Tackling climate change with machine learning with Claire Monteleoni

Claire Monteleoni is an associate professor of computer science at the University of Colorado Boulder as well as the co-founder of the Climate Informatics Workshop in 2011 and its associated hackathon in 2015. Within its first five years, the workshop attracted climate and data scientists from over 19 countries and 30 states.

“I wanted to use machine learning to shed light on climate change. Understanding climate change is an urgent challenge. Meanwhile, climate science is an extremely data-rich field, especially considering the massive amounts of simulation output from physics-driven climate models, providing a lens into the distant past and distant future.”

Can you give us an overview of your work?

I am inspired by a vision that machine learning can shed light on climate change, and other major challenges facing society.

In pursuit of this vision, I have worked to expand the impact of machine learning into fields with societal benefit. My work applying machine learning to the study of climate change advanced the state of the art and helped launch the interdisciplinary field of Climate Informatics: Machine Learning for the study of Climate Change. The 10th International Conference on Climate Informatics will take place next month.

In which ways can informatics help tackle climate change?

Machine learning can be used to improve our understanding and forecasting of climate change, air quality, and extreme weather events, helping communities to better adapt and prepare. 

It can also be used to improve forecasting the power output from solar and wind energy sources, which is crucial for increasing society’s reliance on renewable energy. Here are a few examples from my own research group, but such progress has been happening worldwide. We showed that machine learning can be used to learn from historical temperature data in order to improve the predictions of an ensemble of physics-based climate models. 

We have also used machine learning for detecting avalanches from satellite imagery, forecasting tropical cyclone tracks at 6 – 24 hour lead times, and predicting extreme precipitation spells during the Indian Summer Monsoon.

 

So machine learning can help detect extreme weather events?

Physics-driven climate simulation models have generated more [data] than all satellite measurements of Earth’s weather. These data-driven technologies are actually the most cost-effective way to unlock insights from the massive amounts of both simulated and observed data that have already been collected.

There are a range of other problems on which machine learning can make an impact, and we encourage both machine learning researchers and climate scientists to get involved. The climate informatics endeavor is an exciting experiment in building a new interdisciplinary research field.

 

What about adaptation to the consequences of climate change – a quarter of Bangladesh, one of the most hurt countries by climate change, is currently flooded-. How can informatics help in that area?

The Indian Summer Monsoon has a major effect on the GDP of the entire Indian subcontinent. We have shown that machine learning can improve predictions of total monsoon rainfall, as well as predicting at a much more fine-grained scale. For example, our preliminary results show that machine learning shows significant skill in forecasting multi-day dry spells and active (wet) spells, at a lead time of 10 days.

Can you give us some examples of computer tools already being used to tackle climate change?

To my knowledge, machine learning is being used, in at least some capacity, in the climate and weather forecasting operations of the governments of the US, India, France, the UK, South Korea, and Japan.

How important is open source and open data for climate informatics?

Large and high-quality data sets are extremely important for the training of machine learning algorithms and models. 

Luckily, most climate and weather data is openly available, at least in the US. This has made it much easier for machine learning researchers to get involved.