Seven lithofacies are grouped in this association because the facies are stacked upon each other and jointly show a fining upward trend (Supplementary Figure S2A). The nine lithofacies of FA1 are characterized by, respectively, quartzose sandy conglomerates that show distinct scouring and that are occasionally discontinuous (lithofacies Gp), tabular cross-bedded gravel-bearing coarse-grained sandstones (Sp), trough cross-bedded and horizontally laminated medium-grained sandstones (St), cross-bedded fine-grained sandstone (Sr), horizontally laminated siltstones (Sh), coal or coaly beds (C), and gray, horizontally laminated mudstones (Fm) (Supplementary Figure S2A). Lithofacies Gp and Sp, which form the lowest part of FA1, consist of moderately sorted and poorly rounded particles. Lithofacies St and Sr have, in contrast, well sorted and rounded sand grains. Facies Sh and Fm overly Sr and contain fairly frequent plant fossils. Lithofacies C is about 0.3–0.4 m thick and forms the top of FA1.

The interpretation of these lithofacies is simple, as they have been described frequently in studies from successions ranging from the Carboniferous to the Neogene. They are characteristic of a meandering river system. Lithofacies Gp is common in the channels of a meandering river and in the distributary channels of a delta (Lowe, 1982). Lithofacies Sp typically suggests an environment with high-energy currents. Lithofacies St and Sr are characteristic of point bars. The plant fossils indicate weak currents and settling from suspension, which is common for a flood plain. Consequently, all lithofacies fit in the system of a meandering river depositional environment.

The seven lithofacies of this association are characterized, respectively, by trough and tabular cross-bedded and horizontally laminated medium-grained well-sorted sandstones with rounded grains (St), trough and tabular cross-bedded fine-grained well-sorted sandstones with rounded sand grains (Sr), horizontally laminated siltstones (Sh), gray, horizontally laminated mudstones (Fm), intercalated gray sandstones and grayish-white mudstones (FI) with plant fossils, rooted clay (Fr), and coal beds (C) (Supplementary Figure S2B). Compared with association FA1, the grain size is smaller. Facies Fm and C, which form the lowermost facies of FA2, contain plant fossils. These facies are overlain by facies Sh, which contain a number of thin-layered coal seams in their top parts. The topmost facies of FA2 are facies Fm and C. Facies C is locally intercalated within rooted clay.

The interpretation of the depositional environment of association FA2 is a lacustrine delta. Facies Fm and C with their numerous plant fossils indicate low-energy conditions well comparable to those of a prodelta, deep-marine or lacustrine environment (Reading, 1996; Diessel et al., 2000a; Diessel et al., 2000b). The horizontally bedded facies Sh has been interpreted by Miall (1996) as characteristic of an interdistributary bay. The combination with facies Sr suggests deposition as a mouth bar (Mjos and Prestholm, 1993). Facies St and Sr with their tabular and trough cross-bedding and scours, represent distributary channels (Mjos and Prestholm, 1993; Hwang et al., 1995; Ielpi, 2012). Facies C and Fm at the top of the association indicate a peat-bog environment (Diessel, 1992a; Diessel et al., 2000b; Edress et al., 2021; Guatame and Rincón, 2021). The vertical fine-grained succession of association FA2 thus represents a lacustrine deltaic environment.

The sequence boundaries of the Yan’an Formation in the study area are represented by regional unconformities and abrupt changes of the facies, due to the (also abruptly changing) depositional conditions (Supplementary Figure S3). It appears that three such unconformities can be distinguished (from bottom to top sequence boundaries SB1 to SB3), thus bounding (see Section 4.3.2) two complete third-order sequences (S-1 and S-2).

The lowermost sequence boundary (SB1) is the unconformity that forms the boundary between the Yan’an Formation and the underlying Yanchang Formation. This sequence boundary forms the base of sequence S-1 (Supplementary Figure S3A), which starts with facies Gp and Sp. The quartz grains are oriented more or less horizontally, parallel lamination and cross-bedding occur frequently, and erosion by the currents in the river is distinctly present (Supplementary Figure S1A). Sequence S-1 is bounded at its top by sequence boundary SB2. The transition from the coarse top of sequence S-1 to sequence S-2 is lithologically abrupt: the lower part of sequence S-2 is dominated by silty mudstones. The upper part of sequence S-2 gradually changes into coarse and thick sandstones. The channel erosion is distinct (Supplementary Figure S3C). Sequence S-2 is bounded at the top by sequence boundary SB3, which is the unconformity between the Yan’an Formation and the overlying Zhiluo Formation. This unconformity represents an incised valley (Supplementary Figure S3B).

System tract boundaries are controlled by changes of the base-level (Van Wagoner et al., 1988), and are represented by initial flooding surfaces (IFS) and maximum flooding surface (MFS).

IFS can be identified because of the transition of the succession from progradational to retro gradational (Supplementary Figure S3D), which indicates that a (here: lacustrine) regression changed into a transgression (Catuneanu, 2006; Catuneanu et al., 2009; Li et al., 2018). This development results in a distinct lithofacies change (Figures 2, 3). The beginning of the lacustrine transgression is reflected by a lower part of the succession that commonly consists of fluvial sandstones (Supplementary Figure S3D) and an upper part that is dominated by deltaic sand- and mudstones.

The MFS are recognized by their change from retrogradation to progradation (Supplementary Figure S3E), marked by a lake shoreline trajectory transition from an extended (lake) system tract (EST) to a highstand system tract (HST) (Shanley and McCabe, 1994; Plint et al., 2001; Catuneanu, 2006). The lower part of the MFS of the Yan’an Formation consists of mudstones and a coal seam, whereas the upper part consists of sandstones (Supplementary Figure S3E). The MFS is often accompanied by a condensed layer that represents the end of the lake transgression (Figures 2, 3).

Based on the analysis of the sequence boundaries, the Yan’an Formation can be subdivided into two third-order sequences: S-1 and S-2 (Figures 2, 3). The system tracts were identified by the key surfaces and the sedimentary stacking patterns (Catuneanu, 2002; Catuneanu, 2020), including the lowstand systems tract (LST), extended (lake) system tract (EST), and highstand system tract (HST).

Inertinite is the most abundant maceral group with a content varying from 29.85% to 73.22% by volume (Table 3). The content decreases initially upwards, but then increases again (Figure 9A). The vitrinite content varies from 20.06% to 64.29% by volume (Table 3); it increases initially upwards, but then decreases again (Figure 9A). The vitrinite content is lowest in coal seam 6-2z, and maximum in seam 4-1. The liptinite content varies from 2.47% to 9.45% by volume.

The mineral content in the coals varies from 1.45% to 4.05% by volume (Table 3). Like the liptinite, the mineral content is relatively low and relatively stable.

The GI value, which varies from 0.23 to 9.24, increases initially and then decreases upwards (Supplementary Table S1). This index is relatively high for the 4-1 coal seam. GI indices higher than 1 are found for coal seams 6-1, 5-1, 4-1, and 3-1 coal. These >1 values indicate relatively humid conditions.

The TPI values vary from .32 to 9.35 and are mostly higher than 1 (Supplementary Table S1), indicating that the plant tissues are well-preserved and that xylophytes form a major part.

The vitrinite and inertinite contents in coal are commonly used to reconstruct the depositional environment of the vegetation (Diessel et al., 2000a; Diessel et al., 2000b): vitrinite is formed from the roots and leaves of plants in a reducing environment covered with water, whereas inertinite is formed by the roots and leaves of plants under relatively dry oxidizing conditions. A high content of vitrinite in the coal samples under study thus indicates that the conditions were relatively wet, and that accumulation could take place because the accommodation space increased; in contrast, a high inertinite content indicates that the conditions were relatively dry (Guatame and Rincón, 2021). In other words: the ratio between vitrinite and inertinite (V/I ratio) is a proxy for the degree of reducing conditions to which the peat was exposed (Harvey and Dillon, 1985). If V/I is less than 1, the peat is considered to have accumulated in an oxidizing environment.

The ratio between the increase in accommodation space and the peat production (commonly indicated as the AR/PPR ratio) is also an important index to evaluate the depositional conditions (Bohacs and Suter, 1997; Diessel et al., 2000a; Diessel et al., 2000b; Liu et al., 2020; Wang et al., 2020). The V/I ratio and the GI and TPI indices of the various coal seams, as well as the trends of the vitrinite and inertinite, are shown in Figure 9A and Table 4. More details about the characteristics of the macerals in the six coal seams under study are presented in the following subsections.

As indicated above, the sediment influx and, particularly, the climate conditions and basin subsidence appear to have controlled the lake level, and, consequently, whether the lake expanded or shrank. The influence of the climate is supported by the maceral variation: the coal seams 6-2z, 6-1, and 4-1 formed in the EST of both sequences (Figure 10). The conditions changed from dry to wet during expansion of the lake, which is consistent with EST characteristics. However, the 5-1, 3-1, and 2-2 coal seams formed during the HST of both sequences. The conditions then changed from wet to dry, during shrinking of the lake, which is consistent with HST characteristics. This shows that the fluctuating lake level was the main factor controlling the sequence stratigraphy, and this must have played an important role in the coal formation. Possibly the fluctuations in the lake level were influenced to some degree by climate changes (resulting in e.g., more or less precipitation and consequently more or less inflow of water, more or less evaporation, etc.), but it is physically impossible that the changing meteorological conditions caused the development of the entire succession of the Yan’an coal measures.

Nevertheless, Liu and Li (2007) proposed that the climate was the main factor to control short-lived sequences without sedimentary hiatuses during tectonically quiet stages of lacustrine environments. However, Zhang et al. (2020) have found that astronomical forcing determined the development of a Paleogene coal-bearing succession, and deduced that climate was the controlling factor of these coal-bearing successions. Moreover, Zhang et al. (2021a) found that the coal of the Yan’an Formation developed during time-spans of eccentricity minima in the 405-ka cycle (Figure 10), whereas eccentricity maxima in these cycles are characterized by detrital sedimentation. During phases of large eccentricity, the differences between seasons are large, which, according to Zhang et al. (2021b), would result in increased river discharge and sediment supply, even though this is not supported by actual data. Moreover, phases of limited eccentricity would, gain according to Zhang et al. (2021a), result in limited seasonal differences and in a stable climate, much vegetation, and reduced sediment supply, whereas the maximum long eccentricity would commonly have been accompanied by a short phase of intensive rainfall and a perennial drought climate, with significant seasonal differences. The time-span of long eccentricity minima must, following Zhang et al. (2021b) usually have been warm and humid, with moderate seasonal variation, which favored vegetation growth and thus was suitable for peat accumulation. Zhang et al. (2021a) therefore consider it likely that the climate influenced the sequence development of the Yan’an coal measures as well as the coal formation. Unfortunately, however, they do not pay attention to the process that must have created sufficient accommodation space for the entire coal-measures succession.

Ever since astronomical factors have been found to affect sedimentation on Earth, numerous studies (Berger et al., 1992; Liu and Li, 2007; Fang et al., 2017; Huang et al., 2020; Zhang et al., 2020; Zhang et al., 2021b; Huang et al., 2021; Zhao et al., 2022) emphasize the role of astronomically-induced climate changes on the development of specific successions, including coal measures of different ages and in different parts of the world (Cecil et al., 1985; Cecil, 1990; Fielding and Webb, 1996; Miller and Eriksson, 1999; Pashin and Gastaldo, 2004; Izart et al., 2005; Liu and Li, 2007; Falahatkhah et al., 2021; Zainal Abidin et al., 2022). This is remarkable since another factor, viz., basin subsidence, tends to be completely overlooked, even though the successions could not have developed without such subsidence. The Yan’an Formation in the study area is some 170 m thick, and the succession with the coals seams is some 130 m thick. A geologically simple question is how such a lacustrine succession can develop, particularly if the occurrences of coal seams indicate that the entire succession was deposited at roughly the lake level. The answer is just as simple as the question: accommodation space must have been created that allowed sediment to accumulate, eventually reaching a total thickness of at least the succession with the coal seams. This, in turn, raises the simple question of how such an accommodation space can be created within in and along the marginal part of a huge lake. Also, this question has a simple answer: the accumulation of the sediments must have kept roughly pace with the local subsidence of the lake bottom.

This seems, however, not self-evident, as Chinese literature mentions that the Ordos Basin was (apart from some adjacent areas) tectonically quiet and stable during the Mesozoic (Liu and Li, 2007). This commonly accepted view has, unfortunately, led to the unjustified idea that no tectonic activity (such as subsidence) took place. This is the more remarkable since at least four tectonic phases influenced the evolution of the Ordos Basin, or at least the adjacent areas, during the Mesozoic, viz., the Late Triassic Indosinian Orogeny, and the Yanshan I (Early Jurassic), Yanshan II (Late Jurassic) and Yanshan III (Early Cretaceous) phases of the Yanshan cycle.

In spite of the above considerations, it seems scientifically only appropriate to check whether other processes may have resulted in a succession of over 100 m thick with several coal seams. It has been advocated (Liu and Li, 2007) that climate change must have been the main factor that controlled the development (see also Section 5.2.2). The accumulation of clastic sediment with peat on top is then ascribed to a rising lake level under the influence of climate (high precipitation rate, much inflow of river water). This poses several problems, however: where did the precipitation and inflowing water come from? Considering the paleogeographic situation (almost), all this water must have been derived from evaporated water from the Ordos Lake, which implies that the influence on the lake level was negligible and certainly not enough to create an accommodation space for many meters of sediment. A second problem is that the inflow of huge amounts of water from another water source might, indeed, raise the lake level (though only in a very limited way) so that clastic sediments with peat on top could develop, but that a next lake-level drop (as reflected by the sequence stratigraphical characteristics) would have resulted in erosion of the subaerially exposed peat, so that no coal layer would be formed.

Particularly considering the fact that several coal seams are present in a succession of over a hundred meters thick, it must be deduced that the characteristics of the coal measures, in combination with their geological context, falsify the commonly adhered to hypothesis that climate was the main factor controlling the development of the coal measures. On the other hand, it can be deduced that the basin subsidence, which had taken place already for a long time considering the thicknesses of the Triassic formations in the basin, created the accommodation space in which the coal measures could develop. The subsidence, although quiet and continuous if considered over longer time-spans, may have taken place with small irregularities (steps), which might explain the alternating dry and wet conditions under which the plants grew: slow subsidence could easily result in gradually drier conditions because the sedimentation was faster than the subsidence; when quick subsidence occurred, for instance due to small-scale faulting in the subsoil, lowering of the sedimentary surface resulted in a relative rise of the lake level, and consequently in wetter conditions.

Obviously, basin subsidence and climate change can have acted simultaneously, without leaving traces that prove which process prevailed. It should be realized, however, that only subsidence of the lake bottom could create the conditions in which the Yan’an Formation could develop, and under which successive coal seams could be formed.