Sea‐Level Trend Uncertainty With Pacific Climatic Variability and Temporally‐Correlated Noise. Issue 3 (13th March 2018)

• Record Type: Journal Article
• Authors: Royston, Sam; Watson, Christopher S.; Legrésy, Benoît; King, Matt A.; Church, John A.; Bos, Machiel S.
• Is Part Of: Journal of geophysical research. Volume 123:Issue 3(2018)
• Page Start: 1978
• Page End: 1993
• Publication Date: 2018-03-13
• Subjects: sea level -- trend error -- uncertainty -- climate variability -- temporal correlation -- noise; Oceanography -- Periodicals; 551.4605
• DOI: 10.1002/2017JC013655
• Languages: English
• ISSNs: 2169-9275

Abstract

Abstract: Recent studies have identified climatic drivers of the east‐west see‐saw of Pacific Ocean satellite altimetry era sea level trends and a number of sea‐level trend and acceleration assessments attempt to account for this. We investigate the effect of Pacific climate variability, together with temporally‐correlated noise, on linear trend error estimates and determine new time‐of‐emergence (ToE) estimates across the Indian and Pacific Oceans. Sea‐level trend studies often advocate the use of auto‐regressive (AR) noise models to adequately assess formal uncertainties, yet sea level often exhibits colored but non‐AR(1) noise. Standard error estimates are over‐ or under‐estimated by an AR(1) model for much of the Indo‐Pacific sea level. Allowing for PDO and ENSO variability in the trend estimate only reduces standard errors across the tropics and we find noise characteristics are largely unaffected. Of importance for trend and acceleration detection studies, formal error estimates remain on average up to 1.6 times those from an AR(1) model for long‐duration tide gauge data. There is an even chance that the observed trend from the satellite altimetry era exceeds the noise in patches of the tropical Pacific and Indian Oceans and the south‐west and north‐east Pacific gyres. By including climate indices in the trend analysis, the time it takes for the observed linear sea‐level trend to emerge from the noise reduces by up to 2 decades. Plain Language Summary: We have made improved estimates of the time it takes for a trend signal to emerge from the inherent noise in sea level time series, in the Indian and Pacific Oceans. Sea level has been measured by satellites for over 20 years. There are apparent trends where sea level appears to be rising more in the west and central Pacific Ocean and rising less quickly along the US coast in this period, but these differences are due to long‐period climate variability. Additionally, sea‐level time series show strong temporal‐correlation, meaning each value is affected by previous values in time. We combine this knowledge to make improved estimates of the trend and residual noise in the sea‐level time series and discuss the time it takes for the observed trend to emerge from the inherent noise. Including climate variability in the assessment can reduce this time‐of‐emergence by up to 2 decades. There is an even chance that the observed trend from the satellite altimetry era (1993‐2015) exceeds the noise in patches of the tropical Pacific and Indian Oceans and the south‐west and north‐east Pacific gyres. Key Points: Accounting for ENSO and PDO variability does not affect the type of noise model that best describes the sea‐level trend residual Standard errors may be under (or over) estimated by assuming an AR(1) noise model when compared to a more realistic noise model The observed trend in the satellite altimetry era emerges from the intrinsic noise for some key locations in the Indian and Pacific Oceans