Insight from Tide Gauges

Prior to satellite altimetry, the primary source of sea level data is from historical tide gauge records. These data provide a significantly longer time series relative to that of satellite altimetry, with a few records extending back into the 18th century. Several studies have estimated GMSL from the tide gauge record using a variety of techniques14,15,16,17,18,19,20. The resulting estimates of the GMSL trend from 1900 to 1990 range from 1.2 mm yr−1 to 1.9 mm yr−1, albeit with significant decadal variability about this long-term trend [1, Fig. 13.7b]. Coupled with the higher GMSL trend observed during the satellite altimeter record discussed above, the tide gauge record demonstrates unequivocal acceleration since the early 1900 s, with estimates ranging from 0.009 +/− 0.002 mm yr−2 17 to 0.017 +/− 0.003 mm yr−2 20. Based on these same studies, however, the majority of the acceleration arises from a shift that occurs around 1990 when the rate of sea level rise increases to the satellite-measured trend of 3.3 mm yr−1.

While the tide gauge record may provide a ballpark estimate for what to expect during the altimeter era, it provides only weak quantitative information regarding what acceleration should be expected. In practice, calculating GMSL from 1900 to the present is a challenging problem based on the spatial and temporal sampling characteristics of available gauges. There is little consensus across tide gauge studies on the rate and acceleration of GMSL over the past century, thus making it difficult to interpret the altimeter era in a broader context. As an alternative approach to understanding sea level variability, we therefore seek to estimate and remove effects that obscure a possible underlying acceleration from the altimeter record itself. By doing so, we can potentially estimate the acceleration in GMSL directly from altimetry.

The GMSL Influence of the 1991 Mt Pinatubo Eruption

On June 15, 1991 the second largest volcanic eruption of the twentieth century in terms of aerosol radiative forcing began on the Philippine island of Luzon21,22. Estimates of the amount of ash deposited in the stratosphere ranged from 20 to 30 Tg21 (approximately equal to a tenth the mass of all mankind) inducing a cooling of the globe and especially the world’s oceans. Model-based estimates of the eruption’s cooling effects suggest that the recovery of ocean heat content during the 1990’s may have increased sea level rise by as much as 0.5 mm yr−1 on average in the decade following the eruption23,24,25,26,27. The precise temporal evolution of the increase, and the impacts of the eruption on terrestrial and atmospheric water reservoirs remain largely unknown however; as quantifying these impacts is complicated by contemporaneous climate variability that can mask or amplify the response28. Moreover, as many of the tools now in place for monitoring climate, such as for example the ARGO network of ocean temperature sensors29 and the GRACE satellite for estimating terrestrial water storage30, were yet to be deployed at the time of the eruption, quantifying its precise climatic effects is nontrivial. However, recent specially designed climate model experiments now provide additional insight. What they reveal is that the eruption had a profound and temporally complex influence on several contributors to sea level and especially the amount of heat stored in the oceans, even when compared to background variations in climate, with major consequences for perceptions of GMSL acceleration during the altimeter era.

The CESM Large Ensemble (LE) is a 40-member ensemble of state-of-the-art coupled climate simulations spanning the 20th and 21st centuries using estimated historical forcings including volcanic eruptions [see Methods for a detailed description]. The simulated responses of GMSL and its individual contributors to the 1991 eruption of Mt Pinatubo are summarized in Fig. 2. Immediately following the eruption, aerosols in the stratosphere blocked sunlight and cooled the surface. Surface temperatures quickly dropped, particularly over land due to its relative lack of thermal inertia22, [Fig. S2]. In turn, the atmosphere cooled, reducing the amount of moisture stored within it as water vapor. A cooler surface evaporated less moisture and was less convectively unstable, leading to a subsequent reduction in rainfall globally and disproportionately over land where diminished land water storage and runoff were a consequence of the eruption31.

Figure 2: Simulated sea level rise contributions during and following the eruption of Mt Pinatubo. Shown are changes in (A) clear-sky albedo over the tropical oceans (30 N – 30°S) as an indicator of the eruption’s radiative effects and associated global mean sea level (GMSL) anomalies. In (B) contributions from ocean heat content (OHC) atmospheric water vapor (PW) and terrestrial water storage (TWS) estimated from the LE are shown. The large standard deviation across ensemble members (shaded) highlights the obscuring effect of natural climate variability on the eruption’s influence in observations. Full size image

As these terrestrial and atmospheric changes are associated with reductions in their storage of water, their initial influence was to delay by about six months the eruption’s main effect on sea level, which was a significant and rapid drop arising from a reduction in ocean heat content (OHC). The short timescale of the terrestrial and atmospheric influences relative to the oceans however limited their persistence, and by the beginning of the altimeter era in 1993, a GMSL drop of 5 to 7 mm from the eruption is estimated to have occurred, due largely to cooling of the oceans. While the LE’s estimated OHC deficit is difficult to verify directly, given the large uncertainties and errors inherent in global ocean observations32, confidence in the simulated response is bolstered by satellite estimates of the Earth’s radiative imbalance, which strongly constrain the magnitude of ocean cooling and agree closely with simulated fluxes [Fig. S3]. Confidence in the ability of the LE to capture fundamental features of the eruption is therefore high.