Combining statistical, physical, and historical evidence to improve historical sea surface temperature records
Reconstructing past sea-surface temperatures (SSTs) from historical measurements containing more than 100 million ship-based observations taken by over 500,000 ships from more than 150 countries using a variety of methodologies creates a wide range of historical, scientific, and statistical challeng...
| Published in: | Harvard Data Science Review |
|---|---|
| Main Author: | |
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
2021
|
| Subjects: | |
| Online Access: | https://eprints.soton.ac.uk/484280/ https://eprints.soton.ac.uk/484280/1/51612882520141.pdf |
| Summary: | Reconstructing past sea-surface temperatures (SSTs) from historical measurements containing more than 100 million ship-based observations taken by over 500,000 ships from more than 150 countries using a variety of methodologies creates a wide range of historical, scientific, and statistical challenges. The reconstruction of historical SSTs for studying climate change is particularly challenging because SST measurements are uncertain and contain systematic biases of order 0.1°C to 1°C—these systematic biases are in the range of the historical global warming signal of approximately 1°C. The biases are complicated and have generally been addressed using simplified corrections. In this review, I introduce a history of SST observations, review a statistical method developed for quantifying SST biases, and illustrate scientific insights obtained from adjusted SSTs. This article also documents the scientific journey of my Ph.D. work. As a result, I report personal stories on both successes, difficulties, and setbacks along the way. The statistical method for correcting SSTs (i.e., a linear-mixed-effect intercomparison framework) depends on identifying systematic offsets between intercomparable groups of SST observations. Combining estimated offsets with physical and historical evidence has allowed for correcting discrepancies associated with SSTs, including the North Atlantic warming twice as fast as the North Pacific in the early 20th century and anomalously warm SSTs during World War II. Corrections also permit better hindcasting of Atlantic hurricanes. I conclude with some discussion on how the SST records might be further improved. Given the importance of SSTs for understanding historical changes in climate, I hope that this review can help others appreciate challenges that are present and spark some interest and ideas for further improvement. |
|---|