Introduction

Seeps of liquid (oil) and gaseous hydrocarbons (mainly methane, CH4, ethane, C2H6 and propane, C3H8) are common in sedimentary basins associated with active petroleum systems1,2,3,4. Hydrocarbon generation and surface release may also occur at sediment-hosted geothermal systems (SHGS) where organic matter (OM) maturation is triggered and enhanced by the heat flow induced by magma and hydrothermal fluids intruding or occurring below the sedimentary deposits5 and refs. therein. Tectonic discontinuities (faults), rupture of seals, and erosional processes promote hydrocarbon migration to the surface2,3. Surface hydrocarbon discharges can manifest as focused gas seeps, oil seeps, mud volcanoes, pockmarks (in underwater environments), and diffuse microseepage4,6,7,8,9,10. Recent inventories document 3439 onshore seepage sites and more than 30 zones of submarine seepage in coastal regions and shallow seas, which may affect the atmosphere11,12. Rough estimates suggest that ~10,000 seeps exist globally13.

Natural offshore petroleum seeps represent nearly half of the oil entering the marine environment worldwide, while 15% are related to petroleum extraction and transportation, and 37% to its consumption (i.e. land-based run-off, operational discharges from commercial vessels and recreational craft, and atmospheric deposition)14,15. The significant amounts of oil released in marine environments represent a threat to the ecosystems by increasing water toxicity, thus affecting all trophic levels of the food chain16. Studies monitoring oil discharge in the aquatic ecosystems are limited to a few long-lasting offshore oil seep areas (e.g. California and Mexican Gulf), and several large oil spills (Exxon Valdez, Deep Water Horizon, North Cape, Panama) that took place during the last few decades15,17.

Global emissions to the atmosphere of methane, ethane, and propane from seepage are estimated to be in the range of 45 (27–63) Mt yr−118;, 2–4 Mt yr−1 and 1–2.4 Mt yr−119;, respectively. While methane is a major greenhouse gas, ethane and propane are photochemical pollutants, contributing to surface ozone (O3) production with an effect considered 10 times stronger than that of CO214,20,21. Specifically, ethane and propane degradation in the atmosphere can lead to O3 production in the troposphere via oxidation by OH radicals and reactions with nitrogen oxides22. Anthropogenic and natural sources of ethane and propane are, however, poorly constrained21, and the preliminary emission factors of natural seepage19 need to be refined.

Here, we present the first assessment of oil, ethane, and propane emission from one of the largest natural seepage sites on Earth. The site is named Lusi (NE Java, Indonesia), famous for its mud eruption that started in 200623. Lusi is considered a SHGS24 and the world’s largest active mud eruption (Fig. 1A–D)23. Lusi is located in the East Java petroleum basin, and since 2006 continuously releases copious amounts of gas into the atmosphere, as well as mud, rocks, and oil, that are ultimately discharged offshore (Fig. 1E, F). Previous ground-based and satellite measurements revealed that Lusi is a mega emitter of thermogenic methane, releasing ~0.1 Mt CH4 per year25. We have characterized and quantified the so far unknown amounts of mud and oil released from this large eruption site. A 13-years database of monitored mud flow rates, acquired using various approaches, is combined with geochemical analyses from mud and gas samples collected from the Lusi site during several field trips from 2006 to 2017. Ethane and propane output and emission factors have been estimated based on the methane flux and molecular composition of hydrocarbon gases. The oil, ethane and propane emissions are then compared to major anthropogenic hydrocarbon leaks.

Fig. 1: Lusi seepage views.
figure 1

Google Earth view of the study area in 2006 (A) and 2012 (B). The embankment is highlighted with a red line, black stains in the hydrothermal pond represent the oil slicks. The morphology of the Lusi crater changed during the first ~ 13 years of activity. Map of eastern Java Island in the inset. Drone photo of the Lusi crater in 2006 (C), with an example of the berm construction to contain the mud flow, and in 2013 (D). During gradual subsidence, the berm disappeared, submerged by mud and the crater site remained isolated and not accessible by foot. E, F Discharge of the Lusi mud in the Porong River, April 2017 (see (A) for location).

Full size image

The Lusi seepage system in a nutshell

The Lusi (Lumpur—mud in Indonesian, Si—Sidoarjo region) seepage system is located in the East Java back-arc sedimentary basin, a few kilometers from the volcanic arc related to the subduction of the Indian-Australian plate under the Sunda plate. Two producing hydrocarbon fields (Wunut and Tanggulangin) surround the eruption26 and refs therein. The reservoir intervals are from the Pleistocene Pucangan Formation at 200 to 1000 m depth27,28. The deepest and shallowest reservoirs are filled with oil and gas, respectively, and have been producing since the late 1990-s28. Reservoir formations consist of 3-47 m thick poorly consolidated sandstones and sands intercalated with shales28. Two Holocene volcanoes (Penanggungan and Arjuno-Welirang) are situated 10 and 25 km SW of Lusi that is also intersected by the Watukosek fault system zone which extends from the Arjuno-Welirang volcanic complex29,30. This fault system facilitates the discharge of fluids to the surface31,32,33. In fact, even before the Lusi eruption, this area was already characterized by diffused surface gas seepage manifestations. Additionally, mud volcanism is documented in the northern part of the island30. The sedimentary section at this locality comprises ~5 km of Paleogene-Quaternary sediments (thorough description provided as SOM). Since its birth, Lusi featured a geysering behavior, with regular bubbling activity followed by clastic blowouts, mud bursts, intense vapor release, and quiescent phases34. To date, the erupted mud covers ~7 km2 and the area is entirely confined by an ~11 m tall embankment. The activity of this system is driven by hydrocarbon and CO2 generation occurring at ~4.5 km depth due to magma emplacement and hydrothermal fluids migration in the organic-rich deposits of the Ngimbang Formation24,26,31,35. These actively-generated fluids created overpressure, facilitating the Lusi onset. The subsurface plumbing system and the high surface temperature (~100 °C) of the discharged fluids (i.e. mantle-derived and sedimentary) result from the interaction between sedimentary and magmatic domains, defining Lusi as a sediment-hosted hydrothermal system5,24,32,35,36,37. Geochemical analyses of erupted gas reveal formation and equilibrium temperatures for the CO2 and the CH4 up to 400 °C38, while oils and clasts analyses indicate that the organic-rich Ngimbang Formation (at ~4 km depth) was exposed to temperatures up to 300 °C during the Holocene and that hydrocarbon generation is currently ongoing26,35. The Lusi fluids (gas and oil) and those from the hydrocarbon fields located in the neighboring regions have significant geochemical differences in terms of δ13C-CH4 and oil maturity, indicating that these are two independent systems26,38,39.

Results and discussion

Mud breccia discharge

During the survey period of May 2006 to January 2019, we carried out 805 mud flow rate measurements (Fig. 2A, Table S1). Despite the high variability in short-term scales (days to months), long-term trends are evident in the time series. The highest flow rates occur four months after the inception of the eruption, reaching 180,000 m3/day in September 2006.

Fig. 2: Mud flow monitoring and analyses.
figure 2

A Daily mud flow rate measurements (orange), their interpolation (purple), and measured hydrocarbon content in the dried sediment. During ~ 13 years of Lusi activity, 805 measurements were performed between 2006 and 2017. B Cross-plot of the saturated fraction and the amount of the Extracted Organic Matter (EOM) shows a linear correlation of both parameters indicating gradually increasing oil content from the Ngimbang Fm. Note: the sample collected in 2016 has undergone biodegradation processes during the storage at room temperature. C Biomarker distribution in the EOM from the bulk dry Lusi mud. C29 sterane isomer ratios cross-plot indicates low maturity of the EOM increasing from 2007 to 2017 and forming two maturity clusters (2006–2011 and 2012–2017).

Full size image

The calculated average mud flow rate is 64,450 m3/day, and data interpolation yields a total discharge of ~0.3 km3 of mud breccia. In addition, calculations based on 3D deep electrical resistivity tomography reveal the presence of nearly 0.2 km3 of mud breccia placed below the active vents. This volume represents the collapsed region that should be included in the global Lusi budgets40. This is comparable to the estimated erupted volumes of the Touragay mud volcano in Azerbaijan (0.75 km3), a 400 m tall mountain with a ~500 m wide crater and a total diameter of ~7.5 km2 which is considered the largest known onshore mud volcano on Earth6.

Gas and mud samples were collected between 2006 to 2017. The term ‘mud’ used herein refers to the erupted material that comprises water and solid components, predominantly sandy-silty clay. To include potential sediment content variations through time, an average value of 37 vol.% ±10 was used for oil discharge calculations (see Data and Methods for details). The mercury concentration in the sediment is 33.5 ppb ( ± 1.7).

Oil hydrocarbon composition and origin

The amounts of Extracted Organic Matter (EOM) in the 12 mud samples analyzed vary from 0.7 to 2.0 mg/gsediment (Table 1, Fig. 2A). The compound class distribution changes through time: during the first 5 years of Lusi activity (2006–2011), EOM contains 51–59% saturated hydrocarbons, 5–9% aromatic hydrocarbons, and 35–41% polar compounds. During the following years, the amounts of saturated hydrocarbons increase to 58–79%, with aromatic and polar compounds representing 4-12% and 16-30%, respectively (Fig. 2B, S1, Table S2).

Table 1 Mixing model
Full size table

Previous stratigraphy, petrography, and geochemistry studies of the sediments intersected by the Lusi conduit24,26,41 identified three shale formations as potential sources for the organic matter extracted from mud at Lusi: Eocene-Oligocene black shale deposits of the Ngimbang (~3800–5000 m), the marly shales of Miocene Tuban (~2830 to ~3250 m) and grey shales of the Pleistocene Up. Kalibeng (900–1870 m) Formations. The Ngimbang Fm. black shale clasts at the Lusi surface display high oil generation potentials, with TOC and S2 values reaching 14.6 wt.% and 27.9 mg/g, respectively26. By contrast, Miocene and Pleistocene grey shales have TOCs lower than 2 wt.% and poor generation potentials, with S2 ≤ 1.23 mg/g26. The predominance of heavy n-alkanes (n-C23 − n-C27) in the EOM of ten samples is typical of terrestrial OM (Fig. S2). High Pristane/Phytane ratios (Pr/Ph=2.3–3.8) and dominant C29 over C27 and C28 steranes are also diagnostic of terrestrial OM (Fig. S3, Table S3). Such a terrigenous signature is characteristic of the Ngimbang Fm. Altogether, the Ngimbang Fm. is demonstrably the main regional oil source28 and the most likely source of the Lusi oil films26.

Maturity-related biomarker parameters such as C29 sterane isomer ratios (ββ/(ββ + αα) and 20S/(20S + 20 R)) or re-arranged/regular hopanes Ts/Tm (Fig. 2C, Table S3) show that initially predominantly immature EOM ( < 0.6% vitrinite reflectance equivalent (%Ro)42,43) becomes oil-window mature with time, with a sharp rise around 2012 (Fig. S4). By comparison, diesel-range aromatic hydrocarbon indicators (phenanthrenes, and methyl-dibenzothiophenes) such as MPI-1, MPR and MDR44,45 have a maturity increasing from mid to late oil-window since 2006. The rise in EOM maturity could involve either a change in mud sources or an increasing proportion of oil mixed in the mud that contains immature OM. The co-occurrence of immature biomarkers and high maturity aromatics argues in favor of oil mixed with immature bitumen46. Moreover, biomarkers are abundant in both immature and mature OM (or oil), whereas diesel-range aromatics only become abundant in mature OM/oil. Initially, immature biomarkers represent a non-negligible portion of the total biomarker content of the EOM, but oil-derived diesel-range hydrocarbons already dominate this class of compounds (which is not significant in immature OM). With time, the maturity of diesel-range aromatics increases still (~ late oil-window) as mature biomarkers become predominant, consistent with increasing amounts of oil mixed with the mud immature OM. The rise in the proportion of saturated hydrocarbons in the EOM from 2006 to 2017 also agrees with this hypothesis.

Indications that the Up. Kalibeng Fm. is the main immature bitumen source include: 1) shallow grey shales of the Up. Kalibeng Fm. are the dominant source of the mud erupted at Lusi23,27; 2) a low organic-richness, with maximum EOM yields ~ 0.27 mg HC/grock26; and 3) low contents of saturated hydrocarbons in EOM from Up. Kalibeng Fm. clasts (Table S2)26. The 13C-enrichment of the saturated fraction of the EOM further denotes a high fraction of hydrocarbons from the isotopically heavier Ngimbang Fm. compared to more 13C-depleted Kalibeng Fm. compounds, as indicated by δ13C values of about −23.5 and −24.5‰VPDB, respectively, for black and grey shale clasts at the Lusi surface (Fig. S5, Table S4). While the Ngimbang Fm. is generally isotopically lighter in other parts of the vast East Java Basin than in the black shale clasts analyzed, 13C-rich oils are known to occur in the basin, e.g. in the SE Kangean Field where the oils have a restricted, mostly lacustrine, Ngimbang Fm. source47. Furthermore, Ngimbang-sourced oils in the Porong and Wunut fields (neighboring the Lusi site) are also isotopically heavy, supporting the hypothesis of a local 13C-rich Ngimbang Fm. source interval47.

The potential amount of oil in the EOM is calculated assuming a maximum immature OM content of 0.27 mgHC/grock and subtracting it from each total EOM measurement. Immature bitumen could potentially originate from both Up. Kalibeng and Tuban formations, but the latter is composed of consolidated marl and marly shales which are quantitatively negligible sources for the fluidized mud and are furthermore as organic-poor as the Up. Kalibeng Fm26. The calculation shows proportions of migrated oil increasing from 63% to 86% between 2006 and 2017 (mixing model, Table 1).

Total oil discharge

The total amount of oil hydrocarbons represented by the total EOM in the mud during ~13 years of Lusi activity was calculated using Eq. (1) and equals to 0.20 Mt (min), 0.28 Mt (medium), and 0.36 Mt (max) for 27 vol.%, 37 vol.%, and 47 vol.% of sediment content, respectively (Table 2a). Based on the geochemical evidence and the applied mixing model, the proportion of oil sourced from the Ngimbang Fm. in the total EOM (Table 1) can be estimated at 0.16–0.28 Mt (Table 2a).

Table 2 Estimates of extracted organic matter (EOM) and oil amount
Full size table

This assessment is based on three main parameters (daily mud flow rate, sediment content of the mud, i.e. dry bulk mud, and EOM content) combined with the described mixing model. The variability of the total oil discharge estimates is mainly imputable to difficulties in sediment content measurements. The sediment content measured in mud samples collected in 2006, 2007 and 2010 is 37 ± 1.5 vol.%, but the active vent ceased to be accessible for sampling after 2010 due to the continuously changing morphology of the eruption site. Nevertheless, our field observations suggest that the water proportion has decreased through time23,48 and that the average sediment content of 37 vol. % may be underestimated. Based on these field observations, a more reasonable sediment content value of 47 vol.% is proposed for the ~ 13 years of the study. We therefore suggest that an oil discharge of 0.22–0.28 Mt may represent a more likely estimate for the period of 2006 to 2017 (Table 2a).

Ethane and propane discharge

The amounts of CH4 and CO2 released from Lusi were estimated to be ~0.1 Mt year−1, and ~0.6 Mt year−1, respectively25. These estimates suggest that Lusi is the largest geological methane emitter documented on Earth so far. The CO2 emission is comparable with other volcanic complexes or with large tectonic and hydrothermal/geothermal systems49. Based on the methane emission data and the molecular composition of the gas sampled during different seasons (see Data and Methods), mean ethane emission from Lusi results to be 7551 (5583-9991) ton yr-1 from ground-based methane flux data, and 10,385 (3932-16,839) ton yr-1 from satellite C1 flux data. The propane emission results to be 5612 (4099-7486) ton yr-1 from ground-based methane flux data, and 7700 (2915-12,485) ton yr-1 from satellite C1 flux data (Table S5).

In the gas released from the crater and surrounding seeps, which are the dominant gas emission systems of Lusi, ethane and propane concentrations, and thus the C2/C1 and C3/C1 ratios (0.012 and 0.0045, respectively) are higher than the global average for mud volcanoes and seeps19. This is due to the thermogenic nature of the gas, which was formed in deep and highly mature source rocks, with vitrinite reflectance > 2%Ro, as estimated using the gas formation modeling in Mazzini et al.24 and measured in the collected source rock clasts at the crater surface5,26. The high concentration of alkanes heavier than methane is typical of oil-bearing petroleum systems and enhanced geothermal gradients, such as those of SHGS like Lusi5,24 can shift the “oil window”, with heavy HC and oil production, to shallower depths.

Our new data show that the Lusi emission factors for ethane and propane (1286 t km−2 yr−1 and 943 t km-2 yr−1, respectively) are at least one order of magnitude higher than those from the Barnett shale region (Texas, U.S.; ethane: 1.44 t km−2 yr−150;) and La Brea tar pits of Los Angeles (California, U.S.; ethane: 35 t km-2 yr−1, propane: 17 t km−2 yr−151;) (Table S6). In terms of total gas emission, Lusi has already injected into the atmosphere more than 20 times the amount of ethane released by the Aliso Canyon gas storage blowout that occurred in 2015 in California, which has been considered as the largest gas leak in U.S. history (mean of 3.9 ton/h over 112 days and total leak of 7300 tonnes of ethane52,53). Lusi ethane emission is, then, equivalent to about 10% of the emission from the fossil fuel industry for all Southern Indonesia and Australia sector54,55.

Contribution to the global oil seeps budget

To our knowledge, there are no global estimates of the oil volume released from onshore oil seeps. Offshore natural oil seeps release ~0.6 Mt of oil into the marine environment every year56. Among the various natural and anthropogenic oil discharges (Table 2b), we highlight some representative examples. The Gulf of Mexico: a major contributor, with hydrocarbons leaking since millions of years57, resulting in an estimated discharge exceeding 0.14 Mt yr−115; southern California (Santa Barbara Channel), with an estimated discharge of 17,000 tonnes yr−156;. These values are very similar to our estimates for Lusi (i.e. ~ 17,500–22,200 tonnes yr-1), considering that the oil is transported at the surface together with the mud. The Lusi oil discharge is also comparable to the largest known modern oil spill at the Deepwater Horizon well in the Gulf of Mexico, which was evaluated to be ~0.7 Mt of oil equivalents58. In contrast, oil release from the Exxon Valdez spill was estimated at ~ 37,000 tonnes (Table 2b), similar to our yearly assessment for Lusi. Comparisons can be made also with exploited reservoir volumes. Oil fields with recoverable reserves below 1 Mt are classified as very small oil and gas reserves59. Despite this, some smaller fields, like the Flyndre field in the North Sea, are in production (original recoverable resources of 85,800 tonnes of oil equivalents (Table 2b)). Lusi has discharged approximately 0.22–0.28 Mt of oil between 2006 to 2019, dwarfing the size of the Flyndre field.

Environmental impact of the discharged mud and oil

Since 2006, water and oil-soaked mud erupted at Lusi have been continuously released in the 7 km2 basin framing the crater sites, the north stream running E-W along the embankment, the southern Porong River and, ultimately, the delta located 25 km to the east. Environmental studies in the Porong River and estuary have focused on the impact on sedimentary processes60,61, and on heavy and trace metal contaminations in sediment, water and fish62,63,64,65,66,67,68. Lead, cadmium and zinc concentrations are ~ 10 times higher than the Indonesian regulatory limits66, and we estimate mercury concentrations of roughly 0.013–0.016 mg/l, 6–8 times higher than the US drinking water limit of 0.002 mg/l69, based on Hg levels of ~ 34 ppb in the dry mud and assuming a water content of 53–63 vol.%. However, no oil contamination survey has, to our knowledge, been undertaken despite reports of anomalously elevated phenol levels in the mud and water dumped from Lusi into the Porong River when the eruption started in 200670.

Several processes control the transport of spilled fluids in anthropogenic oil discharges (e.g., spreading, advection, dispersion, entrainment, sinking sedimentation15), resulting in widespread and persistent oil slicks affecting surface waters and soils, e.g. in the Niger Delta Goi disaster caused by pipeline leakages71,72. By contrast, the fate of the hydrocarbons expelled at Lusi is anticipated to be largely controlled by sinking sedimentation via adsorption and water-oil emulsion. Adsorption of petroleum on fluidized mud is the main oil discharge mechanism at Lusi, with the oil-contaminated sediments finally released in the river and estuary. Additionally, vigorous stirring of oil-soaked mud and water during eruption and in the embankment is likely to drive the formation of water-oil emulsions, increasing the density of the free oil fraction and promoting adsorption on particles and settling in the sediments73.

The considerable quantities of (oil-contaminated) mud released in the waterways lead to very high sedimentation rates in the Porong River mouth, where a vertical accretion > 10 cm yr-1 was measured already in 2010-201168. In fact, Lusi mud is used in mangrove plantation projects to fight coastline erosion and foster aquaculture facilities74, highlighting the crucial need for oil-contamination monitoring surveys of the riverine and estuarine sediments.

Materials and methods

Mud flow rate

Between May 2006 and January 2019, 805 measurements of mud flow rate were carried out at Lusi eruption site (Fig. 1A). The measurements were performed 1–4 times per month during the first ten years of activity and were intensified with a daily routine after June 2017 (Fig. 2A). The survey was conducted by two main operators: Lapindo Brantas oil and gas exploration company and Badan Penanggulangan Lumpur Sidoarjo (BPLS) – the agency supervising the mitigation infrastructures at the Lusi site (later renamed to Pusat Pengendalian Lumpur Sidoarjo, PPLS). The mud flow rate measurement methods were modified through time depending on the morphology and the flooding conditions at the Lusi site (Fig. 1). During the initial phases of the eruption, the area around the Lusi crater was subdivided in distinct ponds of various sizes depending on the settlement locations and the morphology of the terrain. The mud level filling the ponds of known dimensions was monitored by calibrated sticks combined with frequently acquired satellite images. The calculated mud volume was compensated with the regional subsidence that was monitored by a network of GPS stations and by the measured rain precipitation. Detailed descriptions on these procedures are reported in Istadi et al., 200927. Since March 2009, the active vent remained isolated at the centre of a 7 km2 area framed by the 11 m tall embankment. Meanwhile, on the dry edges of the embankment, several areas of known dimensions were dredged. The erupted water was diverted through channels that were dug using excavators and then converged at large ponds located on the outskirts of the Lusi area framed by the embankment (e.g. Fig. 1B). These ponds were used as collection areas for the erupted mud. Water from the neighboring Porong River was pumped into these ponds to reduce the mud viscosity and facilitate the removal of the solids. The water-enriched mud was then pumped out to the Porong River (Fig. 1A, E, F). The final mud flow rate estimations accounted for the mud level in the ponds, compensated with the amount of water pumped in and out, the measured precipitations, and the GPS-monitored subsidence coupled with satellite images. To corroborate the validity of the obtained values, flow measurements were also conducted at the outflow of the channels before the collection on the large ponds. More details regarding the approach and the calculations used are provided by Mazzini et al.25, and in the BPLS and PPLS reports to Indonesian Ministry that can be accessed upon request. Considering the significant evaporation ongoing at the Lusi crater and at the adjacent ponds, the acquired flow rate measurements are most likely underestimated.

Mud sampling

Mud sampling was carried out during annual field campaigns between 2006-2017. Hereinafter, ‘mud’ is referred to as the erupted material consisting of water and solid fractions (mainly sandy-silty clay with minor amount of rock clasts). Various sampling techniques were applied depending on the morphology of the crater and its accessibility (Fig. 1). During the 2006 field campaign, mud was collected using an excavator positioned on the edge of the berm around the active vent. Later, the mud sampling was performed using a floating excavator that reached the edge of the active crater, or using a remotely controlled drone75. Other mud samples were collected from the streams that drained the main crater. In order to define the average sediment content to be used for calculations, we considered all the measurements conducted during the various years. The solid fraction of the sampled mud (sediment content) was 57 wt.% in 2006 and 55 wt.% in 2007 and 2010. Using water and sediment densities measured by Manga et al.76, on exactly the same samples (i.e. 1.05 g/cm3 and 2.2 g/cm3, respectively), we calculated an equivalent of 36–39 vol.% of the dry sediment in the erupted mud. The water content was reported to change through time varying from 60 vol.% in 2006 to 30 vol.% in 200723, and reaching 40 vol.% of water by 200848. Mud samples were stored in the refrigerator (4 °C) in bottles ranging in size from 50 to 500 ml. All samples have mild to strong gasoline odour. The majority of the mud samples appears homogenous.

Rock clasts

Rock Clasts were collected throughout the years at the Lusi crater. An extensive set of geochemical, paleontological, and petrographic analyses of the collected clasts revealed the presence of two main organic-rich source rocks: 1) lower maturity and lower organic content (TOC from 0.3 to 2 wt.%) “grey shales” from the Upper Kalibeng Formation and 2) higher maturity and higher organic content (TOC from 1.6 to 14.6 wt.%) “black shales” from the Ngimbang Formation26,35 In this study we have performed additional analyses on the isotopic composition of the extracted organic matter from the collected rock clasts of these two potential source rocks.

Sediment content

Sediment content of the erupted mud was measured for the 2006, 2007 and 2010 samples by weighing the mud sample before and after drying in the oven at 60 °C for 24 h. Hereinafter, ‘sediment’ is referred to as the dry bulk solid fraction of the sampled mud, dominated by silty-sandy clay.

Mercury content

Mercury content measurement of the 2017 dry mud sample was carried out at the University of Oxford, U.K. The analysis was performed using a Lumex RA-915 Portable Mercury Analyzer with an attached PYRO-915 pyrolyzer, following established in-house protocols77,78,79 and calibrated using the NIMT/UOE/FM/001 peat standard with a known Hg concentration of 169 ± 7 ppb. The analytical procedure was performed twice, using 45 and 164 mg of the powdered sample. The aliquots were heated to >700 °C in the pyrolyzer and left for up to 120 seconds to allow full volatilization of the Hg present. Individual sample analytical errors are ±5%.

Liquid and solid hydrocarbon content and composition

Liquid and solid hydrocarbon content and composition of the 12 mud samples were analysed at the Department of Geosciences, University of Oslo, Norway. Mud samples were dried in the oven at 60 °C for 24–48 h, depending on the water content. The dried sediments were subjected to extraction in the Soxtec system HT 1043 using dichloromethane with 7 vol.% methanol (i.e. volume ratio 93:7). The extraction method followed “The Norwegian Industry Guide to Organic Geochemical Analyses” (NIGOGA)80. The glass fibre thimbles with 2.2–8.8 g of sediment were boiled for 1 h and rinsed for 2 h in the solvent. Dissolved samples of the extractable organic matter (EOM) were concentrated by evaporation at room temperature.

The relative concentrations of the saturated, aromatic and polar (i.e. resins and asphaltenes) compound classes of the EOM were estimated using Iatroscan MK-5 equipped with thin layer chromatography-flame ionization detector (TLC-FID)80,81. The method is based on application of 3 μl of the bulk extract dissolved in a dichloromethane-methanol mixture (93:7) on silicagel-coated rods (type Chromarod III). The rods are subsequently submerged in the tanks with n-hexane and toluene to elute the saturated and aromatic fractions. The accuracy of peak detection by the TLC-FID is ±3%.

The bulk EOM was analysed using a Varian 3800 capillary gas chromatograph with flame ionization detector (GC-FID), equipped with a Hewlett Packard Ultra 1 cross-linked methyl silicone column (50 m × 0.2 mm i.d, film thickness 0.33 µm). The initial temperature of the column was 40 °C (held for 2 min), later heated to 320 °C (held for 20 min) with a gradient of 4.5 °C/min.

Saturated and aromatic fractions were separated using pipette chromatography over annealed and dry-packed silica gel (Mesh 230–400) as described in Vinnichenko et al.82. The separated fractions were analysed by gas chromatography-mass spectrometry (GC-MS) on a Thermo Scientific Trace 1310 gas chromatograph coupled to a Thermo Scientific TSQ 8000 Triple Quadrupole MS. The GC was equipped with a low polarity column Thermo Scientific TG-XLBMS (60 m long×0.25 mm i.d. and a film thickness of 0.25 µm). Aromatic fractions were analysed using selected ion monitoring (SIM) mode (m/z 142, 156, 170, 178, 184, 192, 198). Hopanes and steranes in the saturated fractions were analysed by the selected reaction monitoring (SRM) method using M+ →191 and M+ →217 precursor–product transitions, respectively.

Carbon isotopic composition

Carbon isotopic composition (δ13C) of the aliphatic and aromatic fractions of the EOM from mud and rock clasts was completed at the Federal Institute for Geosciences and Natural Resources (BGR, Hannover). Bulk δ13C analyses of the aliphatic and aromatic fractions were performed on the elemental analysis–isotope ratio mass spectrometry (EA–IRMS) using a coupled Flash EA 1112 (Thermo Fisher Scientific), ConFlow IV (Thermo Scientific), and a Delta V Advantage IRMS (Thermo Fisher Scientific). Standard deviation is <0.1 ‰.

Liquid and solid hydrocarbon discharge

Calculation of the liquid and solid hydrocarbon discharge. In order to estimate the total hydrocarbon (i.e. oil + bitumen) discharge from Lusi during the study period, we utilized the daily mud flow rate measurements, sediment content in the mud (dry bulk mud), and EOM from the sediment. Since the daily mud flow rate measurements are sporadic throughout the considered period, the mud flow rate data were interpolated using a shape-preserving linear interpolation method in Matlab for the time periods that were lacking field measurements. The EOM content in the sediments, obtained after the extraction, was linearly interpolated for the studied time interval. The daily hydrocarbon discharge from the Lusi mud eruption during the studied period was then calculated using the following equation:

$${Hydrocabon\; Discharge}({tonnes})=\frac{F* S* D* {EOM}}{10000}$$
(1)

where F—mud flow rate (m3/day), S—sediment content in the mud (dry bulk mud sample, vol.%), D—measured sediment density 2.2 tonne/m3,76, EOM—extractable organic matter from the sediment (weight%).

C2-C3 emission calculation

The ethane (C2) and propane (C3) emissions from LUSI have been estimated using the molecular composition of methane (C1), and C2-C3 concentrations and the C1 emission following the same procedure in Etiope and Ciccioli19 and Etiope et al.51 Specifically, the following formula has been adopted:

$${{{\rm{C}}}}_{{{\rm{n}}}}{{\rm{emission}}}={{\rm{C}}}_{1}{{\rm{emission}}}\,{{\rm{x}}}\,{{\rm{C}}}_{{\rm{n}}}/{{\rm{C}}}_{1}{{\rm{x}}}\,{{\rm{MM}}}_{{\rm{Cn}}}/{{\rm{MM}}}_{{{\rm{C1}}}}$$
(2)

where Cn is C2 or C3 and MM the molar mass.

The C2-C3 emissions were calculated using C1 emission from both ground based and satellite (TROPOMI) measurements25. For the ground-based measurements, the specific emission of each of the four sectorial, gas seepage systems of Lusi, as classified in Mazzini et al.25, i.e., crater, fractured zones, satellite seeps and miniseepage, has been estimated. The C1-C3 composition in the gas released from crater, fractures and satellite seeps has been taken from Mazzini et al.38. For miniseepage, the fracture C1-C3 composition has been considered, as the emission from fractured zones is diffuse, widespread over a wide area, with a process similar to the miniseepage. It, in fact, shows a higher C1/C2 + C3 ratio, typical of molecular fractionation in low and diffuse seepage83. The average C1-C3 concentration values from the samples that are less contaminated with air have been considered for the computation of the C2 and C3 flux.

For the C1 emission derived by TROPOMI satellite, the mean values of the C2/C1 and C3/C1 ratios from the several sectors, weighted for the C1 emission, have been considered (Table S5).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.