Earth Observation for Data Science: How we can Automate Climate Change Assessment from Above

Speaker: Thomas Y. Chen

Track: Data Science

Type: Keynote

As climate change accelerates, one of the most important ways in which we can utilize data science is promoting sustainability. This includes facilitating climate adaptation and mitigation. Machine learning and computer vision models, which are trained using Python libraries like Pandas, NumPy, PyTorch, etc., are key assets in assessing how climate change affects different aspects of the Earth. Earth observation data like satellite imagery serve as good sources of training data for these models. Data science algorithms can help develop natural disaster relief systems, monitor biodiversity, predict future climates, optimize carbon emissions and electricity use, help build smart buildings and cities, and much more.

This is a talk with a very broad focus and should be of interest to most attendees at the conference. This is because climate change is a crisis that affects us all, and the ways in which Python and data science can be used to tackle it are very exciting. I will start with an introduction consisting of describing the devastating the impacts of climate change. Then, I will discuss the basic Python tools that can have real-world impacts (libraries, etc.). I will give a short tutorial about how a convolutional neural network (CNN) can be trained using the PyTorch library, which can help develop disaster relief systems. Then, I will delve into the wide-ranging applications that can be implemented at the intersection of climate change and data science, all while connecting back to how Python can be utilized effectively. I will end with a call to action and an emphasis of how citizen science and crowdsourcing can be really key to addressing global warming; all Pythonistas can chip in and use their skills in data science (and Python in general) to contribute to this burgeoning initiative. The multidisciplinary focus of this talk, which highlights the contributions that Python users can make to the area of scientific computing, geoscience, and climate science, should attract a diverse audience.