AI4ES

Last updated Mar 28, 2023 Edit Source

Applications of AI/ML and AI/AI to Earth Sciences

See:
AI4ES/EO
AI4ES/Geospatial science
AI4G/Challenges in EO and Climate

# Resources

Earth system science
Accelerating progress in Climate Science (Schneider 2021, Physics today)
https://www.scientificamerican.com/article/how-machine-learning-could-help-to-improve-climate-forecasts/
https://e360.yale.edu/features/can-artificial-intelligence-help-build-better-smarter-climate-models
AI - A Game Changer for Climate Change and the Environment
Ten Ways to Apply Machine Learning in Earth and Space Sciences
Supercomputers, Climate Models and 40 Years of the World Climate Research Programme
Thoughtfully Using Artificial Intelligence in Earth Science
Advancing AI for Earth Science: A Data Systems Perspective
- high-priority challenges include:
  - a lack of publicly available benchmark training data sets across all science disciplines
  - a lack of interoperability among data sources, types, and formats (e.g., standard data formats for computer vision algorithms may be different from the standard formats for commonly used Earth science models)
  - limited availability of baseline pretrained models that can be customized for various types or modes of Earth observations
  - label or target values that are not usually structured, such as oceanic measurements from drifting buoys that cannot be adapted easily to the grid systems commonly used in ML algorithms
Weathering Environmental Change Through Advances in AI
Tackling 21st Century Geoscience Problems with Machine Learning
Advancing Application of Machine Learning Tools for NASA’s Earth Observation Data (NASA)
Training the Next Generation of Physical Data Scientists
Guest post (CarbonBrief): How artificial intelligence is fast becoming a key tool for climate science
How machine learning is helping us fine-tune climate models to reach unprecedented detail
Machine learning-based global maps of ecological variables and the challenge of assessing them (Meyer 2022)

# Talks

# Books

# Courses

#COURSE Applied Machine Learning Tutorial for Earth Scientists
- Machine Learning Tutorial for Earth Scientists created by members of the Atmospheric Science Department at Colorado State University
- Google Slides
#COURSE An Introduction to Earth and Environmental Data Science (Ryan Abernathey, Columbia UIniversity, 2021)
#COURSE Research computing in ES (Ryan Abernathey, Columbia UIniversity)
#COURSE Introduction to Environmental Data Science (Jerry Davis, SFSU Institute for Geographic Information Science, 2022)
#COURSE Ocean/Atmosphere Time Series Analysis
- https://github.com/jonathanlilly/time-series

# Infrastructure and data

See AI4ES/AI4ES data
Pangeo Forge. See AI4ES/Pangeo
Microsfot Planetary Computer
European Weather Cloud
- https://www.ecmwf.int/en/newsletter/165/computing/progress-towards-european-weather-cloud
Climate Engine
The OpenWIS Association AISBL
Euro Data Cube
- https://hub.eox.at/marketplace/notebooks
Ocean OPS
- Integrated information, maps and tools to help coordinate and monitor global ocean observation efforts
EUMETSAT data portal
- EUMETView
- Product navigator

# Code

See AI4ES/Pangeo
#CODE MetPy - Collection of tools in Python for reading, visualizing and performing calculations with weather data
#CODE Lexcube - Leipzig explorer of Earth data cubes
#CODE Ncvue - A minimal GUI for a quick view of netcdf files
#CODE xcast - A High-Performance Data Science Toolkit for the Earth Sciences
#CODE climate_indices - Climate indices for drought monitoring, community reference implementations in Python
#CODE CliMetLab - Python package to easy access to weather and climate data
- https://climetlab.readthedocs.io/en/latest/index.html
#CODE Intake - lightweight package for finding, investigating, loading and disseminating data
#CODE intake-esm - An intake plugin for parsing an Earth System Model (ESM) catalog and loading assets into xarray datasets
- https://intake-esm-test.readthedocs.io
#CODE Aospy - Python package for automated analysis and management of gridded climate data
#CODE EarthPy - A package built to support working with spatial data using open source python
- https://earthpy.readthedocs.io/en/latest/
#CODE Psyplot - Python package for interactive data visualization
#CODE PyRain
- See “rainbench” in AI4ES/AI4ES data
#CODE Verde - Processing and interpolating spatial data with a twist of machine learning
- https://github.com/fatiando/verde
- https://www.fatiando.org/verde/latest/index.html
#CODE Radiant MLHub API - Open Library for Earth Observations Machine Learning
#CODE GIBS Downloader
- command-line tool which facilitates the downloading of NASA satellite imagery and offers different functionalities in order to prepare the images for training in a machine learning pipeline
- #CODE https://github.com/spaceml-org/Self-Supervised-Learner

# References

#PAPER Earth System Modeling 2.0: A Blueprint for Models That Learn From Observations and Targeted High‐Resolution Simulations (Schneider 2017)
- http://climate-dynamics.org/earth-system-modeling-2-0/
- ESM with subgrid-scale (SGS) process models that learn automatically from two sources of information
  - Global observations. ESMs can learn directly from space-based global data, augmented and validated with more detailed local observations where available
  - Local high-resolution simulations. Some SGS processes in ESMs are in principle computable, only the globally achievable resolution precludes their explicit computation
- The automated learning from observations and high-resolution simulations will use statistics accumulated in time (e.g., over seasons) to:
  - Minimize model biases, especially biases that are known to correlate with the climate response of models. This amounts to minimizing mismatches between time averages of ESM-simulated quantities and data
  - Minimize model-data mismatches in higher-order Earth system statistics
#PAPER Machine learning for data-driven discovery in solid Earth geoscience (Bergen 2019)
#PAPER Machine learning and artificial intelligence to aid climate change research and preparedness (Huntingford et al., 2019)
- Nine basic dimensions of climate modelling (there include geographical location, the many configurations of ESMs, and alternative future trajectories in emissions policy) : Longitude direction, altitude direction, Vertical direction, Time, Different climate models, Climate model ensembles, Different initial state, Perturbed physics runs, Different future emissions
- Examples of existing AI applications to climate: Earth System modelling, Teleconnections, Weather forecasting, Future climate scenarios, Climate impacts, Climate datasets, Climate extremes
#PAPER Leveraging Modern Artificial Intelligence for Remote Sensing and NWP: Benefits and Challenges (Boukabara 2019)
#PAPER Deep learning and process understanding for data-driven Earth system science (Reichstein, 2019)
- http://www.ccpo.odu.edu/~klinck/Reprints/PDF/reichsteinNature2019.pdf
- ML approaches are increasingly used to extract patterns and insights from the ever-increasing stream of geospatial data, but current approaches may not be optimal when system behaviour is dominated by spatial or temporal context
- Earth system data are exemplary of all four of the ‘four Vs’ of ‘big data’: volume, velocity, variety and veracity
- One key challenge is to extract interpretable information and knowledge from this big data, possibly almost in real time and integrating between disciplines
- Two major tasks in the coming years: (1) extracting knowledge from the data deluge, and (2) deriving models that learn much more from data than traditional data assimilation approaches can, while still respecting our evolving understanding of nature’s laws
- #TALK Machine-learning-model-data-integration for a better understanding of the Earth System (Markus Reichstein) ^ea6338
#PAPER Machine Learning for Clouds and Climate (Beucler 2020)
#PAPER What is next for National Meteorological Services? (Arribas 2020)
- Head of informatics lab at Met office, Prof. U. Exeter, Turing fellow
- #TALK AI, a change in science/technology … or culture?
  - https://vimeo.com/438176732
  - https://www.ecmwf.int/sites/default/files/medialibrary/2020-07/14_July_Arribas.pdf
#PAPER A digital twin of Earth for the green transition (Bauer 2021)
- A digital twin of Earth is an information system that exposes users to a digital replication of the state and temporal evolution of the Earth system constrained by available observations and the laws of physics
- Digital twins must focus exactly on how best to realize this convergence of the modelling and observation worlds
- A methodological framework for the twin’s architecture already exists in the form of data assimilation, which numerical weather prediction has developed with success over decades
#PAPER Machine learning for weather and climate are worlds apart (Watson-Parris 2021)
#PAPER Artificial Intelligence Revolutionises Weather Forecast, Climate Monitoring and Decadal Prediction (Dewitte 2021)
#PAPER Towards neural Earth system modelling by integrating artificial intelligence in Earth system science (Irrgang 2021)
#PAPER Evolution of machine learning in environmental science—A perspective (Hsieh 2022)
#PAPER Machine Learning in Weather Prediction and Climate Analyses—Applications and Perspectives (Bochenek 2022)