Robustness of the modes of Indo‐Pacific sea level variability

Reliable projections of future sea level are vital for coastal communities. Regional deviations from global mean sea level rise are caused primarily by spatial patterns in the various components of sea level change (e.g. melting land ice, thermal expansion, and so on; see Slangen et al. 2012; Perrette et al. 2013; Church et al. 2013) and by wind anomalies associated for example with various climate modes. Whether and how these climate modes might change in the future is an important question, however, understanding their current signature in sea level is the first step. Sea level is observed using both satellite altimeters and tide gauges. Altimetry provides excellent spatial coverage but the time series is of limited length, meaning that interannual to decadal variability remains difficult to resolve. Longer records from tide gauges are available, however, these lack spatial resolution as they measure SSH only at a few points along the coastline. The sparsity of tide gauges along with their uneven spatial distribution makes it difficult to study the spatial patterns of variability in both the empty interiors of the ocean and along the large sections of coastlines where tide gauge records are short or altogether absent. This problem is particularly apparent in the southern hemisphere, where long tide gauge records are extremely sparse. Statistical reconstructions using altimetry and tide gauges are of limited use for studying low frequency variability since these time scales are not sampled in the short satellite record (Ray and Douglas 2011; Meyssignac et al. 2012b; Calafat et al. 2014). The problem of extracting low frequency variability from short time series is exacerbated by the presence of the trend due to anthropogenic climate change [and vice versa, see Cazenave et al. (2014), for example]. Separating the forced trend from the natural variability is not as simple as Abstract This study evaluates the influence of various climate modes on sea level. The altimetry record has excellent spatial coverage but the limited length becomes an issue when evaluating low frequency variability in the presence of a trend. We use altimetry along with two ocean models to study how the relationship between sea surface height (SSH) in the Indian and Pacific Oceans and four climate modes [the Pacific Decadal Oscillation (PDO), the El Niño–Southern Oscillation, the Indian Ocean Dipole and the Southern Annular Mode] depends on the length of the time series. For low frequency variability such as the PDO, a time series on the order of 50 years in length is required to separate variability from the trend. Using shorter time series results in aliasing of the SSH trend and low frequency variability, which has implications for ascertaining the role of the PDO in the SSH trends. We find that the regression of SSH on to the PDO during the altimetry period, which is thought to have been responsible for a large fraction of the recent western Pacific SSH trend, is not representative of the SSH–PDO relationship in the longer-term record.