learning curve you have to go through to use the ESRI ecosystem, but that is part of the fun of learning something new.

Accomplishments that we're proud of

The results are twofold: first, the creation of incredibly granular and up-to-date data on carbon emissions and the second is beautiful maps allowing cities for the first time to intuitively visualize their carbon emission levels. Bringing data and tools to cities to better understand their carbon emissions as well as intuitive ways to visualize them is a game-changer for cities: it allows them to better engage decision-makers and citizens, make more informed decisions and progress faster in mitigating the effects of climate change.

What we learned

We learned how to use the ArcGIS platform and how the javascript API works:

  • Implemented map view
  • Load data as layers (online or locally)

As well as how to create engaging story maps, and

  • Use feature layers to create web layers and interact with the JS component
  • Layout web layers to show the most important information

What's next for Seeing the Unseen

The ESRI Hackathon: Hack for a Sustainable Future allowed us to explore new ways to bring awareness of the CO2 emissions to the policy and decision-makers. Using ESRI technology made it easy to convert the highly scientific content to a format much easier to understand. In addition to helping cities “see” their carbon emissions, this data helps them decide on the most strategic measures to put in place and monitor the impact of these measures. We will share the findings of this hackathon with Everimpact and their city audience to further enable municipalities to monitor, strategize and reduce carbon emissions.

The current idea is that the Everimpact Data Science team will look at the explored methods and assess how we can refine them, do a scientific check for applicability, and hopefully, implement them for a real-world project. Scaling up and automating the process is definitely another aspect we will be looking into as well as modeling multiple cities or a large stretch of motorway to better understand the transport of CO2 emissions.

Share this project: