CarbonTrack: Visualizing AI Upscaled Ecosystem Dynamics
With 30% of atmospheric CO2 being absorbed by our terrestrial environment, it is more important than ever to measure and monitor the amount of environmental CO2 absorption. Researchers use the metric Gross Primary Productivity (GPP) to quantify this absorption. Physical structures, called flux towers, have been placed around the world to acquire this GPP value, however these towers can only measure a small area and are sparsely located, with most of them located in the US and EU. Researchers have taken to advanced technology to upscale this data using machine learning and remote sensing techniques. While these models have become increasingly more accurate, the large amount of data output and lack of tools has prevented further researchers and policy makers from using and building on these models. The CarbonTrack team set off to create a user-friendly data platform for hosting, visualizing, and analyzing GPP data at a global scale.
Our team had the opportunity to collaborate with the Quantitative Ecosystem Dynamics Lab out of the UC Berkeley department of environmental science, management, and policy on this topic and leveraged much of their knowledge and experience in the environmental research area. CarbonTrack was created to handle the large scale GPP data and be able to visualize, analyze, and share this data with research groups and environmental policy makers all over the world.
The team leveraged the Google Earth Engine platform to create features that could help users make the most out of the terabytes of GPP data. CarbonTrack can compare different models for the use of error analysis or to differentiate the model outputs in different land cover types. The user is also able to use the Yearly Trend feature to monitor, quantify, and assess efficacy of reforestation projects or carbon credit investments. The entire CarbonTrack product was design to assess this data on a global scale while also seamlessly allowing the user to assess the data at a more granular kilometer scale.
CarbonTrack was put in the hands of UC Berkeley graduate students with varying areas of study, and the feedback the team received was phenomenal. Many of the early product users were excited to implement CarbonTrack into their work right away. One user who studies wildfires, was excited about the ability for CarbonTrack to draw polygons on the interactive map and track the GPP trends through various seasons of wildfires.
The CarbonTrack team is excited to continue to put their data platform into the hands of researchers and policy makers so that more information can be unlocked and used in our fight against climate change.
The team would like to acknowledge Yanghui Kang, PhD and Maoya Bassioni, Ph.D for their guidance and experience in the world of GPP. The team would also like to acknowledge Professor Alberto Todeschini and Professor Puya Vahabi for their help through product planning and development, as well as their data science knowledge shared throughout the semester.