2.a. The origins of palaeogeography

The term ‘paleogeography’ (palaeogeography) was first coined by Thomas Sterry Hunt in his paper on the palaeogeography of North America (Hunt, Reference Hunt1873) to describe the ‘geographical history’ of the Earth through geological time. Hunt was one of the earliest petroleum geologists and was looking for ways to bring together diverse geological information in the search for oil (Hunt, Reference Hunt1862). Land–sea maps of the past had been available since at least the 1830s, and the Tertiary maps of Northwest Europe by Lyell (Reference Lyell1830, Reference Lyell1837, pp. 214–15 foldout) and Elie de Beaumont (Beudant, Reference Beudant1846, Reference Beudant1872), but Hunt saw palaeogeography more broadly, encompassing all the elements of present-day geography including the oceanography, climatology, biota and orography, as well as land and sea. It was also Hunt who established the basis for a common palaeogeographic mapping workflow, stressing the importance of not only representing the surface of the Earth, but also the underlying crust, the ‘architecture’ of the Earth as he referred to it, which is now termed the ‘crustal architecture’.

Despite Hunt's definition, and the growth of oil exploration, palaeogeographic maps throughout the remainder of the nineteenth century remained simple. Many more maps were published, culminating at the end of the nineteenth century with the first series of global palaeogeography maps in the fourth edition of Lapparent's Traité de Géologie (Lapparent, Reference Lapparent1900). But, these were all land–sea maps. One exception was Jukes-Browne's (Reference Jukes-Browne1888) atlas of 13 palaeogeographic maps showing the geological evolution of Great Britain, which also showed the major palaeo-river systems for select intervals.

It was not until the early twentieth century that Hunt's definition of palaeogeography was more fully realized, beginning with the work of Arldt (Reference Arldt1910, 1917–Reference Arldt1922) and Dacqué (Reference Dacqué1915) in Germany, and Willis (Reference Willis1909) and Schuchert (Reference Schuchert1910) in America. Palaeogeography had been a subject of considerable interest in Germany since the turn of the century (Kossmat, Reference Kossmat1908), and it was here that all the elements needed to draw a modern palaeogeography (see Section 3) came together, including the concept of continental drift (Wegener, Reference Wegener1912a,Reference Wegenerb). Arldt focused his Handbuch der Palaeogeographie (Arldt, 1917–Reference Arldt1922) on palaeobiogeography, echoing ideas started 100 years before by Humboldt & Bonpland (Reference Humboldt and Bonpland1807) and developed further by Wallace (Reference Wallace1876), using land-bridges to explain inconsistencies in observed geographic faunal and floral patterns. Dacqué expanded on Wegener, concentrating on crustal architecture, though oddly he did not draw palaeogeographies on reconstructions. It would be another 20 years before someone put all these elements together on a plate reconstruction map, and then it would be a South African, Du Toit in his 1937 book, which included maps showing the distribution of fossils, lithologies and structures on continental reconstructions (Du Toit, Reference Du Toit1937). This was followed by King (Reference King1958) showing similar Gondwana reconstructions. It would then take until the 1970s before the first systematic atlases of palaeogeography drawn on plate reconstructions would be published in the rest of the world.

Despite these palaeogeographic advances in Germany, the geopolitics of the early twentieth century meant that future developments in palaeogeography, as a science, were American, not German. These developments were first publicly discussed at the Baltimore meeting of the American Association for the Advancement of Science (AAAS) in 1908 and summarized by Bailey Willis in his Phanerozoic North American palaeogeographic atlas (Willis, Reference Willis1909), which not only showed maps of land and sea, but also reconstructions of major ocean currents, made observations about the consequences on the distribution of faunas and provided a graphical indication of mapping uncertainty. Much of this work was actually that of Charles Schuchert, a professor at Yale University, who was to drive palaeogeography and palaeoclimatology through the first half of the twentieth century. Schuchert expanded on the workflows of Hunt, Dacqué and Ardlt, emphasizing again the importance of considering all the geological and geographic components including the past drainage, climate and elevation. This was formalized as a series of guidelines for palaeogeographic mapping published in 1928 (Schuchert, Reference Schuchert1928). But, Schuchert added two new components: first, the need to consider how to reconstruct compressional systems through palinspastic restorations to better represent past landscapes, which was later expanded on by Kay (Reference Kay1945) and Ziegler et al. (Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985); and second, the importance of stratigraphy, especially biostratigraphy, in constraining the temporal resolution of palaeogeographic maps (Schuchert, Reference Schuchert, Dana, Schuchert, Gregory, Barrell, Smith, Lull, Pirsson, Ford, Sosman, Wells, Foote, Page, Coe and Goodale1918) (see Section 3.c). Schuchert was a colleague of Joseph Barrell, and consequently, much of Schuchert's work emphasized the importance of stratigraphy in palaeogeography, including the significance of temporal resolution in mapping. Schuchert's North American atlas of palaeogeography comprised 52 maps spanning Early Cambrian to Pleistocene times and included the names of the stratigraphic formations represented (Schuchert, Reference Schuchert1910).

2.b. Palaeogeography and regional exploration

It was also Schuchert who again stressed the potential benefits of using palaeogeographies in oil and gas exploration (Schuchert, Reference Schuchert1919). The next two decades did see an increase in the application of palaeogeography in exploration, though mostly at the basin or local scale. This included maps to investigate the general geological setting of an exploration area, such as those of Robinson (Reference Robinson1934), who showed land–sea maps for the Upper Cretaceous formations of Montana, to investigate the details of transgression and regression in this part of the Interior Seaway. Other workers, such as Kellum (Reference Kellum1936), used palaeogeographic mapping to identify potential spatial trends, in this case, the possible southward continuation of the (Cretaceous) oil fields of West Texas into Mexico. But, the major use of palaeogeography during this time appears to have been focused on reservoir rocks, especially investigations of shoreline deposits, since shoreline clastic sediments had been identified as potential reservoirs in many areas. Reed (Reference Reed1923) looked at the palaeogeographic setting of the Booch Sand in the Pennsylvanian of Oklahoma, which included detailed sedimentology and isopachs with the palaeogeographies. This focus on clastic sediments, in turn, led to a renewed interest in the reconstruction of palaeo-river systems. Reed (Reference Reed1926), this time in the central Coastal Ranges of California, brought together climate, hinterland topography and drainage to explain the lack of clastic sediments and terrestrial carbon in the local Miocene marine rocks. Zuber (Reference Zuber1934), working in the Black Sea and Caspian Neogene petroleum province, included palaeo-rivers on his maps.

Levorsen (Reference Levorsen1931, Reference Levorsen1933, Reference Levorsen1936) took a slightly different approach by looking at palaeo-geology (including the structure and isopachs of different units), palaeotopography and structure for a series of timeslices in North America. The rationale was to use the palaeo-geology to map out unconformities and then use these to reconstruct palaeotopography and thereby identify potential hydrocarbon traps. Palaeo-geology and palaeogeography are today linked through source-to-sink analysis within the palaeogeographic workflow (Section 3.b.6).

By the 1940s, exploration geologists had started to produce regional- and continent-scale palaeogeographies for different parts of the world as exploration expanded into new frontiers, such as South America (Weeks, Reference Weeks1947) and Australia (Teichert, Reference Teichert1941). In China, Hsieh (Reference Hsieh1948) described the use of palaeogeography to explore for non-petroleum minerals including bauxite, phosphate and coal, coincidentally all to become important geological constraints on the results of climate modelling retrodictions of source rocks during the 1980s and 1990s (Section 2.c).

Despite these examples, palaeogeography was still not widely used within exploration companies, which Adams (Reference Adams1943) ascribed to the fact that regional and especially global palaeogeography maps took a lot of time to prepare and, following Schuchert (Reference Schuchert1910), that the maps were never truly ‘complete’ because the underlying information kept changing. Nonetheless, Adams stressed their value in exploration and proposed a further tool that used palaeogeography, together with climate and all the other components of the Earth system, to retrodict the most favourable areas for source and reservoir lithological deposition, a methodology he called ‘eulithogeology’. This was the precursor of play mapping and also lithofacies retrodiction as can be seen by Adams’ summary description of what a eulithogeologic map is: ‘The eulithogeologic map might well be called a fairway map with many wells still being drilled in the rough’ (Adams, Reference Adams1943).

2.c. Plate tectonics, global palaeogeography and new ventures exploration

Through the 1950s, 1960s and 1970s more detailed palaeogeographic maps were published, including Wills’ A Palaeogeographical Atlas of the British Isles and Adjacent Parts of Europe (Wills, Reference Wills1951), the systematic mapping of the former Soviet Union by Vinogradov and his team (Vinogradov, Grossheim & Khain, Reference Vinogradov, Grossheim and Khain1967; Vinogradov, Reference Vinogradov1968, Reference Vinogradov1969; Vinogradov, Vereschchagin & Ronov, Reference Vinogradov, Vereschchagin and Ronov1968) and the Geological Atlas of the Rocky Mountain Region (Mallory, Reference Mallory1972). These were amongst the most detailed palaeogeographic atlases that had ever been produced and developed a series of graphical methodologies that would be adopted and further developed by Fred Ziegler's Paleogeographic Atlas Project in Chicago (Ziegler et al. Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985) (Section 3.a).

The limited application of palaeogeography in exploration changed radically with the acceptance of sea-floor spreading (Dietz, Reference Dietz1961; Hess, Reference Hess, Engel, James and Leonard1962; Wilson, Reference Wilson1962, Reference Wilson1963a,Reference Wilsonb,Reference Wilsonc; Vine & Matthews, Reference Vine and Matthews1963; Vine, Reference Vine1966) and from this plate tectonics, which resulted in the publication of regional (Bullard, Everett & Smith, Reference Bullard, Everett and Smith1965) and then global plate reconstructions (Smith, Briden & Drewry, Reference Smith, Briden, Drewry and Hughes1973). Onto these were then plotted the past distribution of geological data (Hughes, Reference Hughes1973; Vine, Reference Vine and Hughes1973; Drewry, Ramsay & Smith, Reference Drewry, Ramsay and Smith1974), including by 1982 exploration information (Bois, Bouche & Pelet, Reference Bois, Bouche and Pelet1982) and palaeo-shorelines (Smith, Smith & Funnell, Reference Smith, Smith and Funnell1994).

The oil industry was now quick to pick up on the value of palaeogeographic mapping on the new reconstructions. By the 1980s most major companies were either building their own maps or supporting specialist academic groups who did, or both. This included BP beginning in 1981 (see discussion in Summerhayes, Reference Summerhayes2015), Peter Ziegler's work at Shell (Ziegler, Reference Ziegler1982, Reference Ziegler1990, Reference Ziegler1999), Jan Golonka and colleagues at Mobil and later in collaboration with Chris Scotese and Malcolm Ross (Golonka, Scotese & Ross, Reference Golonka, Scotese, Ross, Beauchamp, Embry and Glass1993; Golonka, Ross & Scotese, Reference Golonka, Ross, Scotese, Embry, Beauchamp and Glass1994; Golonka & Scotese, Reference Golonka, Scotese and Thurston1994) and Tony Doré at Conoco (later at Statoil) (Doré, Reference Doré1991), whose maps, like those of Peter Ziegler, were underpinned by a detailed structural and tectonic framework. In Russia, Ronov, Khain & Seslavinski (Reference Ronov, Khain and Seslavinski1984) and Ronov, Khain & Balukhovski (Reference Ronov, Khain and Balukhovski1989) now placed the Vinogradov maps onto plate reconstructions.

But, it was developments in Fred Ziegler's Paleogeographic Atlas Project at Chicago that would ultimately enable palaeogeography to be more effectively applied to exploration, especially in new ventures exploration. Primarily this was through the reconstruction of palaeotopography and palaeobathymetry (Ziegler et al. Reference Ziegler, Scotese, McKerrow, Johnson and Bambach1979; Ziegler, Scotese & Barrett, Reference Ziegler, Scotese, Barrett, Brosche and Sündermann1982, Reference Ziegler, Scotese, Barrett, Brosche and Sündermann1983) using the group's own plate reconstructions (Scotese et al. Reference Scotese, Bambach, Barton, Van der Voo and Ziegler1979; Rowley & Lottes, Reference Rowley and Lottes1988) and geological databases, which meant that maps could be then used as the boundary conditions for climate modelling, lithofacies retrodiction and source-to-sink analyses; second, but perhaps more importantly, the group used computerized databases for generating palaeogeographies and this led to the development of data management and visualization protocols (see Section 3.a). Ziegler's methods were developed subsequently by his students, an approach which may be referred to collectively as the ‘Chicago school of palaeogeography’.

Ziegler's palaeobathymetric methods were based on physical and biological processes, using sedimentary and fossil records (see Section 3.b.7). Palaeozoic palaeobathymetry using fossil assemblages had been the subject of Fred Ziegler's Ph.D. at Oxford with Stuart McKerrow (Ziegler, Reference Ziegler1965; Ziegler, Cocks & Bambach, Reference Ziegler, Cocks and Bambach1968) and so not surprisingly, Fred's early palaeogeographic reconstructions were for the Silurian Period (Ziegler et al. Reference Ziegler, Hansen, Johnson, Kelly, Scotese and Van Der Voo1977). For the deep-ocean Sclater had just published his palaeobathymetric maps for the Atlantic, based on plate reconstructions and age–depth relationships (Sclater, Hellinger & Tapscott, Reference Sclater, Hellinger and Tapscott1977), which were then followed by those of Thiede (Reference Thiede1979, Reference Thiede, Brosche and Sündermann1982). Putting the Sclater and Ziegler methods together provided a robust way of assessing palaeobathymetry.

The reconstruction of palaeotopography was more problematic because by definition highlands are largely above contemporary base-level and therefore are erosional (Markwick & Valdes, Reference Markwick and Valdes2004). Levorsen (Reference Levorsen1933) had reconstructed palaeotopography by mapping palaeo-geology and unconformities (see discussion in Adams, Reference Adams1943). But, the first systematic large-scale mapping of upland areas was that of Vinogradov, Grossheim & Khain (Reference Vinogradov, Grossheim and Khain1967), Vinogradov, Vereschchagin & Ronov (Reference Vinogradov, Vereschchagin and Ronov1968) and Vinogradov (Reference Vinogradov1968, Reference Vinogradov1969). In the Geological Atlas of the Rocky Mountain Region, Mallory (Reference Mallory1972) showed sediment source areas (highlands) and sediment transport pathways underpinned by well and outcrop data, including industry information. Peter Ziegler distinguished between different tectonic regimes for areas above contemporary base-level (Ziegler, Reference Ziegler1982, Reference Ziegler1990, Reference Ziegler1999). But, in all three cases, ‘highlands’ were shown as sediment source areas without explicit values of elevation. This is what Fred Ziegler and his team added by applying a uniformitarian approach using ‘typical’ present-day elevational ranges to equivalent mapped tectonic settings in the past (Ziegler et al. Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985). This divided the palaeolandscape with distinct elevational breaks at 200 m, 1000 m and 2000 m, which were then digitized.

In 1982, Judy Parrish, another of Fred Ziegler's students, used these new palaeogeographic maps as boundary conditions for parametric climate modelling in order to reconstruct past ocean and atmospheric circulation and thereby ocean upwelling (Parrish, Reference Parrish1982; Parrish & Curtis, Reference Parrish and Curtis1982; Parrish, Ziegler & Humphreville, Reference Parrish, Ziegler, Humphreville, Thiede and Suess1983). The significance of upwelling to exploration was its well-documented link with high primary productivity and the potential for export carbon production and source rocks (Brongersma-Sanders, Reference Brongersma-Sanders1967; Cushing, Reference Cushing1971; Emery & Milliman, Reference Emery and Milliman1978; Müller & Suess, Reference Müller and Suess1979; Summerhayes, Reference Summerhayes1981, Reference Summerhayes, Thiede and Suess1983; Suess & Thiede, Reference Suess and Thiede1983). The key inputs to source rock formation were by now relatively well understood with the balance between productivity and preservation, and the coincidence of known organic-rich mudstones (ORMs) with marine transgression and epicontinental seas and anoxic conditions (Schlanger & Jenkyns, Reference Schlanger and Jenkyns1976; Arthur & Schlanger, Reference Arthur and Schlanger1979; Demaison & Moore, Reference Demaison and Moore1980; Jenkyns, Reference Jenkyns1980; Schlanger et al. Reference Schlanger, Arthur, Jenkyns, Scholle, Brooks and Fleet1987; Arthur, Schlanger & Jenkyns, Reference Arthur, Schlanger, Jenkyns, Brooks and Fleet1987). This initial focus on retrodicting marine source rocks had the additional consequence that most published, global palaeogeographic maps still represent, by default, the maximum extent of the oceans for the time interval represented.

Parrish's parametric climate model was computerized by Scotese & Summerhayes (Reference Scotese and Summerhayes1986) and then further developed within and outside of exploration companies (Ross & Scotese, Reference Ross and Scotese1997). Other parametric models were subsequently also developed (Gyllenhaal et al. Reference Gyllenhaal, Engberts, Markwick, Smith and Patzkowsky1991; Patzkowsky et al. Reference Patzkowsky, Smith, Markwick, Engberts and Gyllenhaal1991). But, increasingly, palaeoclimate modelling was being pursued using more complex, physics-based computer models, GCMs (General Circulation Models). For palaeoclimatology, this research was dominated by Eric Barron and colleagues at the University of Miami and later at Pennsylvania State. Barron's initial interests were very much on how changing land–sea and topography might affect climate on geological times scales (Barron, Reference Barron1981; Barron et al. Reference Barron, Harrison, Sloan and Hay1981; Barron & Washington, Reference Barron and Washington1984), but in 1985 he began to apply these models to the retrodiction of upwelling and potential source rocks (Barron, Reference Barron1985). In 1986, Parrish and Barron were co-editors on the SEPM volume Palaeoclimates and Economic Geology (Parrish & Barron, Reference Parrish and Barron1986). This volume was key because it not only provided information on the models themselves, but also on precipitation and how this might affect coal distribution, kaolinites, bauxites and laterites and evaporites, and in the marine setting how circulation could lead to accumulations of other economic depositions (in addition to source rocks), phosphate and marine iron ore. This link between palaeogeography, palaeoclimatology and source rocks was investigated further by numerous authors leading to Huc's American Association of Petroleum Geologists (AAPG) volume on Paleogeography, Paleoclimate, and Source Rocks (Huc, Reference Huc1995).

The opportunity that Parrish and Barron's modelling work had revealed was instantly recognized in new ventures exploration. Here was a way to use relatively inexpensive palaeogeographies and climate models (at least ‘inexpensive’ relative to shooting seismic and drilling wells) to retrodict what sediments might be present in any part of the world at any time in the past. These retrodictions could then be linked with the known palaeo-position of wells, seismic and fields and other assets to then make rapid assessments of the petroleum potential of any area.

The challenge was in understanding the processes well enough to make this method quantitative and robust, but simple enough to understand and experiment with. This also necessitated an understanding of the uncertainties in palaeogeographic mapping and climate models. There was also the not insignificant problem that the climate models were global, at least initially, and global palaeogeographies took time to build, a problem long recognized (Adams, Reference Adams1943).

In the late 1990s, a team at Fugro-Robertson (now part of CGG) led by Jim Harris took up the challenge by stripping the processes to their simplest level and using Earth system models and global palaeogeographies to retrodict the distribution of source facies for a selection of Mesozoic and Cenozoic stages, which became their Merlin product (Burggraf et al. Reference Burggraf, Harris, Suter, Stronach, Huizinga, Crossley, Markwick, Ghazi, Hudson, Valdes and Proctor2006; Harris et al. Reference Harris, Crossley, Richards, Stronach, Hudson, Burggraf, Suter, Huizinga, Ghazi, Markwick, Valdes and Proctor2006, Reference Harris, Ashley, Otto, Crossley, Preston, Watson, Goodrich, Team, Valdes, AbuAli and Moretti2017). The source facies retrodictions were generated by breaking processes down into three groupings similar to the tripartite workflow published by Tyson with his processes grouped into those responsible for input, preservation and dilution (IPD) (Tyson, Reference Tyson1995, Reference Tyson2001). The palaeogeographic methodologies drew heavily from those of the Chicago school (Ziegler et al. Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985), whilst the climate modelling and data-model testing were based largely on the research of Markwick (P. J. Markwick, unpub. Ph.D. thesis, Univ. Chicago, 1996; Markwick, Reference Markwick, Williams, Haywood, Gregory and Schmidt2007). Scotese and GeoMark Research Ltd took the Merlin methodologies and developed their own system using FOAM (the Fast Ocean-Atmosphere Model) as their climate input into a product called Gandolph (Scotese et al. Reference Scotese, Zumberge, Illich, Moore and Ramos2008; Scotese, Reference Scotese2011). Subsequently, exploration companies started to develop their own systems, most of which remain unpublished. These included ExxonMobil's SourceRER system (Bohacs, West & Grabowski, Reference Bohacs, West and Grabowski2010, Reference Bohacs, West and Grabowski2011, Reference Bohacs, West and Grabowski2013; Bohacs et al. Reference Bohacs, Norton, Gilbert, Neal, Kennedy, Borkowski, Rottman, Burke, Roberts and Bally2012) and retrodictions of upwelling and source rocks at BP (Miller, Reference Miller1989; Summerhayes, Reference Summerhayes2015).

Clastic reservoirs were also modelled using similar methodologies. Heins & Kairo (Reference Heins, Kairo, Arribas, Critelli and Johnsson2007) developed a model for retrodicting the sand character in clastic reservoirs using all of the elements available through climate modelling and palaeogeography: the tectonic setting; hinterland uplift and transport; climate; accommodation space; and depositional environment. Markwick et al. (Reference Markwick, Proctor, Valdes, Wolf and Jacques2006, Reference Markwick, Campanile, Galsworthy, Raynham, Harland, Benny, Bailiff, Bonne, Hagen, Eue and Wrobel2011) used process-based modelling with explicit reconstructions of palaeo-river systems (Markwick & Valdes, Reference Markwick and Valdes2004) together with palaeo-geology, erosion and transport to predict sediment flux and character using empirical models based on databases such as those of Milliman & Syvitski (Reference Milliman and Syvitski1994; Syvitski et al. Reference Syvitski, Peckham, Hilberman and Mulder2003) and Hovius & Leeder (Reference Hovius and Leeder1998). These models were then linked to the offshore shelf transport using models of ocean currents, waves and also tides (Proctor & Markwick, Reference Proctor and Markwick2005). This approach was about following the particle from production in the hinterland by way of weathering and erosion, to deposition in the basin and encompasses what is source-to-sink analysis (Sømme et al. Reference Sømme, Helland-Hansen, Martinsen and Thurmond2009; Martinsen et al. Reference Martinsen, Sømme, Thurmond and Lunt2010, Reference Markwick, Campanile, Galsworthy, Raynham, Harland, Benny, Bailiff, Bonne, Hagen, Eue and Wrobel2011; Martinsen & Sømme, Reference Martinsen and Sømme2013; Sømme, Martinsen & Lunt, Reference Sømme, Martinsen and Lunt2013; Helland-Hansen et al. Reference Helland-Hansen, Sømme, Martinsen and Thurmond2016) (see Section 4.b).

By the 2000s, advances in climate modelling, lithofacies retrodiction and source-to-sink analysis required more detailed reconstructions of palaeogeography, especially elevation and drainage. To address this, Markwick & Valdes (Reference Markwick and Valdes2004) developed a methodology for generating palaeoDEMs (palaeo-Digital Elevation Models) which better represented the surface of the Earth with high-resolution grids that could be directly input into the climate models as boundary conditions. The method was partly based on the workflow of Lillegraven & Ostresh Jr. (Reference Lillegraven and Ostresh1988) combined with a uniformitarian approach similar to that of Ziegler et al. (Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985) to get elevational ranges. The results were then tested and modified with reference to published palaeoaltimetry data, and the evolution of elevation in adjacent timeslices. This method was then applied in various forms within industry (Markwick, Jacques & Valdes, Reference Markwick, Jacques and Valdes2005; Burggraf et al. Reference Burggraf, Harris, Suter, Stronach, Huizinga, Crossley, Markwick, Ghazi, Hudson, Valdes and Proctor2006; Harris et al. Reference Harris, Crossley, Richards, Stronach, Hudson, Burggraf, Suter, Huizinga, Ghazi, Markwick, Valdes and Proctor2006, Reference Harris, Ashley, Otto, Crossley, Preston, Watson, Goodrich, Team, Valdes, AbuAli and Moretti2017; Markwick, Wilson & Lefterov, Reference Markwick, Wilson and Lefterov2008; Raddadi, Hoult & Markwick, Reference Raddadi, Hoult and Markwick2008; Markwick et al. Reference Markwick, Raddadi, Raynham, Tomlinson, Edgecombe, Rowland, Bailiff, Galsworthy and Wrobel2009; Markwick, Reference Markwick2011b).

The last decade has seen new methods for reconstructing palaeo-elevation. These include the following: taking the present-day DEM (Digital Elevation Model) back through time and flooding the landscape using published global sea-level curves (Scotese, Reference Scotese2008, Reference Scotese2014a,Reference Scoteseb); inverting river long profiles (Roberts & White, Reference Roberts and White2010; Paul, Roberts & White, Reference Paul, Roberts and White2014; Rudge et al. Reference Rudge, Roberts, White and Richardson2014; Wilson et al. Reference Wilson, Roberts, Hoggard and White2014); heuristic models of palaeotopography and palaeobathymetry based on tectonic processes (Vérard & Hochard, Reference Vérard and Hochard2011; Vérard et al. Reference Vérard, Hochard, Baumgartner and Stampfli2015); and quantitative, dynamic modelling of mantle-driven topographic change as ‘dynamic topography’ (Moucha et al. Reference Moucha, Forte, Mitrovica, Rowley, Quéré, Simmons and Grand2008; Heine et al. Reference Heine, Müller, Steinberger and DiCaprio2010; Rowley et al. Reference Rowley, Forte, Moucha, Mitrovica, Simmons and Grand2013; Flament et al. Reference Flament, Gurnis, Müller, Bower and Husson2015).

These are all powerful tools but must be constrained and related back to the geology, and so one of the main goals of palaeogeographic mapping is to map the observations which can then be used to test the modelling results (Section 3.b.7).

3.a. ‘Big data’, data management and mapping reliability

One of the main drivers for palaeogeography in the nineteenth century was as a way of managing, visualizing and analysing the increasing volume of geological knowledge stored in museums and libraries (Markwick, Reference Markwick, Duffin, Gill, Howell, Smyth and Sides2018). This is still true today as embodied in the concept of ‘big data’, a result of the explosion in data due to the improved accessibility via the Internet. Managing all this geological information through palaeogeographic maps requires a comprehensive audit trail and easy to understand representations of mapping (interpretation) and data confidence. This is critical for explorationists who want to use the generated maps to re-risk their exploration opportunities and was central to the workflow developed at Chicago (Ziegler et al. Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985). Ziegler's solutions remain elegantly simple and have continued to be used by numerous groups. His group's databases of lithological, tectonic and palaeontological information were underpinned by a comprehensive reference database and library (comprising almost 30,000 references by 1995), stratigraphic summaries, scrapbooks of key figures and a hand-filled map book that showed the coverage of all data in the databases to guide future searches.

Assigning data confidence (reliability) was, and is, more problematic because it can be qualitative and reflects the consequences of three factors: (1) how ‘good’ is the data? (2) how ‘good’ is our understanding and interpretation of that data? and (3) the nature of the geological record. Ziegler et al. (Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985) accomplished this by using a simple alphanumeric code for age dating and mapping confidence; to these codes were added a measure of the geographic precision with which data were located (P. J. Markwick, unpub. Ph.D. thesis, Univ. Chicago, 1996; Markwick & Lupia, Reference Markwick, Lupia, Crame and Owen2001). An example of the latest version of this confidence table, as applied to the maps in this paper, is shown in Table 1.

Table 1. Explanation of confidence codes used in the palaeogeographic mapping workflow

* Based on the scheme described in Ziegler et al. (Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985)

Confidence can also be indicated graphically using different colours, shading or fill symbology on the maps. The solution used here is shown in Figure 2 and is based on that of Vinogradov, Grossheim & Khain (Reference Vinogradov, Grossheim and Khain1967), Vinogradov, Vereschchagin & Ronov (Reference Vinogradov, Vereschchagin and Ronov1968) and Vinogradov (Reference Vinogradov1968, Reference Vinogradov1969), who used symbology to differentiate between the observed (preserved) depositional environments and their lithologies and the former (eroded or non-depositional) extent of those depositional environments. This approach was adopted by Ziegler et al. (Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985) and Markwick & Valdes (Reference Markwick and Valdes2004) and then applied by them across the exploration industry. Data density is typically shown graphically by mapping the location of data, whether point localities such as wells or outcrops, or lines such as seismic or sections. Map annotation can also be used (Mallory, Reference Mallory1972) (see map examples in Sections 4.b and 4.c).

Figure 2. The use of fill patterns to show different levels of mapping confidence based on the mapping methods of Vinogradov, Grossheim & Khain (Reference Vinogradov, Grossheim and Khain1967), Vinogradov, Vereschchagin & Ronov (Reference Vinogradov, Vereschchagin and Ronov1968) and Vinogradov (Reference Vinogradov1968, Reference Vinogradov1969). Increasing mapping confidence is indicated by the addition of lithological symbology using different pattern colours.

Today this can all be accomplished digitally, including more extensive explanations of data provenance and interpretations. This process has been greatly facilitated by developments in spatial databasing, especially GIS. All of the examples shown in this paper have been built and visualized (see also online Supplementary Material available at http://journals.cambridge.org/geo) using ArcGIS (ESRI, 2017). The resulting palaeogeographies can then be ported to other software.

3.b. The palaeogeographic workflow and representation

The full palaeogeographic workflow is shown schematically in Figure 3. Each component is briefly described in the following sections and represented by a series of block diagrams that build to show how the order in which these components are put together in a palaeogeography reflects the way the Earth works to build a landscape (Fig. 4). In practice, building a palaeogeography is an iterative process.

Figure 3. The palaeogeographic mapping workflow described in this paper. Only the top two levels of the workflow are shown. This follows the structure of the Earth by building up from the crustal architecture through the basin fill to drainage and landscape following ideas originally presented in the late nineteenth and twentieth centuries. This is an iterative process.

Figure 4. A block diagram showing the relationship of a hypothetical landscape to the underlying basin form, hinterland form and basin fill. This brings together all the components of basin dynamics, basin analysis and source-to-sink analysis.

The output from each stage of the workflow has value in its own right for exploration.

3.b.1. Crustal architecture

The crustal architecture, comprising the structural framework and crustal type (composition, thickness and origin), is critical to get correct because it is upon this foundation that all the other elements of the palaeogeography are built. The structural framework partitions and defines the crustal architecture, determines how rifts develop, and through this dictates the distribution of depositional play elements (input points for sediment) in rift basins, defines the shape of compressional systems, dictates the pattern (Twidale, Reference Twidale2004) and position of rivers through time (and their changes as the structures evolve), defines the position of canyons and submarine transport pathways, provides conduits or barriers for fluid flow, and acts as hydrocarbon traps.

The structural and tectonic elements are captured using seismic, remote sensing data (Landsat, radar, gravity, magnetics), published literature, geological mapping and primary observations. Features are presented on the maps using standard symbologies, which are largely based on those of the USGS (United States Geological Survey; see online Supplementary Material available at http://journals.cambridge.org/geo). Each feature is underpinned by a detailed audit trail to constrain the kinematics of the feature at the time of the map, the provenance of source information, mapping and age dating confidence and activity history where available. The significance of the feature in the landscape is defined by assigning each structural element to a class that reflects its effect on the crust. Features are also differentiated based on whether they are inferred (no evidence in the data but required to explain the overall kinematics of any area) or defined (a feature has an expression in primary data). Structures need not be active at the time of a past landscape (palaeogeography) to influence it. Today, rivers such as the East African Lurio River, or West African Ntem, Sanaga and Nyong rivers, all flow along the paths of inactive, Precambrian shear zones. Consequently, whether a feature is active or inactive at the time of the map also needs to be recorded.

The crustal thickness, composition (crustal type) and age dictate the strength of the crust (Watts, Reference Watts1982; Watts, Karner & Steckler, Reference Watts, Karner and Steckler1982; Cloetingh et al. Reference Cloetingh, Catalano, D'Argenio, Sassi and Horvath1999, Reference Cloetingh, Burov, Matenco, Beekman, Roure and Ziegler2013; Burov & Watts, Reference Burov and Watts2006), and how the crust responds to geodynamic processes. This is also a function of the past history of the crust through inherited weaknesses (Ashwal & Burke, Reference Ashwal and Burke1989), and the age itself (Mouthereau, Watts & Burov, Reference Mouthereau, Watts and Burov2013). Crustal thickness and composition can be described separately, but in the absence of global observations, they are frequently summarized as (large-scale) crustal type (Fig. 5), which can be defined by remote sensing data, such as gravity and magnetics, seismic, well data or field data. In this workflow, observations (the description of the nature of the crust) are distinguished from the processes acting on the crust (geodynamics, thermo-mechanical events). The classification for rifting processes presented here is based partly on that of Manatschal and co-workers (Péron-Pinvidic & Manatschal, Reference Péron-Pinvidic and Manatschal2010; Unternehr et al. Reference Unternehr, Péron-Pinvidic, Manatschal and Sutra2010; Manatschal, Reference Manatschal2012).

Figure 5. The classification of crustal types used in the palaeogeography mapping workflow. (a) The crustal architecture underpinning the schematic landscape in Figure 4 showing conceptually the relationship of the different crustal types to the landscape. (b) The classification used in this workflow to describe the observed crustal type, its petrology and geometry. Subdivisions of each crustal type are defined where they influence accommodation space, palaeobathymetry and palaeotopography, and can be recognized using remote sensing data and/or seismic. The full legend is provided in the online Supplementary Material available at http://journals.cambridge.org/geo.

3.b.2. Geodynamics

Crustal type dictates first-order heat flow and elevation, but the key for palaeogeographic reconstruction is how geodynamics then acts on this crust. This process is represented by mapping the age and nature of the last thermo-mechanical event with respect to the palaeogeographic timeslice being reconstructed (Fig. 6). This method was first discussed in Markwick and Valdes as ‘tectonophysiography’ (Markwick & Valdes, Reference Markwick and Valdes2004) with regards to defining areas above contemporary base-level and therefore areas of net erosion (sediment source areas in source-to-sink analysis). The age of the last thermo-mechanical event was added to better represent the decay of landscapes (Tucker & Slingerland, Reference Tucker and Slingerland1994; Van der Beek & Braun, Reference Van der Beek and Braun1998; Pazzaglia, Reference Pazzaglia2003; Whipple & Meade, Reference Whipple and Meade2004; Campanile et al. Reference Campanile, Nambiar, Bishop, Widdowson and Brown2007) following the ideas presented in the 1997 USGS thermo-tectonic age map of the world that was used to model heat flow following Pollack, Hurter & Johnson (Reference Pollack, Hurter and Johnson1993) and crustal thickness and structure (Mooney, Laske & Masters, Reference Mooney, Laske and Masters1998). An updated scheme for tectonophysiographic terrains was then presented by Markwick and co-workers in several presentations (Markwick, Wilson & Lefterov, Reference Markwick, Wilson and Lefterov2008; Raddadi, Markwick & Hill, Reference Raddadi, Markwick and Hill2010; Galsworthy et al. Reference Galsworthy, Raynham, Markwick, Campanile, Bailiff, Benny, Harland, Eue, Bonne, Hagen, Edgecombe and Wrobel2011; Markwick, Reference Markwick2011b; Markwick, Galsworthy & Raynham, Reference Markwick, Galsworthy and Raynham2015).

Figure 6. Classification of the thermo-mechanical state (geodynamics) of the crust. (a) A block diagram showing schematically the relationship of the geodynamic classification used to generate the landscape shown in Figure 4. The crustal type (Fig. 5) defines the state of the crust upon which geodynamics acts, which is here represented by the age and type of the last thermo-mechanical event acting on that crust. The combination of the two builds the palaeolandscape to give the tectonophysiography (Markwick & Valdes, Reference Markwick and Valdes2004). (b) The different types of thermo-mechanical event (extension, compression, vertical) operate at a variety of scales spatially and temporally, which have a direct expression in the landscape. This is why each process is mapped to reflect its age of activity relative to the age of the landscape being reconstructed. The full legend is provided in the online Supplementary Material available at http://journals.cambridge.org/geo.

3.b.4. Depositional environments

The mapped depositional environment in this workflow represents the full spatial extent of each environment at the time of the palaeogeography (see Section 3.c). The extent represents areas below contemporary base-level and potentially able to accumulate sediments (Markwick & Valdes, Reference Markwick and Valdes2004), but will also include areas with no deposition due to by-passing. Base-level and depositional environments can vary rapidly in time and space, especially in tectonically active areas, and this must be considered (see Section 3.c and Section 4.c). Strictly speaking, depositional environment maps are distinct from a facies map, the latter representing the products of deposition (the rock record); a palaeogeography in this definition would show a submarine canyon system, slope, rise and abyssal plain as environments, but not refer to a ‘turbidite’ depositional environment, which is a facies. With digital systems, these can be kept separate and overlain later during analysis. In reality, many published palaeogeographies ‘mix’ facies, gross depositional environments (GDEs) and depositional environments, and users should be aware of this.

The classification and representation of depositional environments in palaeogeography is relatively standard and shown in Figure 7 (the full legend is given in the online Supplementary Material available at http://journals.cambridge.org/geo). Marine environments are generally classified by depth following Ziegler et al. (table 2 in Ziegler et al. Reference Ziegler, Rowley, Lottes, Sahagian, Hulver and Gierlowski1985), who related environments to sedimentological and fossil evidence. Similar definitions were used by Vinogradov, Grossheim & Khain (Reference Vinogradov, Grossheim and Khain1967), Vinogradov, Vereschchagin & Ronov (Reference Vinogradov, Vereschchagin and Ronov1968) and Vinogradov (Reference Vinogradov1968, Reference Vinogradov1969), and to lesser or greater degrees by Scotese (Scotese & Golonka, Reference Scotese and Golonka1992; Reference ScoteseScotese, 2014a,Reference Scoteseb), Golonka (Golonka et al. Reference Golonka, Ross, Scotese, Embry, Beauchamp and Glass1994; Golonka, Reference Golonka, Spencer, Embry, Gautier, Stoupakova and Sørensen2011), Peter Ziegler (Ziegler, Reference Ziegler1982, Reference Ziegler1990), Markwick (Markwick et al. Reference Markwick, Rowley, Ziegler, Hulver, Valdes and Sellwood2000, Reference Markwick, Galsworthy and Raynham2015; Markwick & Valdes, Reference Markwick and Valdes2004; Markwick, Reference Markwick, Williams, Haywood, Gregory and Schmidt2007, Reference Markwick2011b), and most other palaeogeographers. For coastal and near-shore environments this paper follows the definitions in Boyd, Dalrymple & Zaitlin (Reference Boyd, Dalrymple and Zaitlin1992) given the importance of such environments to reservoir systems (Armentrout, Reference Armentrout, Beaumont and Foster2000). This detailed level of division is most appropriate for high-resolution maps at the basin to play scale. Delta tops also fall into the group of coastal or near-shore environments. Terrestrial environments have been divided into fluvial, lacustrine, glacial and desert environments (Fig. 7). Subdivisions to allow differentiation between meandering and braided fluvial systems, different lacustrine types (based on salinity) and alluvial as distinct from fluvial systems, have been included given the needs of source-to-sink analysis and climate modelling.

Figure 7. (a) The representation of depositional environments that constitute the landscape in Figure 4. In map view, this will define the basin fill for those areas below contemporary base-level. (b) The classification for terrestrial and marine depositional environments used in this workflow. The full depositional legend, including the detailed coastal environments, is provided in the online Supplementary Material available at http://journals.cambridge.org/geo.

Table 2. Pros and cons of global-scale palaeogeographic maps in exploration

3.b.6. Transport pathways: palaeo-rivers and canyons

Rivers are the primary onshore transport pathways providing clastic sediments to the shelf where sediments can be redistributed. Sediment transport is now encompassed within source-to-sink analysis (Sømme et al. Reference Sømme, Helland-Hansen, Martinsen and Thurmond2009; Martinsen et al. Reference Martinsen, Sømme, Thurmond and Lunt2010 Reference Martinsen, Sømme, Thurmond, Skogseid, Lunt, Leith and Helland-Hansen2011; Martinsen & Sømme, Reference Martinsen and Sømme2013; Sømme, Martinsen & Lunt, Reference Sømme, Martinsen and Lunt2013; Helland-Hansen et al. Reference Helland-Hansen, Sømme, Martinsen and Thurmond2016). As well as affecting reservoir systems, nutrients from rivers can affect local productivity; freshwater discharge can result in (especially seasonal) water column stratification and anoxia, or conversely, clastic sediments from rivers can dilute organic carbon accumulations and source rock quality. Palaeo-rivers are a key boundary condition in most of the latest GCMs.

Markwick & Valdes (Reference Markwick and Valdes2004) provided an overview of the methodologies used to construct the palaeo-rivers in their Maastrichtian palaeogeography, which are represented schematically in Figure 8. The methodologies can be applied at all map resolutions, but which method is most appropriate within this workflow will depend on the age of the palaeogeographic timeslice and the data available; drainage network analysis is used to work back through time, and analogues and geological data are used to work forward through time. A worked example is presented as a case study for West Africa in Section 4.b (see also Markwick, Reference Markwick2013).

Figure 8. Basic palaeodrainage workflow. The reconstruction of palaeodrainage is generated forward from the past (conceptual models and palaeogeography) and back in time from an analysis of the present day. The term ‘network analysis’ encompasses the analyses of drainage patterns, drainage morphometry (shape) and hypsometry (long profiles). How far back or forward each method can be applied depends on the drainage basin and underlying geological history. A key difference in this workflow, compared with other palaeogeographic methodologies, is that the palaeodrainage is an input into the reconstruction of the landscape (palaeoDEM) and not an automated output from the palaeoDEM (for more details of the workflow see Markwick & Valdes, Reference Markwick and Valdes2004).

Fundamental to this workflow is the drainage network analysis, which establishes the hypotheses that are then tested by the geology (provenance and palaeobiogeography) and tectonic and depositional mapping. This analysis of the present-day drainage networks and landscape follows the maxim of Summerfield, ‘that landscapes inherit’, and uses geomorphological geometric relations (‘laws’) established in the 1960s by geomorphologists including Strahler (Reference Strahler1952, Reference Strahler1957, Reference Strahler and Chow1964) and Horton (Reference Horton1932, Reference Horton1945), and summarized by Twidale (Reference Twidale2004). Analysis of the present day provides indications of disjuncts (from ‘equilibrium’) that may be due to changes in the network at some point in the past. This provides hypotheses that can be tested with other data, including fish DNA and palaeobiogeography, geology, sedimentological palaeo-current directions and provenance data. This then provides potential explanations (causation and age) that can be used to retrodict the river systems back through time.

The palaeogeography itself also provides additional information, because, by definition, the mapping of depositional systems and tectonics shows the overall downslope direction in this methodology (Markwick & Valdes, Reference Markwick and Valdes2004). The landscape analysis relates to the tectonics through the hypsometry (Fig. 9) and river long profiles. These provide an indication of the discrepancy today from base-level (the ‘equilibrium’ profile). These curves can also be constructed for any area through time, as a check on the response of the palaeogeography as drawn to the mapped tectonics. This is a major strength of the overall palaeogeographic workflow described in this paper.

Figure 9. The relationship of landscape hypsometry to tectonics and palaeogeography. Strahler (Reference Strahler and Chow1964) recognized different developmental stages in the landscape that are expressed by the hypsometry. As landscapes decay, they will seek to reach an ‘equilibrium’ state with no net erosion or deposition. This is the local base-level which is explicitly defined on the palaeogeographies in this workflow following Markwick & Valdes (Reference Markwick and Valdes2004). This can be applied to any palaeogeographic landscape as a check on the continuity and evolution of palaeolandscapes. The volume change in 3D indicated by the hypsometries of adjacent palaeogeographic timeslices provides an indication of eroded volume when corrected for isostatic factors.

The older the palaeogeography represented, the less of that history is preserved in the present-day landscape. This varies depending on the stratigraphic and tectonic activity of the area. Some parts of Africa are purported to preserve early Mesozoic landscapes (King, Reference King1950; Partridge & Maud, Reference Partridge and Maud1987), but areas of active tectonics, such as SE Asia, may only record the latest Cenozoic period. In NW Europe, the record of rivers of areas that were flooded during Cretaceous and early Cenozoic times are most likely ‘young’: transgressions tend to reset the drainage networks.

In the offshore, submarine canyons provide an important conduit for transporting sediment off the shelf and into the deep water and are especially important for predicting deep-water sandstone turbidites. Present-day positions, constrained by seismic, can be used to look at the longevity of canyons through time. Some, but not all canyons have a structural influence.

3.c. Scale, resolution and the nature of the geological record: implications for exploration

In his summary of the 1908 AAAS palaeogeography discussions, Willis drew attention to the fact that each palaeogeographic map covered a long period of time within which features changed, which meant that palaeogeographies ‘do not truly represent any special geographic phase of the continent’ (Willis, Reference Willis1909, p. 204). This was further discussed by Schuchert (Reference Schuchert1910), who stated that unless a palaeogeography was temporally finely constrained, ideally to the level of a stratigraphic formation (Schuchert, Reference Schuchert, Dana, Schuchert, Gregory, Barrell, Smith, Lull, Pirsson, Ford, Sosman, Wells, Foote, Page, Coe and Goodale1918), it would not be of much use. Kay (Reference Kay1945) was more explicit about what an ‘ideal’ palaeogeography should represent in time: ‘the ideal paleogeographic maps are of a moment in the past’. This idea of a palaeogeography as a ‘moment in time’ draws a distinction with palaeogeographies as representations of an interval of time or a rock unit, which is more akin to a facies map. The problem is that palaeogeographies can rarely represent a moment in time, if ever, because of the nature of the geological record (Schuchert, Reference Schuchert1928; Markwick & Lupia, Reference Markwick, Lupia, Crame and Owen2001; Markwick & Valdes, Reference Markwick and Valdes2004), especially as larger geographic extents are considered. This was summarized in the concept of the ‘palaeontological uncertainty principle’ in which the temporal precision of an event in the geological record decreased as the event was considered over larger and larger areas (Markwick & Lupia, Reference Markwick, Lupia, Crame and Owen2001). Other authors have challenged this view of global correlation and argued that a eustatic interpretation of sequence stratigraphy provides a way of correlating sequence boundaries globally and building global palaeogeographies that approximate a ‘moment in time’ at intervals of millions of years. That level of temporal resolution requires glacio-eustacy (Haq, Hardenbol & Vail, Reference Haq, Hardenbol and Vail1987, Reference Haq, Hardenbol, Vail and Wilgus1988; Rowley & Markwick, Reference Rowley and Markwick1992; Simmons, Reference Simmons2011). But, as numerous authors have stated, this is not tenable, going back to Miall's original critique (Miall, Reference Miall1986), and the lack of evidence for glacio-eustacy during large parts of the Phanerozoic Eon (Markwick & Rowley, Reference Markwick, Rowley, Pindell and Drake1998), but also the fact that the proposed sea-level changes of 10–20 m can easily be swamped by local effects. These local effects include margin flexure, heterogeneous sediment supply along a margin, differential subsidence and dynamic topography. Recent high-resolution mantle modelling has demonstrated how short-wavelength (10s–100s km), high-amplitude changes in palaeo-elevation of up to 60 m Ma⁻¹ (around the Norfolk Arch in Rowley's study) migrate through time affecting different parts of the crust (Rowley et al. Reference Rowley, Forte, Moucha, Mitrovica, Simmons and Grand2013). These rates of change exceed the rate of change indicated by the long-term global sea level (Haq & Al-Qahtani, Reference Haq and Al-Qahtani2005), with the logical inference that even over short-distances on margins, correlation of sequence boundaries needs to be made with care.

If global and, potentially, regional-scale palaeogeographic maps do not represent ‘a moment in time’ (Kay, Reference Kay1945), then how can they be used in exploration or as boundary conditions for Earth system models? Three arguments need to be considered: (1) the position in the exploration workflow requires different palaeogeographic resolutions and this dictates the temporal resolution needed; (2) palaeogeographic variability and uncertainty are investigated using modelling experiments; and (3) palaeogeographic processes and features have different response times.

The resolution required of a palaeogeographic map is dictated by its application and this, in turn, reflects where in the exploration cycle the map is being used (Fig. 1). For global explorationists, concerned with high grading areas to identify new opportunities, the fact that a global map does not represent an instant in time may not be as significant as it would be if the same were true of a prospect map for an asset geologist working up that prospect. For example, in global exploration, it is not usually about trying to map the distribution and reservoir properties of every individual turbidite channel on a global scale but knowing the probability that such a depositional environment might exist at a particular time and place, and through analogues what that prospectivity might be. Key is understanding what that variability within a ‘timeslice’ is, and what effect it has for exploration pertinent processes and systems.

The variability and uncertainty in the palaeogeographies, and how this might impact geological risk in exploration, can be investigated by running modelling experiments that use different palaeogeographic boundary conditions: for example, running multiple climate modelling experiments to understand the response (sensitivity) of depositional play elements (source, seal and reservoir facies) to changes in hinterland relief or palaeobathymetry. There is a growing portfolio of modelling experiments that have already examined the response of the climate system to palaeogeography (Lunt et al. Reference Lunt, Farnsworth, Loptson, Foster, Markwick, O'Brien, Pancost, Robinson and Wrobel2016).

Some features in a palaeogeography are more long-lived than others and may not change ‘significantly’ within a stratigraphic stage or over several stages. This variability in the response of processes and geographic features in the landscape has been long known in landscape dynamics (Fig. 10). This shows that many tectonic features are long-lived, which means that for much of a palaeogeography, the hinterland is probably relatively stable. In contrast, other geographic features may have shorter response times, such as some flooding events in epeiric seas. This is where the sensitivity experiments can be focused. The significance of this operationally depends on where in exploration you are. As an example, in the central Pyrenees (Section 4.c), the early Eocene (Ypresian) palaeogeography is dominated by the Axial zone in the north as highlands, which also show up on global palaeogeographies and is resolved by most climate models. But, in detail, the depositional story is also affected by the southward motion of a series of thrust sheets with related uplifts that changes the geometry of the source area and depositional basins, which in turn affects the response of deep-water sandstone turbidites. The thrust sheets are below the resolution of global climate models, but the overall resulting climate patterns can be superimposed on this high-resolution mapping and at least first-order conclusions made: for example, is this area susceptible to high precipitation and flood events, as indicated by the sedimentology of the Castissent Formation, or is it arid, as indicated by the red beds of the uppermost Cretaceous and Paleocene? These are resolvable at a global scale and can be applied up through the higher resolutions.

Figure 10. Temporal resolution of different processes and features in palaeogeography. Different features in the landscape have different response times and can remain unchanged in the landscape for different intervals of time (longevity of features). In this figure, a range of types of palaeogeographic features is shown. The mean, minimum and maximum lengths of Mesozoic and Cenozoic stages are included for comparison to illustrate which features will vary within a global map duration, which is usually a stratigraphic stage. The duration of ‘climate’ (30 years), human evolution and the Aptian stage are shown for comparison. Drainage features can vary on a variety of scales depending on the degree of structural control and tectonic setting; river mouths for rivers with broad floodplains can migrate in a single event (hours or days).