Award Amount: $573,955
Duration: 3 years
Project title: Valuation of Surface Ocean pCO2 Observations For Machine Learning Applications
Why we care
The ocean absorbs about 25% of annual anthropogenic carbon dioxide emissions. To monitor this key climate service, we must understand air-sea carbon dioxide exchanges across the globe. Machine learning allows us to interpolate surface ocean carbon dioxide (called pCO2) by estimating relationships with more commonly measured ocean variables like temperature and salinity. This project uses machine learning to determine how to best deploy observing assets like buoys, autonomous vehicles, and ships to measure the ocean’s uptake of carbon dioxide globally.
What we will do
The Surface Ocean CO2 ATlas (SOCAT) is the largest global database of pCO2 measurements, but only covers 1.6% of the global ocean and is biased towards the northern hemisphere. Machine learning (ML) is now widely used to interpolate these data to derive monthly estimates of spatially-resolved air-sea CO2 fluxes, and from these, estimates of the integrated ocean carbon sink. This project will use machine learning approaches to quantify the value of each individual pCO2 observation in SOCAT, evaluate interpolation algorithms, and use a model testbed to suggest improvements to the distribution of existing observations.
Benefits of our work
Limited resources means there are limited numbers of ships, buoys, and autonomous vehicles that can be deployed. In addition, global shipping routes do not cover inaccessible regions and unsafe waters. This work will assess how we can optimize observing resources from the global fleet to support improved, efficient, and cost-effective monitoring of the ocean carbon sink and minimize uncertainty.
Investigators
Galen A. McKinley, Columbia University
Amanda R. Fay, Columbia University
Thea Hatlen Heimdal, Columbia University
Tian Zheng, Columbia University
Viviana Acquaviva, Columbia University and CUNY-Brooklyn Tech
Resources
Access the carbon cycle data.
Read the carbon cycle explainer.
Get the carbon cycle story with animations.
Bipartisan Infrastructure Law
The funds for this project are provided by the Bipartisan Infrastructure Law through President Biden’s Investing in America agenda. This project will allow NOAA to better monitor surface ocean CO2 and ultimately improve understanding and forecasting of global climatic and environmental change. For more information on this investment by the Biden-Harris administration, please see the associated press release here.
Image: Map of SOCAT (v1.5) surface fCO2 values released on September 14, 2011. The map shows lines from ships of opportunity equipped with sensors that measure ocean carbon. Credit: NOAA PMEL