The Indonesian seas occupy a unique position in the Earth’s climate engine. Warm, low-salinity water from the tropical Pacific flows through the narrow straits and deep channels of the Malay Archipelago and spills into the Indian Ocean, a movement known as the Indonesian Throughflow. This interbasin exchange modulates the global thermohaline circulation, redistributes heat between ocean basins, and influences everything from Indian Ocean dipole events to the pace of global surface warming. Yet the observational network in the region remains thin, and before the intensive INSTANT mooring program of 2004 to 2006 and subsequent Makassar Strait moorings, sustained measurements of throughflow transport and upper-ocean structure were essentially absent. That observational gap has made it extraordinarily difficult to determine whether recent changes in the region are part of natural multi-decadal variability or something new.

The research team’s solution was to turn to corals. Massive coral colonies growing in the main pathways of the Indonesian Throughflow, including sites in the Makassar Strait and near Lombok Strait, incorporate oxygen isotopes into their calcium carbonate skeletons in proportion to the temperature and isotopic composition of the seawater in which they grow. Because coral δ18O responds to both sea surface temperature and sea surface salinity, these skeletons act as natural recorders of the upper-ocean environment, layer by layer, year by year, extending back well before any instrument touched the water. Paired with independent coral Sr/Ca measurements, which isolate the temperature signal, the δ18O records can be deconvolved to reconstruct past salinity variations with remarkable fidelity. The team synthesized coral δ18O records from sites directly within the throughflow pathways, drawing on compilations such as the CoralHydro2k database of tropical hydroclimate proxies.

To interpret what the corals were recording, the researchers turned to eddy-active ocean model simulations performed with the NEMO modeling framework at the North-German Supercomputing Alliance and the Computing Centre at Kiel University. These simulations were driven by the JRA55-do atmospheric reanalysis dataset from the Japan Meteorological Agency, providing realistic surface forcing of winds, heat fluxes, and freshwater inputs. Crucially, the model’s simulated upper-ocean temperature and salinity variability showed consistency with the limited observational data available near the coral sites, including the Labani Channel mooring observations in Makassar Strait. This validation step is what makes the proxy-model synthesis approach credible: if the model reproduces the observed variability where observations exist, then the model can be trusted to fill in the spatial and temporal gaps where corals and instruments cannot.

The comparison between coral records and model output yielded a striking result. On multi-decadal timescales, coral δ18O co-varies with simulated salinity and with the depth of the thermocline, the boundary layer separating warm surface waters from cooler deep waters. This coupling means the corals are not merely recording surface conditions but are sensitive to the vertical structure of the upper ocean, including how deep the warm layer extends. Thermocline depth in the Indonesian seas is controlled by a delicate balance of wind-driven processes in both the Pacific and Indian Oceans, monsoon dynamics, and buoyancy fluxes from rainfall and evaporation. The fact that a surface-dwelling organism’s skeleton tracks this subsurface structure opens a window onto vertical ocean variability that no other proxy archive in the region can provide.

To disentangle the physical drivers, the team ran model sensitivity experiments that separated wind forcing from buoyancy forcing, allowing them to attribute simulated variability to each mechanism independently. The verdict was clear: wind forcing dominates the multi-decadal variability of upper-ocean temperature and salinity structure in the Indonesian seas, with buoyancy forcing, driven by changes in rainfall, evaporation, and surface heat fluxes, contributing episodically throughout the twentieth century. Winds exert their influence by piling up or drawing down water across the basin, modulating the pressure gradient that drives the throughflow and adjusting the depth of the thermocline through Ekman pumping and planetary wave dynamics. This finding aligns with earlier work showing that Pacific trade wind variations, tied to the Interdecadal Pacific Oscillation, exert remote control over Indian Ocean heat content and sea level.

Indeed, the Interdecadal Pacific Oscillation emerges as a central character in this story. Both the coral-reconstructed and simulated variability show sensitivity to the phase changes of this basin-scale climate mode, which alternates between states resembling persistent El Niño and persistent La Niña conditions over the Pacific. When the IPO shifts phase, the winds over the tropical Pacific and the Maritime Continent reorganize, and the consequences propagate through the Indonesian seas as changes in throughflow transport, surface salinity, and thermocline depth. The study highlights that these IPO-related impacts extend beyond the well-documented interannual ENSO teleconnections into the realm of multi-decadal modulation of the regional thermohaline vertical structure, a dimension of variability that short observational records simply cannot capture.

Perhaps the most consequential finding is that the recent multi-decadal modulation of wind-driven surface freshwater operates through fundamentally different mechanisms than those acting at interannual timescales. In other words, the processes that make the Indonesian seas fresher or saltier from one decade to the next are not simply scaled-up versions of the processes that drive year-to-year swings. This non-stationarity has profound implications for how scientists interpret both modern observations and paleoclimate proxies, and for how climate models should be evaluated against the region’s hydrological evolution. As the Indo-Pacific freshwater pool expands under global warming and the global water cycle intensifies, understanding which mechanisms govern freshwater delivery through the Indonesian seas becomes essential for projecting future changes in interbasin heat and salt exchange.

The study’s methodological achievement deserves emphasis as much as its physical findings. By demonstrating that coral δ18O records from throughflow pathways can be meaningfully compared against eddy-resolving model simulations, and that the two archives agree where they overlap, the team has established a template for future proxy-model syntheses in data-sparse ocean regions. The approach transforms corals from isolated curiosity into quantitative constraint, and transforms models from tools of the present into instruments for interrogating the past. As the authors note, future proxy-model comparison will be critical for evaluating whether the drivers of variability they identified persist or shift as Indo-Pacific heat and freshwater exchange continues to evolve far beyond the reach of the short observational record.

For a region that regulates the flow of heat between two of the world’s largest ocean basins, the Indonesian seas have long been under-observed and under-appreciated in the public imagination. This work changes that. It shows that the skeletons of corals growing in Indonesian waters carry a legible, century-scale record of how winds and rains have reshaped the ocean’s vertical architecture, and that state-of-the-art ocean models can read that record and extend it. As climate change accelerates the intensification of the global water cycle and reshapes Pacific wind systems, the multi-decadal memory preserved in coral reefs and reproduced in simulations offers one of the few reliable guides to what the Indonesian Throughflow has done, and what it may do next, in the service of the planet’s climate.

Subject of Research: Multi-decadal upper-ocean temperature and salinity variability in the Indonesian seas reconstructed from coral δ18O records and ocean model simulations

Article Title: Drivers of multi-decadal upper-ocean thermal and haline structure in the Indonesian seas: a synthesis of coral δ18O and ocean models

Article References: Murty, S. A., Ummenhofer, C. C., Wang, S., Gruenburg, L., Sprintall, J., Biastoch, A., & Böning, C. W. (2026). Drivers of multi-decadal upper-ocean thermal and haline structure in the Indonesian seas: a synthesis of coral δ18O and ocean models. Climate Dynamics, 64(10), Article 440. https://doi.org/10.1007/s00382-026-08394-8

Image Credits: AI Generated

DOI: 10.1007/s00382-026-08394-8

Keywords: Indonesian Throughflow, coral δ18O, paleoceanography, ocean modeling, Interdecadal Pacific Oscillation, thermocline, sea surface salinity, Makassar Strait, climate variability, NEMO model, freshwater flux, Indo-Pacific climate

The research team, led by Taufan Wiguna of the Bandung Institute of Technology and the National Research and Innovation Agency of Indonesia, together with colleagues from IPB University, focused on two elegant diagnostic variables. The first is the depth of the 26.5 degree Celsius isotherm, abbreviated T26.5, which marks the top of the thermocline where warm surface water gives way to cooler water below. The second is the depth of the 34.4 practical salinity unit isohaline, or S34.4, a boundary in the salinity profile that separates fresher near-surface water from saltier water masses of Pacific origin. When these boundaries shoal, or move closer to the surface, cooler and saltier water is brought within reach of wind-driven mixing and upwelling; when they deepen, that supply is cut off.

To track these boundaries through time, the team used the GLORYS12V1 global ocean physics reanalysis from the Copernicus Marine Service, a state-of-the-art dataset that blends ocean model simulations with observations at a horizontal resolution of one-twelfth of a degree. Monthly anomalies of T26.5 and S34.4 were computed relative to the corresponding monthly climatological means over the full 1993 to 2025 analysis period, a step that removes the seasonal cycle and isolates the low-frequency signal of interest. The resulting anomaly fields were then passed through a 13-year low-pass filter, a mathematical treatment that suppresses anything oscillating faster than the interdecadal band and allows the slow IPO signal to emerge cleanly from the noise of El Niño, the monsoon, and other shorter-lived phenomena.

The IPO itself was quantified using the Tripole Index, derived from the NOAA Extended Reconstructed Sea Surface Temperature version 5 dataset and maintained by the NOAA Physical Sciences Laboratory. This index captures the characteristic seesaw pattern of the IPO, in which sea surface temperature anomalies in the central tropical Pacific vary out of phase with those in the northwestern and southwestern Pacific. By correlating the filtered IPO index with the filtered isotherm and isohaline depth anomalies, and by examining lagged cross-correlations to test whether the ocean response follows the climate index in time, the researchers could map where and how strongly the interdecadal pulse penetrates into the Maluku Sea.

The results are strikingly consistent. Across the regions analyzed, the correlation between the IPO and the depth of the 26.5 degree Celsius isotherm ranged from negative 0.73 to negative 0.83, while the correlation with the 34.4 PSU isohaline depth ranged from negative 0.57 to negative 0.72. These are strong relationships by the standards of ocean-climate teleconnection studies, and the negative sign carries a clear physical meaning: when the IPO is in a positive phase, the thermocline and halocline in the Maluku Sea tend to shoal, bringing cooler, saltier water upward and creating conditions favorable for upwelling. In the negative phase, the boundaries deepen and the upwelling-favorable configuration relaxes. Notably, the temperature boundary responded more coherently than the salinity boundary, suggesting that heat content in the upper ocean is the more faithful recorder of the interdecadal signal in this basin.

What makes the study particularly valuable is its honesty about what the IPO does not control. The team also examined horizontal divergence, a dynamical quantity related to the vertical stretching of surface waters, and satellite-derived chlorophyll-a, a proxy for phytoplankton biomass that often blooms where nutrient-rich water upwells. Here the correlations were weaker and spatially patchy, and phase composites, which average conditions during positive and negative IPO states separately, revealed far less consistent differences. In other words, the clearest expression of low-frequency IPO variability in the Maluku Sea is written in the thermohaline structure itself, not necessarily in the biological response at the surface. This distinction matters for anyone hoping to predict fisheries productivity from climate indices alone, because a shoaling thermocline does not automatically translate into a chlorophyll bloom.

To probe the vertical dimension of the response, the researchers turned to monthly vertical velocity fields from the ECCO2 Cube92 data synthesis, accessed through the Asia-Pacific Data-Research Center. Composites of vertical motion during different IPO phases revealed phase-dependent patterns of upward and downward movement that differed among subregions of the sea and among depth ranges. The upwelling system of the Maluku Sea, it turns out, does not respond to the interdecadal forcing as a single coherent block. Instead, the vertical-motion response is modulated by local circulation features, including the intermediate western boundary current that recent moored observations and numerical modeling have shown connects Pacific circulation to the Indonesian Throughflow through this very basin.

The Maluku Sea occupies a special position in the global ocean conveyor. It is one of the eastern gateways of the Indonesian Throughflow, receiving North Pacific water masses through passages between Halmahera and Sulawesi and exporting them toward the Banda Sea and ultimately the Indian Ocean. Because the thermocline waters carried by the Throughflow help set the temperature and salinity of the Indian Ocean’s upper layers, decadal changes in the Maluku Sea’s stratification can ripple far beyond Indonesian waters. Previous work has documented warming trends in the Indonesian seas and interannual to decadal shifts in the Throughflow’s vertical profile, but the specific link between the IPO and the Maluku Sea’s thermohaline boundaries had remained largely unquantified until now.

Methodologically, the study is a careful exercise in teleconnection analysis. The choice of a 13-year low-pass filter is well matched to the dominant IPO timescale, and the use of lagged cross-correlation acknowledges a well-known pitfall in climate science: correlated signals do not by themselves establish causation, and spurious correlations can arise when two filtered time series share energy in overlapping frequency bands. By testing multiple lags and by comparing the strong thermohaline response against the weaker divergence and chlorophyll responses, the authors build a case that the IPO-isotherm relationship is physically meaningful rather than a statistical artifact. The phase composites add a complementary line of evidence, showing that the distribution of T26.5 and S34.4 anomalies genuinely differs between IPO regimes rather than merely correlating with them in a linear sense.

The practical implications extend from climate science into resource management. The Banggai upwelling region in the southern Maluku Sea supports locally important fisheries, and interdecadal shifts in thermocline depth could modulate the nutrient supply that underpins the marine food web over timescales relevant to management planning. Because the IPO swings slowly, its state provides a degree of predictability that faster modes such as ENSO cannot offer at decadal horizons. The datasets underpinning the study, including the GLORYS12 reanalysis, the Copernicus biogeochemistry hindcast, the NOAA Tripole Index, and the ECCO2 vertical velocities, are all publicly available, and the derived regional time series can be requested from the corresponding author. Funded by Indonesia’s National Research and Innovation Agency and Institut Teknologi Bandung, the work exemplifies how open reanalysis products can illuminate climate dynamics in data-sparse seas. As the Pacific potentially transitions between IPO phases in the coming decades, the thermohaline boundaries of the Maluku Sea will continue to rise and fall in response, a slow tide of climate change written not in sea level but in the depth of an isotherm.

Subject of Research: The influence of the Interdecadal Pacific Oscillation on upper-ocean thermohaline structure and upwelling variability in the Maluku Sea, Indonesia

Article Title: Interdecadal pacific oscillation and upwelling in the Maluku Sea, Indonesia: Evidence from isotherm and isohaline variability

Article References: Wiguna, T., Putri, M. R., Rachmayani, R., Zuraida, R., & Atmadipoera, A. S. (2026). Interdecadal pacific oscillation and upwelling in the Maluku Sea, Indonesia: Evidence from isotherm and isohaline variability. Ocean Dynamics, 76(10), Article 100. https://doi.org/10.1007/s10236-026-01854-z

Image Credits: AI Generated

DOI: 10.1007/s10236-026-01854-z

Keywords: Maluku Sea, Interdecadal Pacific Oscillation, upwelling, thermocline, isotherm depth, isohaline depth, Indonesian Throughflow, GLORYS12 reanalysis, ocean dynamics, climate variability, chlorophyll-a, Indonesia