Optimal fingerprinting of the upper 2000m of the ocean for forecasting heat content trends

• Main Author: Dowling, Henry
• Other Authors: Huybers, Peter, Tziperman, Eli
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: Climate change; Meteorology; Pacific; North Atlantic; Sea ice
• Online Access: https://nrs.harvard.edu/URN-3:HUL.INSTREPOS:37372698

Anthropogenically-induced ocean heat uptake is an important climate variable, whose future value has implications in the expected pattern of sea-level rise, climate sensitivity, and extreme weather events. However, trends in ocean heat content (OHC) due to forcing are difficult to detect. As a result of its large thermal mass, the three-dimensional ocean heats slowly, whereas the near-surface ocean experiences great variations in temperature on annual and interannual timescales. Fingerprinting is a useful technique for identifying a weak signal amidst noisy data. It has been used successfully for detection of anthropogenic influence in a variety of climate variables, such as near-surface air temperature, sea surface temperature, and sea ice extent. OHC is therefore a promising potential application for this technique, yet horizontally-sliced fingerprinting on three-dimensional OHC has not been attempted. An optimal fingerprint analysis is applied on 100 years of synthetic historical OHC data from the ACCESS-ESM1-5 climate model from the time period 1850-1900 and 15 years of Argo data 2004-2018, on 15 5-degree-gridded constant-pressure level surfaces in the upper 2000m. We propose a modified fingerprinting approach in which the primary modes of thermal variability—those caused by the annual cycle and ENSO—are controlled for before the fingerprint is computed. We derive an optimal fingerprint for detecting change in OHC in 34 pressure levels in the upper 2000m of the ocean. We find that the optimal fingerprint for OHC has strongly positive coefficients in the western Pacific; this can likely be ascribed to the relative quiescence of this region. Despite a generally positive trend, the optimal fingerprint is characterized by negative coefficients in the upper North Atlantic. The optimal fingerprint resembles the anticipated signal direction increasingly with depth, highlighting the value of examining three-dimensional OHC. We compute optimal detection variables corresponding to these fingerprints, and determine ...