Last glacial maximum and deglacial transition
Geological reconstructions from the Sunda Shelf and Singapore reveal the response of sea level to climate forcing as the Earth transitioned from glacial to interglacial conditions. Between ~ 26 kyr and ~ 19 kyr BP, global ice sheet complexes grew to their maximal extent19,27,34 and GMSL reached a lowstand of ~ 120 to ~ 130 m below present level (Fig. 5a, b). Atmospheric CO2 concentrations were between ~ 188 and ~ 194 ppm and global mean surface temperature (GMST) cooler (relative to 1850–1900 CE) by 5°–7°C35,36.
Increasing northern summer insolation began the onset of the LDT towards the early Holocene12,37 as atmospheric CO2 concentrations increased to ~ 270 ppm and GMST rose 1°–1.5°C kyr36. In response, GMSL rose as ~ 50 million km3 land-based ice was transferred to the global ocean primarily from deglaciating Northern Hemisphere ice masses27,34,38,39. This ice melt caused RSL on the Sunda Shelf to rise ~ 93 m over ~ 12,000 years from 21.5 kyr to 9.5 kyr BP at an average rate of ~ 7.6 mm/yr.
Superimposed on long-term secular rising GMSL have been several periods of short-term rapid increases40,41. An initial rapid increase in GMSL between 19.5 kyr and 18.8 kyr BP (MWP1Ao, Fig. 5b) caused RSL on the Sunda Shelf to rise ~ 8 m at rates of RSL rise up to 7 mm/yr (Fig. 3d, e). Similar rapid increases in RSL following termination of the LGM have been reported from geological reconstructions in both low27 (Fig. 5a, b) and high42 latitude settings.
A second rapid increase in GMSL between 14.8 kyr and 13 kyr BP (MWP1A, Fig. 5b) caused RSL on the Sunda Shelf to rise ~ 32 m at rates of RSL rise up to ~ 15 mm/yr (Fig. 3d, e). This averaged rate is conservative compared to annual rates of RSL rise of ~ 40 mm/yr26,41,43. Rising GMSL during MWP1A was the most rapid of the last deglacial period and was driven by the sudden influx of meltwater as deglaciation of Northern Hemisphere ice sheets continued and ice-dammed and subglacial lakes drained increasing ocean volume by ~ 470,000 km3 19,41,44,45. Geological reconstructions to support MWP1A are global in scale with widespread evidence within equatorial and tropical latitudes (Fig. 5 and references therein) that are also corroborated at higher latitudes46,47.
An additional rapid increase in GMSL during the LDT is postulated between ~ 11.5 kyr and ~ 11 kyr BP (MWP1B, Fig. 5b). Evidence from coral reconstructions in Barbados suggests RSL rose ~ 13 m at rates of RSL rise 20–40 mm/yr24,25,48. The occurrence of MWP1B, however, is currently unresolved in other equatorial and tropical reconstructions with an equivalent change in RSL not recorded on the Sunda Shelf and Tahiti26,43. While SLIPs (n = 5) from the Southern Vietnam Shelf originally presented in ref. 26 provide evidence of RSL change between ~ 13 kyr and ~ 11 kyr BP, they were not included here due to their spatial distance from North Sunda River and Singapore SLIP datasets and their proximal location to the Mekong River Delta and its associated influence of subsidence19,49.
The rise in GMSL after the LGM had profound impacts on shelf margins50 including the Sunda Shelf and significantly transformed the paleogeography of the region51–53. The low shelf gradient allowed sea level to rapdily transgress laterally increasing from an average rate of ~ 57 m/yr to ~ 335 m/yr during MWP1A. Rising RSL submerged coastal landscapes53,54 and segregated insular islands of Southeast Asia from the continental mainland dislocating flora and fauna migration routes52,55. In the Singapore Strait, the land bridge that existed during low GMSL was severed as rising seas flooded across sills to the east from the South China Sea and to the west from the Malacca Strait when RSL reached ~ − 30 m toward the end of the LDT56 (Fig. 1b). The flooding of the Sunda Shelf also altered the interchange of water between the Indian Ocean and South China Sea distrupting regional oceanographic and atmospheric climate systems57–59.
Holocene
Climate forcing during the Holocene up until the pre-industrial Common Era (~ 1850 CE) was relatively mild compared to the preceding LDT. Atmospheric CO2 concentrations were between ~ 260 and ~ 285 ppm36 and variability in GMST was reduced60 showing a slight but steadily warming trend of ~ 0.25°–0.5°C35,61. Despite relative climate stability, GMSL continued to rise tens of meters during the early Holocene as Northern Hemisphere ice sheets entered their final stages of disintegration and coastal ice streams broke19, 62–64.
In Singapore, RSL rose ~ 21 m between 9.5 kyr and 7 kyr BP at rates of RSL rise up to ~ 15 mm/yr (Fig. 3d, e). Ref. 65,66 suggested rapid RSL rise during the early Holocene in Singapore was temporally punctuated by a near cessation in RSL rate between 7.8 kyr and 7.4 kyr BP before continuing to rapidly increase again thereafter. Ref. 19, however, noted that the near-zero RSL rise during this period possibly reflects local processes because of an absence of similar trends at the global scale. Furthermore, ref. 67 concluded accurate verification of oscillating RSL in Singapore is precluded by large vertical and temporal scatter of SLIP data following their standardization68.
Rising RSL reached near present level by ~ 7 kyr BP and continued rising to a mid-Holocene highstand that is characteristic of far-field regions distal from ice sheets and driven by regional hydro-isostatic processes when meltwater input decreased21,22. The magnitude and timing of the highstand varies around the Sunda Shelf67,69,70 and in Singapore, RSL reached ~ 4.6 m at ~ 5 kyr BP. Falling RSL from the mid-Holocene highstand was driven by both hydro- and glacio-isostatic loading of the Earth’s surface (equatorial ocean siphoning and continental levering)71 and rotational feedback72. Late-Holocene SLIPs show RSL below present at −~2 m between ~ 2.5 kyr and ~ 0.5 kyr BP. Evidence to support RSL below present during the late Holocene is corroborated by mangrove SLIPs from Peninsula Malaysia which show RSL at ~ − 0.7 m ~ 0.8 kyr BP73–76.
The rise and fall in RSL during the Holocene continued to alter the paleogeography of the region. Rapid RSL rise during the early Holocene flooded lowlands surrounding Singapore and in the Johor Strait segregating Singapore from Peninsula Malaysia56,77. As the rate of RSL rise declined below ~ 7 mm/yr after ~ 8.5 kyr BP, widespread development of mangrove forests commenced as mangroves began to maintain their vertical position through sediment accretion78. The rise in RSL above present during the mid-Holocene continued to encroach and submerge low elevations of Singapore decreasing the terrestrial land area. The marine sediments that were deposited were subsequently weathered and partially eroded by sub-aerial processes as RSL fell during the late Holocene increasing land area as previous shorelines became progressively exposed79.
Twentieth and twenty-first century
Instrumental records cover changes in Earth’s climate driven by anthropogenic forcing80. Atmospheric CO2 concentrations have increased to ~ 421 ppm in 2022 CE81, which is unprecedented in at least the last two million years, and GMST rose ~ 1.1°C between 1850–1900 and 2011–2020, which is warmer than any multi-centennial interval during the LDT36. The response of GMSL to this forcing has so far been an increase in rate from 1.35 mm/yr82–84 between 1901 and 1990 to 3.7 mm/yr between 2006 and 2018 that was driven by accelerating land-ice losses and thermal expansion4. Long-term (i.e., > 50 years) instrumental records coupled with overlapping geological reconstructions have also demonstrated rising GMSL during the 20th century was faster (P ≥ 0.999) than the proceeding 3 kyr17, with a timing of emergence above background variability centred on 1863 CE85. Geological reconstructions using mangrove sediments from Peninsula Malaysia have suggested an increase in RSL rate from 1.26 mm/yr to 3.2 mm/yr after 1900 CE74.
The addition of geological reconstructions from Mapur Island, Indonesia provides an important methodological step toward further extending instrumental records of RSL change in equatorial and tropical latitudes such as Singapore86. The combined dataset shows RSL rose 0.15 m between 1915 and 2020 CE increasing from a rate of ~ 1.7 mm/yr to ~ 2.2 mm/yr. Singapore’s temporally limited instrumental records provide only a brief history of RSL changes with variability on different timescales87–89. Seasonal sea-level variability is influenced by meteorological forcing across the region90. In the Singapore Strait, the influence of Northern Hemisphere and Southern Hemisphere monsoon systems and their prevailing wind direction causes sea levels to vary up to ± 20 cm, the highest level of which coincides with northerly winds between December to early March87. On interannual timescales, sea-level variability is dominated by the El Niño–Southern Oscillation and during El Niño and La Niña events, changes in sea-surface height (SSH) can vary by up to ± 5 cm, with lower SSH observed during El Niño events87. On decadal and inter-decadal timescales, sea-level variability in Singapore and wider Southeast Asia is influenced by basin-scale climate modes including the Pacific Decadal Oscillation (PDO) and Interdecadal Pacific Oscillation (IPO)91,92.
Coastal landscape changes following British colonial establishment in 1819 and during the 20th and 21st century in Singapore largely reflect anthropogenic modifications to accommodate industrial and urban development93. Land reclamation projects expanded land area ~ 25% from 581.5 km2 in 1960 to 733.2 km2 in 202293 that significantly reduced coastal habitat extent94–96. Between 1922 and 2011, tidal flats and coral reefs reduced in area from 33 km2 to 5 km2 and 32 km2 to 9.5 km2, respectively. Furthermore, the damming of mangrove-fringed estuaries to create freshwater reservoirs resulted in a 91% decrease in mangrove forest extent, reducing in area from 75 km2 to 6.4 km2 95,96.
Perspectives of future sea level
The magnitude RSL rise since the end of LGM through the LDT, Holocene and towards the present demonstrates the long-term commitment and sensitivity of sea levels to climate forcing on timescales of centuries to millennia12. Rising GMSL initiated during the 20th century by anthropogenic forcing4 will continue even if CO2 emissions are drastically reduced due to the lagging integrated response time of deep ocean heat uptake and ice sheets12, 97–100. Indeed, over the next 2000 years, GMSL is committed to rise 2–3 m if GMST is limited to 1.5°C warming, 2–6 m if limited to 2°C warming and 19–22 m with 5°C of warming4. Projections of GMSL rise are consistent with geological reconstructions during past warm climate periods when GMST were higher101. For example, during the last interglacial ~ 126 kyr to 116 kyr BP, GMSL was 5–10 m higher than present when GMST were just 0.5°–1°C warmer than today4, 101–103.
Future sea-level rise will primarily be caused by global increases in ocean mass and volume associated with meltwater input from land-based ice sheets and glaciers and thermal expansion to warming temperatures4. Faster-than-projected disintegration of marine ice shelves may also exacerbate sea-level rise through Marine Ice Cliff Instability (MICI) processes104,105 for which there is low confidence4.
Under the very low emissions shared socioeconomic pathway (SSP)1-1.9 scenario, future RSL rise (relative to a 1995–2014 baseline) in Singapore will increase 0.58 m (likely [at least 66% probability] range of 0.31–0.93 m) at a rate of 3.4 mm/yr (1.2–6.2 mm/yr) by 2150 (Fig. 6a, b, Table 2). Magnitude 0.5 m and 1.0 m thresholds under SSP1-1.9 are expected to be surpassed by 2127 (likely range of 2084–>2300) and 2279 (2162–>2300), respectively. The geological past provides probability perspectives to when equivalent rates of RSL rise were last exceeded (Fig. 7). Rates of RSL rise increasing at greater than 3.4 mm/yr were very likely (at least 90% probability) between ~ 20 kyr and ~ 19.5 kyr BP and about as likely as not (between 33 to 66% probability) between 18.75 kyr and 16 kyr BP (Fig. 7a). Rates of RSL rise exceeding 3.4 mm/yr were virtually certain (at least 99% probability) to have occurred between 15.25 kyr and 13.5 kyr BP and between 9.5 kyr and 8.25 kyr BP. During the last ~ 5 kyr, it is unlikely (less than 33% probability) rates of RSL rise exceeded 3.4 mm/yr.
Table 2
Future projections of relative sea-level (RSL) rise to 2150 for Singapore. Magnitudes and rates of RSL rise to 2150 for Singapore showing 50th percentile and likely (17th − 83rd percentile) ranges for Shared Socioeconomic Pathways (SSP) 1-1.9, SSP2-4.5, SSP5-8.5 and SSP5-8.5 low confidence (LC) scenarios. Projections are relative to a baseline period of 1995–2014 for the Sultan Shoal tide gauge station4,33 (Fig. 1b).
SSP | Year | Magnitude m | Rate mm/yr |
1-1.9 | 2030 | 0.10 (0.07–0.14) | 4.3 (2.8–6.3) |
| 2040 | 0.14 (0.09–0.19) | 4.2 (2.7–6.4) |
| 2050 | 0.19 (0.13–0.26) | 4.1 (2.3–6.6) |
| 2060 | 0.22 (0.15–0.32) | 4.0 (2.0–6.6) |
| 2070 | 0.27 (0.18–0.40) | 4.2 (2.1–6.8) |
| 2080 | 0.31 (0.20–0.46) | 4.2 (2.2–7.0) |
| 2090 | 0.36 (0.23–0.54) | 3.9 (1.7–6.8) |
| 2100 | 0.38 (0.23–0.59) | 3.8 (1.5–6.7) |
| 2110 | 0.43 (0.25–0.67) | 4.0 (1.8–6.8) |
| 2120 | 0.47 (0.27–0.74) | 3.8 (1.5–6.6) |
| 2130 | 0.51 (0.29–0.8) | 3.6 (1.4–6.4) |
| 2140 | 0.54 (0.30–0.86) | 3.4 (1.3–6.3) |
| 2150 | 0.58 (0.31–0.93) | 3.4 (1.2–6.2) |
2-4.5 | 2030 | 0.09 (0.06–0.13) | 4.3 (2.9–6.2) |
| 2040 | 0.14 (0.09–0.20) | 5.1 (3.4–7.4) |
| 2050 | 0.20 (0.14–0.29) | 5.8 (4.0–8.5) |
| 2060 | 0.26 (0.19–0.37) | 6.6 (4.4–9.7) |
| 2070 | 0.33 (0.24–0.47) | 7.1 (4.6–10.6) |
| 2080 | 0.41 (0.29–0.58) | 7.4 (4.8–11.4) |
| 2090 | 0.48 (0.34–0.69) | 8.1 (5.2–12.4) |
| 2100 | 0.57 (0.40–0.81) | 8.2 (5.2–12.4) |
| 2110 | 0.65 (0.43–0.95) | 7.8 (4.8–12.1) |
| 2120 | 0.72(0.48–1.07) | 7.7 (4.7–12.0) |
| 2130 | 0.80 (0.52–1.19) | 7.6 (4.6–11.9) |
| 2140 | 0.88 (0.57–1.31) | 7.5 (4.5–11.8) |
| 2150 | 0.95 (0.62–1.42) | 7.3 (4.4–11.6) |
5-8.5 | 2030 | 0.10 (0.06–0.14) | 5.1 (3.6–7.0) |
| 2040 | 0.16 (0.11–0.22) | 6.0 (4.2–8.4) |
| 2050 | 0.23 (0.17–0.32) | 7.0 (4.9–9.9) |
| 2060 | 0.31 (0.23–0.42) | 8.4 (5.9–12.0) |
| 2070 | 0.40 (0.30–0.56) | 10.0 (7.0–14.4) |
| 2080 | 0.51 (0.38–0.71) | 11.3 (7.8–16.6) |
| 2090 | 0.64 (0.49–0.88) | 12.9 (8.7–19.4) |
| 2100 | 0.79 (0.60–1.09) | 13.5 (9.0–20.4) |
| 2110 | 0.87 (0.61–1.28) | 13 (8.7–19.4) |
| 2120 | 1.02 (0.70–1.48) | 12.6 (8.4–19.2) |
| 2130 | 1.14 (0.79–1.66) | 12.2 (7.8–18.7) |
| 2140 | 1.26 (0.87–1.84) | 11.6 (7.1–18.2) |
| 2150 | 1.37 (0.94–2.02) | 11.0 (6.6–17.6) |
5–8.5 LC | 2030 | 0.10 (0.06–0.16) | 5.3 (3.6–10.2) |
| 2040 | 0.16 (0.11–0.28) | 6.5 (4.2–13.3) |
| 2050 | 0.24 (0.16–0.42) | 7.8 (4.9–17.1) |
| 2060 | 0.33 (0.22–0.60) | 9.6 (5.9–21.8) |
| 2070 | 0.44 (0.30–0.83) | 11.8 (7.0–26.4) |
| 2080 | 0.57 (0.38–1.10) | 14.2 (7.8–29.9) |
| 2090 | 0.73 (0.49–1.40) | 17.1 (8.7–33.0) |
| 2100 | 0.92 (0.60–1.74) | 19.3 (9.0–41.7) |
| 2110 | 1.10 (0.61–2.05) | 20.7 (8.7–53.3) |
| 2120 | 1.32 (0.70–2.36) | 22.6 (8.4–70.0) |
| 2130 | 1.55 (0.79–3.09) | 25.0 (7.8–93.0) |
| 2140 | 1.81 (0.87–4.14) | 27.5 (7.1–109.4) |
| 2150 | 2.11 (0.94–5.28) | 29.2 (6.6–111.2) |
Under the moderate emissions SSP2-4.5 scenario, future RSL in Singapore will increase 0.95 m (0.62–1.4 m) at a rate of 7.3 mm/yr (4.4–11.6 mm/yr) by 2150 (Fig. 6a, b, Table 2). Magnitude 0.5 m and 1.0 m thresholds under SSP2-4.5 are expected to be surpassed by 2092 (2072–2125) and 2157 (2114–2264), respectively. Rates of RSL rise exceeding 7.3 mm/yr were about as likely as not between ~ 20 kyr and ~ 19.5 kyr BP and unlikely up to 15.75 kyr BP (Fig. 7b). Rates of RSL rise exceeding 7.3 mm/yr were virtually certain between 15 kyr and 14 kyr BP and between 9.25 kyr and 8.75 kyr BP. During the last ~ 8 kyr, it is unlikely rates of RSL rise exceeded 7.3 mm/yr.
Under the very high emissions SSP5-8.5 scenario, future RSL rise in Singapore will increase 1.37 m (0.94–2.02 m) at a rate of 11 mm/yr (6.6–17.6 mm/yr) by 2150 (Fig. 6a, b, Table 2). Magnitude 0.5 m and 1.0 m thresholds under SSP5-8.5 are expected to be surpassed by 2080 (2066–2097) and 2119 (2096–2159), respectively. Rates of RSL rise exceeding 11 mm/yr were very unlikely (less than 10% probability) between 21.5 kyr and 15.75 kyr BP while it very likely occurred between 15.25 kyr and 14.75 kyr BP and virtually certain at ~ 9 kyr BP (Fig. 7c). During the last ~ 8.25 kyr, it is extremely unlikely (less than 5% probability) rates of RSL rise exceeded 11 mm/yr.
Considering ice-sheet processes in which there is currently low confidence in the scientific ability to model raises the potential sea-level rise contributions, particularly under high emissions scenarios. The 83rd percentile projection for SSP5-8.5 including low confidence MICI processes reaches 5.3 m and 111 mm/yr by 2150 (Table 2) and moves the crossing of a 1.0 m magnitude threshold as soon as 2076 (Fig. 6a). Such high rates of RSL rise have no precedent in the last ~ 21.5 kyr.