Climate Modelling Clouds With AI
If you are looking for a technical article on the topic of clouds the best paper award at NeurIPS within the climate change track is not the worst place to start. The article is called Cumulo: A Dataset for Learning Cloud Classes. It won within a track at one of the most prestigious AI conferences in the world, with such a large amount of papers submitted that is a feat worth recognition. Yet let us consider for a moment why this is important.
Why do we look at clouds with machine learning techniques?
To help I have brought in the abstract.
“One of the greatest sources of uncertainty in future climate projections comes from limitations in modelling clouds and in understanding how different cloud types interact with the climate system.”
One of the great aspects was Cumolo, the dataset they had provided:
“Cumulo, a new dataset which combines the global 1km-resolution imagery of the Moderate Resolution Imaging Spectroradiometer (MODIS) with the accurately measured properties of the CloudSat products. It contains one year of 1354 x 2030 pixel hyperspectral images from MODIS combined with pixel-width ‘tracks’ of cloud labels from Cloudsat, corresponding to the eight World Meteorological Organization (WMO) genera (Fig. 1). While both datasets are publicly available, the extraction, cleaning and alignment of the data required specialist domain knowledge and extensive compute resources.”
In doing so their goals was:
“[…] enabling the Machine-Learning community to develop innovative new techniques which could greatly benefit the Climate community.”
Their dataset contains 105,120 geolocated and hyperspectral images and provides a combination of channels from different sources.
The data looks rather stunning too.
Here are a few bullet points from the paper:
- Clouds play a crucial role in the climate system. They are the source of all precipitation and have a significant impact on the Earth’s radiative budget.
- There is a limited understanding of the mechanisms and relationships between clouds, climate and global circulation
This work is the result of the 2019 ESA Frontier Development Lab 7 (FDL) Atmospheric Phenomena and Climate Variability challenge.
By understanding the weather and attempting to predict how it unfolds we may better mitigate and address the emerging issues that are inbound in the coming years due to the ignorance of climate issues in the past.
This is #500daysofAI and you are reading article 282. I am writing one new article about or related to artificial intelligence every day for 500 days. My current focus for 100 days 200–300 is national and international strategies for artificial intelligence. I have decided to spend the last 25 days of my AI strategy writing to focus on the climate crisis.