RESEARCH @ LEAP
Overview
LEAP research aims to reduce errors in near-future Earth system projections by developing novel machine learning (ML) algorithms which will better extrapolate by including domain and causal knowledge, while aggressively leveraging the wealth of recent datasets.
Research Focus
LEAP will focus on hybridizing ML by integrating physical knowledge for implementation with the open-source Community Earth System Model (CESM) by:
- Reducing the existing model structural errors related to the lack of comprehension of the process at play (e.g., clouds + microphysics);
- Optimally estimating the model parameters using a Bayesian approach; and
- Developing new observational products which will be used to evaluate the CESM skill.
Research Impacts
The Problem
Earth scientists and Earth System modelers struggle to fully integrate the wealth of existing datasets into their models, while ML algorithms have been good at replacing and interpolating, but less targeted at extrapolating.
Our Solution
LEAP’s research triggers a significant advancement for data science applied to physical problems, by incorporating physics into ML algorithms to develop a next-generation CESM for better generalization and extrapolation, while optimally using the wealth of data available to Earth science. Our scientists merge physical modeling with ML across a “Knowledge-Data Continuum,” benefiting both scientific communities by providing a template for other Earth System modeling centers, as well as offering a more accurate approach to predicting the future of our Earth’s systems.
LEAP’s research triggers a significant advancement for data science applied to physical problems, by incorporating physics into ML algorithms to develop a next-generation CESM for better generalization and extrapolation, while optimally using the wealth of data available to Earth science. Our scientists merge physical modeling with ML across a “Knowledge-Data Continuum,” benefiting both scientific communities by providing a template for other Earth System modeling centers, as well as offering a more accurate approach to predicting the future of our Earth’s systems.
The Problem
Compounding historical shortcomings in Earth science and computing capacities that limit the accuracy of Earth system projections, inaccessibility to trustworthy, relevant, and accurate Earth system-related information makes it challenging for communities to optimally adapt and develop resiliency in the face of Earth system changes.
Compounding historical shortcomings in Earth science and computing capacities that limit the accuracy of Earth system projections, inaccessibility to trustworthy, relevant, and accurate Earth system-related information makes it challenging for communities to optimally adapt and develop resiliency in the face of Earth system changes.
Our Solution
By pursuing deep collaboration and engagement between climate data scientists, national research labs, public and private stakeholders, and other partners, LEAP will develop tailored and relevant Earth system-related information that will provide improved climate projections 10 to 40 years into the future. Communicated in ways that are more digestible and relevant, this information will benefit industry, government, and local communities with increased mitigation knowledge, more precise risk quantification, and the ability to better adapt to Earth system change to improve life for generations to come.
By pursuing deep collaboration and engagement between climate data scientists, national research labs, public and private stakeholders, and other partners, LEAP will develop tailored and relevant Earth system-related information that will provide improved climate projections 10 to 40 years into the future. Communicated in ways that are more digestible and relevant, this information will benefit industry, government, and local communities with increased mitigation knowledge, more precise risk quantification, and the ability to better adapt to Earth system change to improve life for generations to come.
Support LEAP
Help support our grand challenge to develop the next generation of climate model and climate projection for tailored adaptation.
Please contact us to start the conversation.