Spatio-Temporal Local Interpolation for Quantifying Global Ocean Heat Transport from Autonomous Observations

• Activity Number: 587 - Ocean Statistical Methodology and Application
• Type: Topic Contributed
• Date/Time: Thursday, August 6, 2020 : 3:00 PM to 4:50 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Beomjo Park* and Mikael Kuusela and Donata Giglio and Alison Gray
• Companies: Carnegie Mellon University and Carnegie Mellon University and University of Colorado Boulder and University of Washington
• Keywords: spatio-temporal statistics; nonparametrics; model misspecification; oceanography; climatology

Abstract

We investigate nonparametric Spatio-temporal interpolation techniques for estimating the global ocean heat transport based on in-situ observations from the Argo profiling float array and regional Spray gliders. The proposed methods are motivated by the major statistical challenges—global non-stationarity, massive data and model misspecification—arising in the data-driven estimation of ocean heat transport. Specifically, we adopt a two-stage locally stationary space-time regression model to capture the globally non-stationary heat transport process and to handle a large number of observations efficiently. A nonparametric debiasing method is applied to resolve mean-field misspecification by which sharp ocean fronts are insufficiently identified and the anomalous field overestimated. Our estimates are validated using alternative data sources, including a satellite altimetry product, to demonstrate the performance of the proposed approach. We discuss the significance of our improved heat transport estimates for global environmental and climate science.