Earth surface temperatures, including in the deep sea increased by 5–10°C from the late Paleocene ca. 58 Myr ago to the Early Eocene Climatic Optimum (EECO) centered at about 51 Myr ago. A large (∼2.5‰) drop in δ13C of carbonate spans much of this interval. This suggests a long-term increase in the net flux of13C-depleted carbon to the ocean and atmosphere that is difficult to explain by changes in surficial carbon cycling alone. We reveal a relationship between surface temperature increase and increased petroleum generation in sedimentary basins operating on 100 kyr to Myr time scales. We propose that early Eocene warming has led to a synchronization of periods of maximum petroleum generation and enhanced generation in otherwise unproductive basins through extension of the volume of source rock within the oil and gas window across hundreds of sedimentary basins globally. Modelling the thermal evolution of four sedimentary basins in the southwest Pacific predicted an up to 50% increase in petroleum generation that would have significantly increased leakage of light hydrocarbons and oil degeneration products into the atmosphere. Extrapolating our modelling results to hundreds of sedimentary basins worldwide suggests that globally increased leakage could have caused a climate feedback effect, driving or enhancing early Eocene climate warming.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
1. Early Eocene Climate and Carbon Isotope Excursions
 After reaching a long-term high in the late Paleocene, marineδ13C values overall decline by 0.5‰/Myr [Cramer et al., 2009; Nicolo et al., 2007; Zachos et al., 2010]. Carbon flux modeling relates the increase in δ13C values during the early Paleogene to a change in net organic carbon burial through sedimentary systems of up to 5,000 Gt C/Myr [Hilting et al., 2008; Kurtz et al., 2003]. The subsequent overall decrease in δ13C values leading to the Early Eocene Climate Optimum (EECO) suggests an equivalent net flux of 13C-depleted carbon into surface systems [compareHilting et al., 2008]. This notion is supported by an apparently contemporaneous drop in the lysocline and calcite compensation depth [Leon-Rodriguez and Dickens, 2010]. Such a flux is difficult to explain without invoking a huge external reservoir of 13C-depleted carbon that can slowly release carbon [e.g.,Dickens, 2011; Kroeger et al., 2011]. While not generally considered to be emitting significant amounts to the ocean and atmosphere, subsurface sources such as microbial methane (δ13C = −60 to −110‰) or thermogenic methane (−20 to −50‰) have the potential to cause a long-term shift in marineδ13C values, given the 15,000,000 Gt of organic carbon buried in sedimentary basins [Berner, 1989]. A possible source for 13 C-depleted methane could be carbon mobilized by microbial methane generation or petroleum that was not trapped in conventional accumulations [Kroeger et al., 2011]. The abundance of seep structures such as pockmarks at the modern seafloor and within the sedimentary record and the abundance of mud volcanoes supports the notion that methane flux from subsurface reservoirs contributes significantly to fluxes to atmosphere and oceans [Davy et al., 2010; Dimitrov, 2002; Hovland and Judd, 1988]. A considerable part of this flux is thermogenic methane [Etiope et al., 2009; Milkov, 2005].
 It has recently been shown that carbon burial and remobilization rates through microbial methane generation in the shallow (<1000 m) subsurface significantly increase during warm climates [Gu et al., 2011]. Deeply buried organic matter, on the other hand, is usually thought to be largely decoupled from surface processes. However, any increase in surface temperature will propagate to depths sufficient for thermal generation of oil and gas in less than 1 Myr [Barker, 2000] (Figure 1a). Hence, given an average temperature gradient of 3°C/100 m, a 10°C increase in surface or water-sediment interface temperature will shift the depth window suitable for oil and gas generation upward by approximately 300 m. For source rocks located within or just above the generation window, this can significantly increase the rate of transformation to oil and gas (Figures 1b and 1c). In this paper, we investigate the effects of climate warming in the early Eocene on petroleum generation in southwest Pacific basins. We then discuss whether globally increased petroleum generation could have increased leakage and caused a climate feedback effect.
2. Model Construction
 The southwest Pacific region has a well-established record of warming through the Paleocene-Eocene transition that can be used for a regional case study in an area with widespread Mesozoic organic rich sediments that are known to be important petroleum source rocks. Sea surface temperature (SST) data derived from TEX86 [Bijl et al., 2009] and other paleotemperature proxies [Hollis et al., 2009a, 2009b] indicate a pronounced temperature rise of ∼10°C (20 to 30°C) from late Paleocene to early Eocene. Preliminary studies of benthic foraminiferal Mg/Ca paleothermometry, offshore eastern New Zealand, indicate a parallel rise in seafloor temperatures from ∼5 to ∼15°C [Hollis et al., 2009b]. The extended SST record from Tasman Rise [Bijl et al., 2009] south of Lord Howe Rise (Figure 2) is consistent with shorter New Zealand records [Hollis et al., 2009a; Nicolo et al., 2007] and thus may be taken to represent the regional climate regime for late Paleocene to mid Eocene times [Bijl et al., 2009]. For modelling the impact on petroleum generation, the absolute increase in temperature indicated by both shallow and deep water proxies [Hollis et al., 2009a, 2009b] is the critical factor. Although recent recalibration of the TEX86 proxy [Kim et al., 2010] may bulk-shift calculated SSTs towards lower values, this shift would therefore not significantly affect model results.
 We examine how pronounced climate warming may have affected oil and gas generation using PetroMod™ software, a package widely used in petroleum exploration and research [e.g., di Primio and Neumann, 2008; Kroeger et al., 2008]. PetroMod™ is designed to reproduce basin evolution through time using 3D numerical forward modelling to predict thermal history, and petroleum generation and migration. Further information can be found on http://www.slb.com and in the work by Hantschel and Kauerauf . Four large petroleum generating basins in the region were investigated: the Taranaki and Great South basins, located on the New Zealand shelf [Cook et al., 1999; King and Thrasher, 1996], and the Capel and Faust basins, located on the Lord Howe Rise between New Zealand and Australia [Funnell and Stagpoole, 2011] (Figure 2). The reconstruction of the four sedimentary basins relies on seismic, biostratigraphic and well data, and published previous models [Cook et al., 1999; Funnell and Stagpoole, 2011; King and Thrasher, 1996]. In the New Zealand basins, several 100 m to several 1000 m thick Late Cretaceous organic-rich sediments were deposited [Cook et al., 1999; King and Thrasher, 1996], whereas in the Capel and Faust basins, the age of potential source rocks ranges from Late Jurassic to Late Cretaceous [Funnell and Stagpoole, 2011]. The inferred average total organic carbon content (TOC) of the source rocks was 1–3% in the Capel and Faust basins and up to 8% in Taranaki and Great South basins. Parameters defining hydrocarbon (HC) yield used in the models as defined by the hydrogen index HI (in mg HC/g TOC) were 250–300. Models were calibrated using corrected in-house and published well temperature and vitrinite reflectance data [Funnell et al., 1996; Sykes et al., 1992].
 3D forward modelling of basin reconstructions has predicted the Cretaceous to Recent temperature history of each of the four basins and calculated petroleum (oil and gas) generation through time (Figure 3). A late Paleocene-early Eocene climate warming scenario, using the TEX86data was compared against the results of a constant 20°C surface temperature scenario. Petroleum generation and expulsion in Capel and Faust basins is predicted to have begun in the Cretaceous and continued throughout late Paleocene and early Eocene time. In the Great South Basin most petroleum is predicted to have been expelled in the Eocene, whereas in the Taranaki Basin expulsion began in the Eocene with increasing rates predicted in the Miocene-Recent after a period of tectonic quiescence and slow burial [King and Thrasher, 1996]. The primary result from modelling the two scenarios is the occurrence of an intense period of petroleum generation and expulsion coincident with the EECO in the climate warming scenario. In the Great South Basin the peak in expulsion was advanced by 8 Myrs in the climate warming scenario, coinciding with the EECO (Figure 3). The predicted early Eocene expulsion rates during the EECO are up to 4 times higher for oil and 5 times higher for gas than in the constant 20°C surface temperature model (Figures 1 and 3). Modelled oil expulsion rates reach 7500 Mtons/Myr during the early Eocene, amounting to 238 Gt of total expelled oil in the Great South Basin to the present day. The very large amounts of petroleum generated in these southwest Pacific basins are consistent with the view that source rocks generate several orders of magnitude more oil and gas than the amount found in conventional accumulations [Price, 1994]. For example, only 0.04–4.0% of the petroleum generated from Upper Jurassic source rocks is estimated to be recoverable in sedimentary basins worldwide [Klemme, 1994].
4. Petroleum Generation and Leakage
 The models suggest that an additional 37 Gt of oil and 8.3 Gt of gas were generated and expelled within the latest Paleocene and early Eocene from all four basins as a result of climate warming, 50% more than in the constant surface temperature scenario. Increased generation and expulsion is likely to have resulted in greater petroleum fluxes within the basin promoting an increase in oil and gas leakage at the surface. Estimates of 600,000 tons per year of present-day global oil seepage [Kvenvolden and Cooper, 2003], enough to deplete all produced, discovered and estimated undiscovered conventional resources worldwide in less than 1 Million years, emphasize the significance of leakage in sedimentary basins. This reinforces the view that only a small proportion of the oil and gas generated in sedimentary basins is trapped while a much larger proportion leaks to the surface driven by its buoyancy. Little is known about the average time it takes for oil and gas to migrate through a sedimentary succession to the surface. While horizontal flow in carrier beds can reach velocities in the order of 1000–10,000 km/Myr [England, 1994; Sylta, 2002], vertical flow on a basin scale is typically several orders of magnitude slower, due to intercalated less permeable layers. However, flow capacity through impermeable layers is enhanced by trapped petroleum accumulations increasing the pressure on sealing strata which may lead to break-through or leakage of oil or gas. This process tends to focus vertical flow and may lead to conduits of concentrated leakage to the surface [Kroeger et al., 2011]. Furthermore, phase separation and biodegradation of oil at shallow depth increases the amount of gas relative to liquid petroleum. Given that flow velocities of a continuous gas phase may be as high as 100–1000 m/yr [Brown, 2000] these processes may not only considerably increase the amount of gas leaking to the surface, but also reduce the time between petroleum expulsion and surface leakage of a gaseous phase.
5. Global Implications
 If we can assume that the varied burial histories of the modelled southwest Pacific basins represent a reasonable proportion of the more than 600 sedimentary basins in the world, and that they are similarly affected by increased surface or ocean temperatures, it is reasonable to expect that climate warming has influenced global oil and gas leakage. Remobilization, through expulsion and migration, of only a fraction of the 15,000,000 Gt of organic carbon stored in sedimentary basins could result in globally significant leakage. An indicator that a relationship between globally increased rates of petroleum generation and leakage has existed is the cumulation of paleo-seepage structures in the early Eocene part of many sedimentary successions worldwide [e.g.,Cole et al., 2000; Hartwig et al., 2011; Van Rensbergen et al., 2007]
 Increased surface temperature begins to affect petroleum generation within less than 1 Myr, but leakage of oil and gas to the atmosphere in the form of methane and other volatile components is expected to lag in time due to the migration process. This lag may be reflected by the age of the reversal in the benthic δ18O trend, which appears to lead the reversal in the δ13C trend by 1–2 Myr [Cramer et al., 2009] (Figure 4). However, the decrease in global benthic δ13C predates significant increase in predicted expulsion (Figure 4). This is in part a reflection of the initially slow increase in SST indicated by the Bijl et al.  record. The direct comparison of records is further complicated by the use of data from different sites with different age models. Moreover, while our models represent high latitude sites that were terrestrial or in shallow water during the Eocene [Cook et al., 1999; King and Thrasher, 1996; Funnell and Stagpoole, 2011], the timing and magnitude of the increase in petroleum generation may have been different at other deeper water or lower latitude locations. Then again, increased rates of petroleum generation and leakage may have been the driving force in the initial late Paleocene global warming process, primarily through rapid burial of prolific Jurassic and Cretaceous source rocks as a result of globally increased Cretaceous-Paleocene sediment burial rates [Klemme and Ulmishek, 1991]. After beginning to decrease with the increase in surface temperature in the latest Paleocene, δ13C values began to increase again early during the EECO (Figure 4). Our models, on the other hand, predict only a gradual decrease in petroleum expulsion after the EECO with expulsion rates continuing to be higher than in the constant surface temperature scenario for 5–10 Myrs. A possible reason for the increase in marine δ13C during the EECO, despite a continuous influx of isotopically light carbon into the atmosphere, could be a significant increase in terrestrial sink strength and buffering by increased influx of dissolved carbonate from weathering. Mechanisms that could have increased terrestrial sink strength include expansion and diversification of tropical forests in the Early Eocene and increase of peat formation [Jaramillo et al., 2006; Kurtz et al., 2003; Zachos et al., 2010]. In addition to increased rates of weathering, the release of HCO3− with a heavy carbon isotope signature originating from anaerobic oxidation of methane could have buffered δ13C values [Dickens, 2011].
 Other changes in global carbon cycling such as changes in peat formation or its destruction due to wildfires [Hilting et al., 2008; Kurtz et al., 2003] may help to explain the early Paleogene δ13C record. However the masses involved exceed the amount of carbon estimated to have been contained in the terrestrial soil and vegetation reservoir by 10 fold (compare Beerling , Hilting et al. , and Kurtz et al. ). Our models suggest that a warming climate will affect petroleum generation in sedimentary basins at a global scale; firstly by synchronizing periods of maximum generation in time, and secondly by enhancing generation in otherwise unproductive basins through extension of the volume of source rock within the oil and gas window. Over time an additional carbon flux of both microbial and thermogenic origin into the surficial carbon cycle would have caused a feedback process whereby higher surface temperatures further enhanced rates of carbon mobilization. An inverse process could be invoked for times of climatic cooling, where mobilization of organic carbon slowly decreased in response to decrease in temperature, such as during the Paleocene (Figure 4). We therefore propose that the early Paleogene δ13C record may be better understood as reflecting a change in the ratio of organic carbon burial versus remobilization. It has yet to be determined by analyzing a large number of basins at different paleo-latitudes and waterdepths worldwide whether this feedback effect was globally significant enough to drive climate change in the past or whether it merely enhanced climate warming.
 This research was undertaken with Crown funding provided through the New Zealand Ministry of Science and Innovation. We thank R. Hashimoto and Geoscience Australia for providing data on Capel and Faust basins, C.J. Hollis and G. Browne for comments and discussion, and G.R. Dickens and M. Lyle for constructive reviews of the paper. We are grateful for IES/Schlumberger in providing access to PetroMod™ software for petroleum generation modelling.
 The Editor thanks Gerald Dickens and Mitchell Lyle for their assistance in evaluating this paper.