Tackling climate change with machine learning [part 4] – Farms & Forests

No Comments

On 10th of June, 2019, twenty-two AI researchers, including Andrew Ng and Yoshua Bengio, published a paper on how climate change can be tackled with machine learning. I really enjoyed reading it and I am convinced that the paper as well as the climatechange.ai initative, which emerged from it, deserve more attention. For that reason i created a series of blog posts and videos which provide a dense summary, listing many of the proposed solutions and linking research work as well as ongoing projects. In the big picture, all solutions aim to reduce greenhouse gas emissions.

As my contribution to the global #ClimateStrike week from September 20th to 27th, i will post one chapter (video and blog post) on every working day. You can subscribe to our YouTube channel or follow me on Twitter to be notified when new content is published.

This is part four of a six-part series:

  1. Electricity Systems
  2. Transportation
  3. Buildings & Cities
  4. Farms & Forests
  5. Industry & Carbon Dioxide Removal (available from Thursday, Sept. 26th)
  6. Datasets & further resources (available from Friday, Sept. 27th)

Farms & Forests

Overall, land use is estimated to be responsible for about ¼ of greenhouse gas emissions. Melting permafrost is expected to add 12-17% of global greenhouse gases within the next decades. Furthermore, an increase of forest fires releases sequestered carbon as well. Better land management and more efficient agriculture could achieve ⅓ of all potential greenhouse gas reductions (from all climate solutions) according to Project drawdown.

Machine Learning can help reduce the carbon footprint of farms and forests by

Enhancing precision agriculture

[high leverage] [high risk]

Robot detects plant diseases and reduces fertilizer use and therefore nitrous oxide

Agriculture alone is responsible for about 14% of greenhouse gas emissions. There are three major causes of emission: 1. Cutting of trees, 2. Tilling of soil 3. Fertilizers, releasing nitrous oxide, which is 300 times more potent than CO2. Observing, measuring and responding to crop conditions is called precision agriculture.

Detecting lettuce with di-wheel robot

Autonomous lettuce detection and analysis [Sukkarieh, 2017]

ML can help

  • predict crop yield [Paper – Convolutional Neural Network, LSTM] [Paper]
  • detect plant diseases [Paper / GitHub-Repo – Fast variational Bayes inference for Latent Dirichlet Allocation]
  • sense soil e.g. for levels of nitrogen, water, carbon, texture and mineralogy [Paper]
  • reduce the use of fertilizers by only applying it to plants that need it [Paper]

Protecting peatlands

Peatlands

Peatlands cover only 3% of Earth’s land area, yet they hold twice as much carbon as all the world’s forests combined. A single peat fire in Indonesia in 1997 caused between 20% and 50% of all emissions of that year.

ML can help

  • protect peatlands by monitoring them, estimating peat thickness and predicting risks of fire [Paper – multiple techniques]

Estimating forest carbon stock

[high leverage]

Most carbon is stored above ground. The height of trees can be estimated with LiDAR sensors mounted on drones / unmanned planes. However, LiDAR technology is not scalable.

ML can help

  • predict the LiDAR’s outcome from satellite imagery [Paper] [Blog post]

Enabling automated afforestation

[long-term] [high risk]

According to research, more than one trillion additional trees can be planted in existing and new forests. They could sequester a decade of global CO2 emissions.

ML can help

  • enable tree planting drones to locate appropriate planting sites and monitor plant health [Blog post] [Project] [Project]

Helping with forest fire management

Forest fires

Forest fires obviously release carbon into the atmosphere and reduce the number of trees that can sequester carbon in the future.

Satellite and thermal image of forest fire

Satellite image (left) and thermal image (right) of a forest fire [Crowley et al., 2018]

ML can help

  • forecast droughts which helps to locate forests at risk [Paper – Decision Tree, Random Forest, Extremely Randomized Trees]
  • predict spatial progression of fires [Paper – Reinforcement Learning]

Tracking deforestation

[high leverage] [high risk]

Rainforest Connection Project

Deforestation contributes to approximately 10% of greenhouse gas emissions. Logging is the major cause. Tracking deforestation can inform policy makers and furthermore prevent illegal logging.

ML can help

  • prevent illegal logging by detecting it and its type (clearcutting vs. selective cutting) from satellite imagery [Paper – Random Forest]
  • prevent illegal logging by detecting chainsaw sounds using old smartphones [Project]

Credits

Many thanks to all researchers of the paper:

David Rolnick, Andrew Y. Ng, Priya L. Donti, Lynn H. Kaack, Kelly Kochanski, Alexandre Lacoste,  Kris Sankaran, Andrew Slavin Ross, Nikola Milojevic-Dupont, Natasha Jaques, Anna Waldman-Brown, Yoshua Bengio, Alexandra Luccioni,  Tegan Maharaj, Evan D. Sherwin, S. Karthik Mukkavilli, Konrad P. Kording, Carla Gomes, Demis Hassabis, John C. Platt, Felix Creutzig and Jennifer Chayes.

Continue to Part 5: Industry & Carbon Dioxide Removal (available from Thursday, Sept. 26th)

Paul Strobel

Paul is passionate about computer vision and neural networks for image recognition. In addition, he likes to create explanatory videos to easily convey complex content through stories. In his spare time he likes to play volleyball and guitar and occasionally does light painting photography.

Comment

Your email address will not be published. Required fields are marked *