A new method, S-DEIM, improves the estimation of global sea surface temperatures from scarce observational data.
Image Credit: Mohammad Farazmand
Scientific Frontline: Extended "At a Glance" Summary: Sparse Discrete Empirical Interpolation Method (S-DEIM)
The Core Concept: S-DEIM is a model-free data assimilation method designed to reconstruct high-resolution global sea surface temperature (SST) fields from scarce observational data.
Key Distinction/Mechanism: S-DEIM improves upon older empirical methods by utilizing historical data to train recurrent neural networks (RNNs) to estimate a kernel vector for missing data points. It is 40% more accurate than the standard Discrete Empirical Interpolation Method (DEIM) and slightly more accurate than top convolutional neural networks (CNNs), requiring only a fraction of the computational training time (approximately one minute).
Major Frameworks/Components:
- Empirical Interpolation: Calculates instantaneous in situ observations.
- Recurrent Neural Networks (RNNs): Utilizes historical time-series data to learn and compensate for missing information.
- Historical Datasets: Trained using the National Oceanic and Atmospheric Administration’s (NOAA) high-resolution SST datasets from 1989 to 2021.
Branch of Science: Oceanography, Climatology, Mathematics, Computer Science, and Computational Science.
Future Application: Enhanced short-term weather forecasting, more reliable long-term climate modeling, and efficient monitoring of marine ecosystems using limited sensor arrays.
Why It Matters: Accurate sea surface temperatures are crucial for predicting weather patterns and understanding climate change; however, physical data collection via buoys and satellites is often limited by location or atmospheric interference. S-DEIM provides a highly accurate, computationally inexpensive tool to fill these critical data gaps.
Researchers have developed a method for extrapolating sea surface temperatures from sparse data; this method is significantly more accurate than other commonly used computational methods and slightly more accurate than the best-performing artificial intelligence (AI) model, all while taking a fraction of the time to train. The work has implications for both short-term weather forecasting and longer-term climate predictions.
“Sea surface temperatures (SSTs) are a key factor in understanding everything from marine ecosystems and climate to weather predictions, but there are limitations to our ability to collect those data, so we often have sparse or limited data,” says Mohammad Farazmand, an associate professor of mathematics at North Carolina State University and the corresponding author of the research.
SST data are collected from sources such as buoys and satellites, but there are limitations to each method. Buoys are more accurate but limited in number, while satellites cover a larger area, but atmospheric conditions can interfere with their accuracy. Therefore, oceanographers use complex mathematical models to obtain the most accurate estimations of sea surface temperatures.
“Historically, federal agencies such as the National Oceanic and Atmospheric Administration (NOAA) have used a combination of complex differential equations to calculate these temperatures from sparse data,” Farazmand says.
“Recently, some AI or machine learning models have been developed, but they are expensive both computationally and in terms of the time needed to train them. We wanted to see how our method stacked up against different computational methods, such as the Discrete Empirical Interpolation Method (DEIM), and some of the newer AI models.”
DEIM does not rely solely on complex mathematical models. Instead, it specifies a basis, or combination of patterns, that encodes information about the specific field researchers are trying to estimate—in this case, SSTs. However, DEIM does not work well with sparse data.
Farazmand and the team developed a new method, the Sparse Discrete Empirical Interpolation Method (S-DEIM). To compensate for missing or sparse data, S-DEIM utilizes historical data to estimate a so-called kernel vector, for which there is no closed-form mathematical formula.
The team compared the S-DEIM method to both DEIM and the highest-performing AI model, a convolutional neural network (CNN), using a dataset containing 30 years of NOAA data. They withheld the last year of data from the models and asked them to predict what that final year’s SSTs would be. Then, they compared the models’ predictions to the historical data.
S-DEIM was 40% more accurate than DEIM and 2% more accurate than the CNN. Additionally, the S-DEIM model took only 1 minute to train, compared to 1.5 hours for the CNN.
The researchers hope to continue improving the S-DEIM method’s accuracy.
“This work shows that S-DEIM is capable of utilizing sparse data to provide accurate results while reducing training and computational time,” Farazmand says.
Funding: Partially supported by the National Science Foundation (NSF) under award DMS-2349611, as well as through grant DMS-2220548 (Algorithms for Threat Detection program) and award DMS-2342344.
Published in journal: Journal of Geophysical Research: Machine Learning and Computation
Title: Rapid Estimation of Global Sea Surface Temperatures From Sparse Streaming In Situ Observations
Authors: Cassidy All, Kevin Ho, Maya Magnuski, Christopher Nicolaides, Louisa B. Ebby, and Mohammad Farazmand
Source/Credit: North Carolina State University | Tracey Peake
Edited by: Scientific Frontline
Reference Number: es090826_02
