The Indian Ocean Dipole (IOD) describes an aperiodic oscillation of sea surface temperatures (SST) in the equatorial Indian Ocean [1, 2]. A positive IOD (pIOD) causes upwelling and cooling in the eastern equatorial Indian Ocean, off the coast of Java and Sumatra (Fig. 1), and droughts in adjacent land areas of Indonesia and Australia [1–3]. The western and central Indian Ocean shows moderate warming, which nevertheless causes above-average precipitation and flooding in the central Indian Ocean and equatorial East Africa due to the warm mean SSTs in the region [2, 4–6] (Fig. 1). The negative phase of the IOD (nIOD) causes opposite conditions, with warmer water and greater than average precipitation in the eastern Indian Ocean, and cooler and drier conditions in the west [1, 2].
A well-known feature of the IOD index is its skewness, as pIOD events may grow much larger than negative IOD events (nIOD), so that the IOD is positively skewed [7, 8]. The positive skewness of the IOD index reflects the negative skewness of SST in the eastern pole of the IOD (IODE; 90°E–110°E, 10°S-Eq.), as western Indian Ocean SSTs show only a weak positive skewness (Fig. S1) [9, 10]. The negative skewness of IODE SSTs is caused by a positive Bjerkness feedback involving the SST response to the depth of the thermocline in the eastern Indian Ocean: cold IODE SST anomalies lead to a zonal SST gradient that drives an easterly wind anomaly in the equatorial Indian Ocean, which further shoals the thermocline and reinforces the cold SST anomalies in the IODE region [7, 8, 11]. Due to this asymmetry, pIOD events tend to have stronger cold sea surface temperature anomalies over the eastern pole of the IOD than warm SST anomalies during nIOD events [8]. In addition, pIOD events display strong inter-event differences, with extreme events dominated by westward‐extended strong cold anomalies along the equator (Fig. 1), and moderate events with weakened cooling confined to the region off Sumatra‐Java [11, 12]. This is due to non-linear zonal and vertical advection of cold water during extreme pIOD events in the IODE region, which occurs in addition to the Bjerkness feedback [11]. Climate models suggest that extreme pIOD events may increase in frequency under greenhouse warming, due to the faster warming of the western Indian Ocean that favors nonlinear advection in the eastern equatorial Indian Ocean [12, 13].
Extreme pIOD events have particularly devastating impacts in the countries surrounding the Indian Ocean. For example, the extreme pIOD event of 2019 caused extreme droughts and bushfires over Indonesia and Australia, as well as severe flooding in equatorial East Africa followed by plagues of locusts [14]. The extreme pIOD event of 1997 led to large-scale warming in the western Indian Ocean (Fig. 1), which caused extensive coral bleaching and mass mortality [15]. Given the severe socio-economic impacts of extreme pIOD events, their adequate representation in instrumental SST products is of primary importance. However, the magnitude of cooling in indicated in the IODE region during extreme pIOD events varies between temperature products (Fig. 1). In particular, historical SST products based on interpolation from sparse historical observations underestimate the magnitude of the cooling, while reanalysis products that include subsurface oceanographic processes and their non-linear dynamics capture the magnitude of extreme pIODs more realistically [11]. However, beyond the start of the satellite era in 1982, the IOD index is based on historical SST products [16] that do not adequately capture the cooling in the IODE region [11]. Therefore, the magnitude of extreme pIOD events prior 1982 is difficult to assess.
Tropical corals can be used to reconstruct past changes in SST at monthly resolution by measuring the Sr/Ca ratios in skeletal aragonite, which have been shown to be a very reliable paleothermometer and provide independent constraints on historical SST observations [5, 17, 18]. The high temporal resolution of coral proxy data allows the reconstruction of seasonal climatic phenomena such as the IOD [19–22], and as coral Sr/Ca ratios are inversely correlated to ambient water temperatures during the corals’ growth, they should mirror SST variability in the IODE region, including its non-linearity. To date, however, most coral reconstructions from the IODE region focused on coral δ18O [21], which reflects a combination of SST and δ18O seawater and does not allow quantitative estimates of SST anomalies. In a recent study, 40-year coral Sr/Ca record from Enggano island, located off the coast of Sumatra (Indonesia) has been shown to track IODE SST variability [22]. Here, we extend this record back until 1930, present a replication core and reconstruct SST variability in the IODE region from Enggano coral Sr/Ca ratios. We use a Monte Carlo simulation to estimate the uncertainties of the multi-colonial reconstruction [23, 24]. The coral SST reconstruction is compared with various SST products (satellite observations, reanalysis products that include ocean dynamics and historical products based on statistical interpolation from sparse data), with a particular focus on extreme pIOD events. We will show that despite their uncertainties, the corals provide a better indicator of IOD-induced cooling in the eastern equatorial Indian Ocean than historical SST products interpolated from sparse data.