El Niño Detection Via Unsupervised Clustering of Argo Temperature Profiles. Issue 9 (1st September 2020)

• Record Type: Journal Article
• Authors: Houghton, Isabel A.; Wilson, James D.
• Is Part Of: Journal of geophysical research. Volume 125:Issue 9(2020)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-09-01
• Subjects: data clustering -- Argo floats -- ENSO dynamics; Oceanography -- Periodicals; 551.4605
• DOI: 10.1029/2019JC015947
• Languages: English
• ISSNs: 2169-9275

Abstract

Abstract: Variability in the El Niño‐Southern Oscillation (ENSO) has global impacts on seasonal temperatures and rainfall. Current detection methods for extreme phases, which occur with irregular periodicity, rely upon sea surface temperature anomalies within a strictly defined geographic region of the Pacific Ocean. However, under changing climate conditions and ocean warming, these historically motivated indicators may not be reliable into the future. In this work, we demonstrate the power of data clustering as a robust, automatic way to detect anomalies in climate patterns. Ocean temperature profiles from Argo floats are partitioned into similar groups utilizing unsupervised machine learning methods. The automatically identified groups of measurements represent spatially coherent, large‐scale water masses in the Pacific, despite no inclusion of geospatial information in the clustering task. Further, spatiotemporal dynamics of the clusters are strongly indicative of El Niño events, the east Pacific warming phase of ENSO. The fitting of a cluster model on a collection of ocean profiles identifies changes in the vertical structure of the temperature profiles through reassignment to a different group, concisely capturing physical changes to the water column during an El Niño event, such as thermocline tilting. Clustering proves to be an effective tool for analysis of the irregularly sampled (in space and time) data from Argo floats and may serve as a novel approach for detecting anomalies given the freedom from thresholding decisions. Unsupervised machine learning could be particularly valuable due to its ability to identify patterns in data sets without user‐imposed expectations, facilitating further discovery of anomaly indicators. Plain Language Summary: The climate phenomenon known as El Niño leads to variable temperatures and rainfall amounts around the world and occurs at unpredictable intervals. The most commonly used measurement to determine an El Niño is occurring relies on the difference between the 3‐month average temperature and the 30‐year average at the surface of the ocean in a rectangular region near the equator. However, as climate changes, these historically defined ways of measuring an El Niño may no longer be helpful. In order to develop a more flexible way to observe an El Niño, we use tools from the field of machine learning. Specifically, temperature measurements in the Pacific Ocean from the surface down to a depth of 1, 000 m are grouped automatically (i.e., without predefined rules) using machine learning methods. Without using information about the location of the measurements, this process groups measurements that are also close together in space. Changes over time of group assignments closely matches an El Niño happening and also point to physical changes to that region in the ocean. The automatic grouping of ocean profiles works very well to signal an El Niño and could potentially be a useful tool for future study of data from the ocean. Key Points: Unsupervised clustering based solely on temperature profiles effectively partitions water masses in the Pacific Ocean The temporal evolution of the clusters reveals spatial oscillations associated with El Niño events Unsupervised machine learning serves as a flexible and robust approach to anomaly detection in oceanographic data