国际期刊论文｜DionisosFlow应用案例

刘萝卜锅

品质源于技术 服务源于态度

这是阿什卡微信公众号的第778篇原创文章

首发于2020年6月7日

▽

作者：胡勇 等

Application of stratigraphic-sedimentological forward modeling of T sedimentary processes to predict high-quality reservoirs within tight sandstone

Hu Yong-a, Ma Yongning-b, Guo Bincheng-c, Gao Zhaopu-d, He Wenxiang-a,∗

-a Yangtze University, Wuhan, Hubei, 430100, China
-b Third Oil Production Plant of Changqing Oil Field, China Petroleum, 750006, China
-c PetroChina Exploration & Development Research Institute, Beijing, 100083, China
-d Exploration and Development Research Institute, SINOPEC North China Company, Zhengzhou, Henan, 450006, China

ABSTRACT

The prediction of high-quality reservoirs within tight sandstone is presently a popular and challenging research topic. Many geophysical characteristics of tight sandstone reservoirs are similar to those of the surrounding rock, making it difficult to predict the distribution of these reservoirs. In this paper, a stratigraphic-sedimentological forward modeling is proposed to predict the distribution of high-quality tight sandstone reservoirs in the Xihu Depression, East China Sea. Stratigraphic-sedimentological forward modeling can evaluate the influences of different geological parameters on sedimentary processes and the results. To reproduce the spatial distribution of sedimentary strata in a progradational delta and the overall stratigraphic stacking pattern, different parameters were tested by numerical simulation. According to the influence of the parameters on the sedimentary process, the parameters can be divided into two types: high-sensitivity parameters (transport capacity of the river and sand ratio) and low-sensitivity parameters (maximum initial water depth, lake level change, and tectonic sub- sidence). The analysis results are suitable for continental sedimentary basins with relatively high deposition rates. In the process of optimizing the simulation parameters, 236 simulations were carried out by adjusting the high-sensitivity parameters and then the low-sensitivity parameters. An optimal model that can reflect the de- velopment characteristics of high-quality tight sandstone reservoirs was obtained. According to the character- istics of high-quality reservoirs, the product of the sand ratio and reservoir thickness from the sedimentary simulation results is used as the parameter for evaluating high-quality reservoirs. The ratio can effectively overcome the difficulties of insufficient amounts of data and seismic prediction and effectively predict the lo- cations of high-quality reservoirs in low-exploration areas of tight sandstone reservoirs. This method also pro- vides a new idea for the prediction of high-quality reservoirs in this kind of area in the future.

1. Preface

With the increasing global demand for oil and gas resources and the increasing difficulty in exploration and development of conventional oil and gas resources, the proportion of unconventional oil and gas re- sources, such as coalbed methane, tight sandstone gas and shale gas, in the oil and gas supply has been increasing. In particular, tight sandstone gas has a dominant advantage because of its “three highs” (namely, technologically recoverable reserves, proven reserves and production) (Dai et al., 2012). In recent years, the production of tight sandstone gas in China has been steadily increasing at an annual increment of 5 billion m3 (Zou et al., 2015). Therefore, tight sandstone gas has more potential for growth than shale gas over the next 10 years and will be an im- portant resource with which to replace conventional oil and gas re- sources in the future, depending on improvements in the understanding of relevant geological patterns and/or in exploration and development technology. Although tight sandstone gas reservoirs are abundant, the rocks often become strongly compacted during their deposition; therefore, these reservoirs are generally dense and hard, and the geo- physical features of these reservoirs are not significantly different from those of the surrounding rock, making them difficult to identify using seismic data (Li and Xu, 2008; Dong et al., 2008; Ye et al., 2009; Liu, 2014; Huang et al., 2016; Zhu et al., 2015). Consequently, accurate prediction of the occurrence of high-quality tight sandstone reservoirs has become the main bottleneck in the exploration and development of tight sandstone oil and gas.

Fig. 1. Study area.

According to domestic scholars, the sedimentary environments of tight sandstone reservoirs in China are mostly related to deltas (He et al., 2003; Zou et al., 2010, 2012; Jia et al., 2012). These deltaic sedimentary bodies formed in a transitional environment, so the theory of sequence stratigraphy can be used to study the deposition and evo- lution of these deltas. Sequence stratigraphy mainly considers the ac- commodation space, sea/lake level changes and the source supply to study the filling process of sedimentary basins (Emery and Myers, 1996). Stratigraphic-sedimentological forward modeling is a computer- based stratigraphic simulation method developed on the basis of se- quence stratigraphy theory that can quantitatively describe the basin formation and filling process in multidimensional space and quantita- tively evaluate the influence of various geological factors on the sedi- mentary sequence and distribution of sedimentary facies. Stratigraphic- sedimentological forward modeling has unique advantages in the study of stratigraphy and factors that control deposition, and by comparing multiple simulation results, the uncertainty in the distribution of sedimentary facies and in the deposition of sedimentary sequences in a basin can be reduced (Harbaugh and Bonham-carter, 1970; Kenyon and Turcotte, 1985; Bosence and Waltham, 1990; Martinez and Harbaugh, 1993; Granjeon, 1996, 1999; Lin et al., 1995, 1998, 2010; Griffths et al., 2001; Syvitski and Hutton, 2001; Plint et al., 2001; Warrlich et al., 2002; Hoy and Ridgway, 2003; Pelletier, 2004; Overeem et al., 2005; Burgess et al., 2006; Vesely et al., 2007; Hutton and Syvitski, 2008; Torra et al., 2009; Xiu et al., 2012; Csato et al., 2013; Seard et al., 2013; de Paula Faria et al., 2017). Sedimentary modeling techniques can therefore be used to reproduce the sedimentary characteristics of del- taic sedimentary systems and to predict areas with high-quality re- servoirs.

This paper intends to use three-dimensional sedimentary simulation technology to study the sedimentary characteristics and predict the locations of high-quality reservoirs within a tight sandstone gas field in eastern China. By analyzing the characteristics of sequence develop- ment in the study area, carefully analyzing the input parameters af- fecting the results of the sedimentary simulations and the degree of influence of different parameters on the model, and carrying out

optimization research for the parameters, a conceptual model for pre- dicting high-quality tight sandstone reservoirs is proposed and tested.

2. Geological setting

The study area is located in the Xihu Depression of the East China Sea Shelf Basin. The structural features of the East China Sea Shelf Basin are oriented along the NE-SW direction and are characterized by a north-south block and east-west zonation. The Xihu Depression is a faulted basin with faulting in the east and overlapping strata in the west. As with the entire East China Sea Shelf Basin, the Xihu Depression has features typical of a north-south block and east-west zoning. The depression extends along the NNE direction, and the interior of the depression can be divided into five secondary structural units from west to east: the western slope belt, the western subsidence belt, the central inversion structural belt, the eastern subsidence belt and the eastern fault belt. The study area is located in the western slope zone, and the porosity of the reservoir in the target bed of the Oligocene Huagang Formation is 2.44–13.90%, with an average of 9.44%, whereas the permeability distribution ranges from 0.01 to 1.71 × 10−3 μm2, with a mean of 0.14 × 10−3 μm2 (Xu et al., 2012). According to the “People's Republic of China oil and gas industry standards - natural gas reservoir evaluation method” (SY/T5601-93,1993), this reservoir is composed of tight sandstone. In recent years, drilling in the block has not achieved the expected results primarily due to an unclear understanding of the distribution of high-quality reservoirs in this area and the difficulty in identifying new exploration opportunities. The X gas field in the study area is a fine-grained deltaic sedimentary system of western provenance with an area of approximately 200 km2, and 2 wells have been drilled in the area (Fig. 1). In the target area, the Huagang Formation was de- posited in a freshwater lake sedimentary environment (Zhou, 1994; Hu et al., 2010; Wang et al., 2002). After years of exploration, the gas fields near the X gas field have not been thoroughly explored. Both wells in the area exhibited good daily gas production after the initial production test. Therefore, the reservoir characteristics need to be evaluated to guide exploration and drilling in the next step of development.

2.1. High-quality reservoirs of tight sandstone

To maintain domestic oil and gas production, China has funded several projects to carry out unconventional reservoir research since 2010 (Zhao and Du, 2012). By analyzing 36 oil and gas fields (Du, 2016) in 11 basins in China, Du Jinhu discovered that the reservoirs of wells with high oil and gas production from tight sandstone were mostly deposited as delta progradational sand bodies because the delta depositional area was expanding and the distribution of sand bodies was gradually expanding during the progradational period. He also noted that sand with relatively good physical properties was distributed within a large area in a parallel plane to form the high-quality re- servoirs in the tight sandstone, which has been confirmed by many scholars (Wang et al., 2013; Zhang et al., 2015; Chen et al., 2016).

The reservoir of the Huagang Formation is tight sandstone, and due to poor reservoir properties and especially the high cost of offshore drilling, only reservoirs with relatively good physical properties and large area distributions can be classified as high-quality reservoirs. Under the guidance of this theory, exploration and research work concerning the tight sandstone in the Huagang Formation has begun.

刘萝卜锅

2.2. Lithology and sequence development characteristics

Fig. 2 shows the stratigraphic and sequence characteristics of Well A. The lower part of the Huagang Formation is mainly composed of silt and fine sandstone, and the upper part of the Huagang Formation is dominated by fine sandstone. Moreover, the values of the gamma ray (GR) logging curve of the lower part of the Huagang Formation are clearly higher than that of the upper part, which indicates that the water depth during the sedimentation period of the upper part of the Huagang Formation was shallower than that of the lower part. All these characteristics indicate that the Huagang Formation was deposited in the stage of water withdrawal in a progradational sedimentary system.

In recent years, drilling in the Huagang Formation has been mainly concentrated in the H1 layer. The H1 layer is also the simulation target in this research. The H1 layer in Fig. 1 shows the characteristics of progradation, both in lithology and in the GR logging curve. The thin section photograph in Fig. 3 shows that the lithologic grain size is coarse and the porosity improves from bottom to top. The value of the GR curve in Fig. 2 gradually decreases upwards. The seismic profile in Fig. 4 also shows the progradational sedimentary architecture of the H1 layer (faults did not develop in these areas), and during this period, the delta deposits were the largest and the most likely to form high-quality reservoirs.

Fig. 2. Characteristics of the strata in the Huagang Formation (The lower part of the Huagang Formation is mainly composed of silt and fine sandstone, and the upper part of the Huagang Formation is dominated by fine sandstone. The gamma ray (GR) logging curve values of the lower part of the Huagang Formation are higher than those of the upper part.).

2.3. Reservoir prediction

Because of the high drilling cost, offshore oilfield data are usually scarce in the exploration stage, and seismic data are often used in re- servoir prediction. Seismic reservoir prediction can be performed using many methods, and wave impedance inversion has become a key technology in predicting the lateral continuity of reservoirs based on seismic data. At present, the development of the seismic inversion technique and the update speeds of various inversion software systems are very fast (Liu et al., 2013; Gu et al., 2016; Wang et al., 2017a,b; Fang et al., 2017; Zheng et al., 2017; Li et al., 2017; Wang et al., 2017a,b). However, wave impedance inversion requires the wave im- pedance of sandstone and mudstone to be different, as in Fig. 5a; therefore, a threshold can be determined between the sandstone and mudstone (approximately 7.5 × 106 m/s.kg/m3) so that the sandstone can be depicted. However, the actual situation in this area is shown in Fig. 5b: the wave impedance overlapping area of the sandstone and mudstone is large, and determining a threshold value of wave im- pedance to distinguish between sandstone and mudstone is difficult, so reservoir prediction by wave impedance inversion in this case has great uncertainty.

Fig. 3. Lithology and grain size characteristics of the H1 layer in the Huagang Formation (Well A - slice). a 2192 m, fine sandstone, average particle size 0.1 mm, strong compaction, close contact between particles, low porosity. b 2160 m, fine sandstone, average particle size 0.2 mm, pore development.

Tight sandstone reservoirs generally have associated difficulties with reservoir prediction. The large area, minimal wells and unclear seismic reservoir response characteristics in this study bring further challenges to reservoir research. Fortunately, the succession examined in this study is a typical delta depositional environment, which makes it possible to study the reservoir by stratigraphic-sedimentological for- ward modeling. In addition, the changes in the delta sedimentary ar- chitecture can be clearly observed from the seismic data, which also provides a way to verify the sedimentary simulation results and thus helps us adjust the parameters of the stratigraphic-sedimentological forward model and reproduce the three-dimensional distribution of reservoirs to carry out the analysis and prediction of favorable re- servoirs.

Fig. 4. Prograding depositional architecture of the H1 layer of Huagang Formation (The seismic profile shows that the seismic reflection axis extends from left to right with one reflection axis covering another reflection axis, showing the prograding depositional architecture. The reflection axis at the end of deposition is more continuous, and the distribu- tion area is larger.).

Fig. 5. Sandstone and mudstone impedance distribution histogram of the H1 layer (In the left figure, the wave impedance of sandstone is high and that of mudstone is low, so a threshold value of wave impedance can be determined to distinguish sandstone from mudstone. However, the wave impedance of sand- stone and mudstone overlaps in the right graph, and wave impedance cannot be used to distinguish lithology.).

Three-dimensional seismic data from the area are available. By analyzing the impedance characteristics of the reservoir from the wells, we found that the overlap of sand-shale impedance is considerable, making it difficult to predict the location of the reservoir using seismic inversion (Fig. 2).

刘萝卜锅

3. Study area and methods

3.1. Study area

The research area is a square area of 14 km * 14 km, the simulated succession is the H1 layer of the Huagang Formation, and the H1 thickness in the drilling area is approximately 60 m (Fig. 6). The area contains logging, core, slice and paleontology analysis data from two wells, for which we show some data in Figs. 2–5. In addition, the study area contains three-dimensional seismic data, which we can use to analyze the sedimentary architecture (Fig. 4) and verify the reliability of the simulation results.

Fig. 6. Stratigraphic thickness distribution map.

In this study, the main parameters used in the forward simulation are determined by analyzing the existing data, and these parameters are input into the simulation software. The sensitivity analysis of these parameters is carried out by the following procedure.

3.2. Forward modeling and methods

Forward modeling of sedimentary systems is the process of re- constructing strata as they were deposited and evolved over time (Hawie et al., 2015). There are numerous stratigraphic-sedimentolo- gical forward modeling programs, such as Fuzzim (U Nordlund, 1999), Simsafadim (K Bitzer and Salas, 2002), and Sedsim (Xiu et al., 2012), all of which have unique advantages and disadvantages (Li, 2009). Com- pared with other software, Dionisos has three main technical ad- vantages. (1) The software can quantitatively describe the accom- modation space, which is the potential volume that can be filled with sediments. Variations in the accommodation space constitute one of the main factors controlling the geometric shape and lithological distribu- tion of the stratigraphic units. The Dionisos software fully considers various geological factors, including basin deformation (i.e., sub- sidence), sea/lake level rise and fall, compaction and flexure, in addi- tion to the presence of growth faults and the structural deformation of salt, all of which affect the changes in the accommodation space. Therefore, the original appearance and stratigraphic distribution of the basin can be effectively restored. (2) The software can provide a de- tailed description of the supply of sedimentary sources. The source number, supply and type all control the overall lithological composition of the detrital materials involved in each time period. Dionisos can define the supply rate, type, and location of the source, in addition to the variation in the source inflow width with time and the changes in the rainfall and water load. Through a detailed description of the source supply, we can analyze and simulate the distribution of sediments under different source supply conditions. (3) The software can quanti- tatively simulate the sediment transport process. The diffusion equation is used to describe the transport and deposition processes. The transport capacity of clastic particles is directly proportional to the water flow and slope, and the transport capacity varies for different lithological particles; thus, the diffusion coefficient also differs. The diffusion coefficient is an empirical constant describing the deposition of the entire depositional system. Through a reasonable estimation of the diffusion coefficient and a subsequent simulation adjustment, the transport process model, which is more consistent with the geological law of the target area, can be obtained, and the distributions of lithologies and lithofacies in a three-dimensional space can be de- scribed more thoroughly with regard to the geological period. Many

scholars have carried out studies of stratigraphic-sedimentological forward modeling with Dionisos software (Berne et al., 2004; Rabineau et al., 2005; Alzaga-Ruiza et al., 2009; Sømme et al., 2009; Csato et al., 2013; Seard et al., 2013). Based on the above characteristics, this paper uses the Dionisos software to conduct sedimentary simulation research.

3.3. Input parameters

There are five main parameters for the forward modeling of sedi- mentary processes. The first three parameters are tectonic subsidence, initial bathymetry and lake level change, which are the main control- ling factors of accommodation space changes. The fourth main para- meter is the source of the sediment supply, including the area, direc- tion, location and lithologic composition of the source material, which can affect the macroscale deposition pattern. The fifth main parameter is the transport mode of the water depth. Different particle transport mechanisms produce different sedimentary architectures, affecting the spatial distribution of the sediments.

According to the above characteristics, the following 5 input para- meters are mainly considered in this research.

3.3.1. Initial bathymetry

The basin must have some accommodation space before sediment filling can occur. The size of the accommodation space is determined by the water depth during basin sedimentation. Therefore, sedimentary simulation is actually the process of reproducing sediment filling under different water conditions. The depth of the water before sedimentation is the initial bathymetry of the simulation.

Due to the considerable subsidence of continental faulted basins in eastern China, the depocenter and subsidence center are relatively consistent. The center of a basin has thick deposits, whereas the areas with shallow paleo-water depths at the edge of a basin have thinner sedimentary deposits. Thus, the thickness of sedimentary deposits can reflect the paleo-water depth of the basin, as demonstrated by Dong and He (2010). The thickness of the right side of the work area shown in Fig. 6 is large and indicates deep-water deposition. In addition, the depth of the oxidation environments in lake sediments in China is generally less than 15 m and can reach 20 m within estuaries. The depth range of an environment characterized by weak oxidation and weak reduction is generally 15–25 m, and the depth of a reducing environ- ment is generally greater than 25 m (Xue and Liu, 1999). Thus, the initial average depth of the study area is 20 m, and the maximum depth of the basin is 30 m (Fig. 7).

Fig. 7. Initial bathymetry.

刘萝卜锅

3.3.2. Lake level change

Lake level changes affect not only the macroscopic accommodation space but also the small-scale sedimentary architecture. Generally, a retrogradational architecture is formed when the water level rises, and a progradational architecture is formed when the water level falls (Yu et al., 2013). The absolute lake level is a difficult parameter to de- termine. At present, the methods used to determine the changes in lake level primarily provide estimates of the relative water level (Li and Zhang, 1999; Zhao et al., 2000). The two exploration wells in the study area have provided a wealth of analytical laboratory data, of which the paleontological data can effectively reflect changes in the sedimentary environment and water depth and can be used as an important basis for determining the lake level change curve. Sporopollen data can reflect the paleoclimate and paleo-water environment to a certain degree (Li et al., 2005). Generally, sporopollen data are used to characterize the sporopollen in one stratum. A higher pollen degree indicates an abun- dance of sporopollen, reflecting a rise in the lake level, expansion of the basin and an increase in the abundance of species. A lower pollen abundance indicates a lack of species, a decrease in the lake level, and shrinkage of the basin (Lu et al., 1998; Ren et al., 2001). According to the statistical characteristics of sporopollen differentiation, the paleo- water characteristics of the target layer were consistent between the two wells in the study area. As the depth increased, the sporopollen differentiation and paleo-water level gradually decreased (Fig. 8).

Fig. 8. Relationship between sporopollen differentiation and depth in the study area (Both Well A and Well B show a decrease in depth as the sporopollen differ- entiation and paleo-water level gradually decrease).

Poyang Lake is the largest freshwater lake in China, with an area of approximately 3690 km . Ouyang and Liu (2014) examined the lake level variation characteristics over the past 50 years and found that the average variation amplitude of Poyang Lake is 5 m. In this study area, the lake level variation amplitude was set to 5m, and the relative variation curve of water depth in the study area was determined (Fig. 9).

Fig. 9. Curve of lake level changes.

3.3.3. Subsidence and compaction

Tectonic subsidence is another major controlling factor that affects the accommodation space. Changes in lake level control the develop- ment patterns of the stratigraphic sequences and subsequences, the sedimentary system and the lithofacies distribution. During the filling and evolution of sedimentary basins, the thicknesses of sedimentary deposits are an integrated effect of tectonic subsidence, isostasy and water level change. The total subsidence of a basin is the sum of the tectonic subsidence and load subsidence, wherein the latter is related to the crustal isostasy caused by the sediment load. In this study, the tectonic subsidence is primarily calculated using the Airy equilibrium formula (Yang et al., 2017). The subsidence caused by lake level rise, tectonic action and the load balance is calculated. Faulted lacustrine strata with a thickness of approximately 1000 m experienced only se- dimentary compaction; they did not suffer from structural inversion in the later period and typically exhibit a large degree of tectonic compaction. The depositional center and subsidence center coincide, and the depositional thickness basically represents the level of sub- sidence. Fig. 10 shows the subsidence map of the target layer. The tectonic subsidence center of the study area coincides with the depo- center of the basin. When the sedimentary strata are thicker, the amount of tectonic subsidence is also larger. In general, the amount of structural subsidence in the eastern part of the study area is greater than that in the western part.

Fig. 10. Subsidence of the target layer in the study area.

compaction. The depositional center and subsidence center coincide, and the depositional thickness basically represents the level of sub- sidence. Fig. 10 shows the subsidence map of the target layer. The tectonic subsidence center of the study area coincides with the depo- center of the basin. When the sedimentary strata are thicker, the amount of tectonic subsidence is also larger. In general, the amount of structural subsidence in the eastern part of the study area is greater than that in the western part.

Compaction is another factor that affects the accommodation space. In a compaction model, a compaction curve is usually used to represent the compaction characteristics of different lithologies, as it shows the relationship between the porosity and depth. The following formula describes this relationship, including the three following parameters: the initial porosity (porosity at the surface), minimum porosity, and decay factor.

Φ = Φf+Φa × Exp(-Z/Cd)

Where Φ represents the porosity, Φf represents the minimum porosity, Φa represents the initial porosity, Z represents the depth, and Cd re- presents the attenuation factor.

3.3.4. Supply of source sediments

In this work, the type and lateral distribution of the sedimentary system is influenced by the source supply model, which greatly impacts the model. The area of the study location (14 km × 14 km) and the average thickness of the target H1 layer can be used to calculate the total amount of sediment material supplied by the source. Because the thickness of H1 layer in the study area is not stable but gradually thickens along the provenance direction, the total sediment volume calculated here is only an approximate value. The approximate total volume of the H1 layer is 11.76 km3, and the research results by Zhou and Qian (1996) indicate that the period of deposition for the H1 layer in the Huagang Formation was 36–35.5Ma and that sedimentation started before 36 Ma and ended before 35.5 Ma. Therefore, the rate of sediment supply from the provenance area was 23.52 km3/Ma. In ad- dition, the core observations show that the reservoirs in the study area are primarily composed of fine sandstone containing 64.5% sandstone and 35.5% mudstone.

3.3.5. Sediment transport

Sediment transport in the basin is described by the diffusion equa- tion. The diffusion equation primarily considers the average particle size of the sediment, river transport capacity and slope gradient. Among these factors, the terrain gradient is the most influential. Under different slope conditions, the shape of the sedimentary body will also vary. The steeper the slope is, the thicker the sedimentary body and the smaller its area (Yu et al., 2013). The structural trends and tectonics influence the particle size and spatial distribution of the sediments: under the same hydrodynamic conditions in the study area, a shorter particle transport distance will result in the deposition of finer sand- stone. The sediment particle size characteristics are shown in Table 1. To determine the total sediment volume, the transport capacity of the river is related to the particle density. The detrital material con- centration in mountain rivers is lowest among continental rivers, being 0.1 g/L (Yin et al., 2017), whereas the concentration of debris in un- derwater distributary channels is generally less than 0.1 g/L due to the diffusion effect of lake water. The detrital concentration in this study area has an empirical value of 0.065 g/L.

刘萝卜锅

4. Discussion and results

Stratigraphic-sedimentological forward modeling has gone through a period of development; in this process, with improved technology, the models are becoming more complex as more factors are being con- sidered (Cross and Harbaugh, 1989; Granjeon, 1996; Griffths et al., 2001; Cheng and Si, 2013; Rahman et al., 2014). Thus, many factors affect sedimentation, but the final sedimentary model is the result of the interaction of many factors. The method for determining a set of rea- sonable parameters is the key point in the model process and requires repeated analyses. Many models do not achieve the best results in the first simulation (Robustelli et al., 2014; Yin et al., 2015; Cloetingh et al., 2015; Faria et al; 2017; Cooper et al., 2018), as these models need re- peated analyses of the geological conditions in the work area and parameter adjustments. For example, Faria (2017) set up a parameter table and made four models, and after careful analysis, determined the most appropriate simulation parameters. Therefore, analyzing the parameters is necessary before determining the simulation results.

A similar problem also exists in the study area, and the results of the first simulation are not satisfactory. Five parameters are used in the first simulation (initial bathymetry, lake level change, subsidence, source supply, and river transport capacity), and the initial values are used. We did not carefully examine the simulation results. We only compared the simulated H1 thickness with the well data (Table 2) and found that the error was greater than 15%, which is enough to show that the current parameters are unreasonable.

According to the above analysis, a detailed analysis of the impact of the above five parameters (initial bathymetry, lake level change, sub- sidence, source supply, and river transport capacity) on the simulation results is necessary to determine the set of parameters that can best reflect the actual geological conditions in the simulation results.

4.1. Sensitivity analysis of the simulated parameters and discussion

Geological parameters and data are often highly uncertain. These parameters are generally simplified according to qualitative geological models and concepts. Different parameters in geological models often lead to different results. In fact, determining precise geological para- meters, such as settlement history and precise settlement amplitude, is difficult, especially for small-scale tectonic activities. Therefore, strati- graphic-sedimentological forward modeling can help us understand the effects of uncertain parameters on sedimentary processes (Burgess et al., 2006).

Parameter sensitivity analysis is the prerequisite of model optimi- zation. Sensitivity analysis is performed to find the parameters out of many uncertain parameters that have the greatest influence on the analysis objects and analyze their influence on the simulation results to overcome or reduce the uncertainties caused by these factors to a cer- tain extent (Caers, 2011; Fenwick et al., 2014).

Because we cannot determine the appropriate parameter values at first, we assign the parameters several values, simulate each parameter value, and determine the sensitivity of the parameters according to the influence of the different values on the model. Classifying the sensitivity level is difficult. Thus, in this study, the sensitivity is simply divided into two levels: low sensitivity and high sensitivity.

The following describes a simulation to determine the sensitivity of parameters with uncertainties.

4.1.1. Original water depth

We defined the maximum original water depth as 25 m, 30 m (re- ference value) and 35m and then compared the simulation results, which are shown in Fig. 11a, b and 11c, respectively.

Fig. 11. Simulation results after changing the initial water depth. (The max- imum initial water depth is 25 m in (a), 30 m in (b), and 35 m in (c). From a to c, the sand body moves slightly towards the source direction.).

The results show that there is little difference between the three results, but there are some differences in the location of sand bodies. From Fig. 11a–c, the water level gradually rises, and the sand bodies corresponding to the two comparison lines (line 1, line 2) show a trend of moving to the left. This result is due to rising water level, resulting in sediment tending to be deposited towards the source direction.

Because the maximum water depth of lacustrine basins in China is mostly concentrated at 30 m (Xue and Liu, 1999), the three maximum water depth values we tested represent the possible scenarios. Although the sand bodies have moved, the movement distance is very short. Thus, the distribution of sand bodies is not greatly affected by the water depth, so the original water depth is a low-sensitivity parameter.

4.1.2. Changes in lake level

We defined the variation in lake level as 2 m, 5 m and 8 m, and the simulation results are shown in Fig. 12a, b, and 12c, respectively.

Changes in water depth usually change the sedimentary architecture and lithology distribution. Because the water depth has a small range in this experiment, the three simulation results are not very different for the sedimentary architecture, but water depth changes do influence the sand ratio. When the lake level falls, the sediments move towards the center of the basin. Therefore, the sand bodies corresponding to the two comparison lines (line 1, line 2) in Fig. 12 tend to move towards the basin. At the same time, the sediments are more severely eroded in the shallower water, so the sand ratio increases gradually from Fig. 12a–c.

Fig. 12. Simulation results after changing the lake level (The variation in lake level is defined as 2m in (a), 5m in (b), and 8m in (c). From a to c, the sand bodies move slightly to the center of the basin, accompanied by an increased sand ratio.).

Generally, the change in lake level has little influence on the sedi- mentation of the sand body, so this parameter can be classified as a low- sensitivity parameter.

4.1.3. Tectonic subsidence

Together, tectonic subsidence and lake level changes control the changes in accommodation space, which affect the development of the sequence. Based on the existing data, the range of tectonic subsidence is 5 m to −5 m. The simulation results are shown in Fig. 13a and b.

Fig. 13. Simulation results after changing the tectonic subsidence (The range of tectonic subsidence is the initial value increased by 5 m in (a), and the initial value decreased by 5 m in (b). Similar to the effect of the initial water depth, the movement of the sand body is smaller due to gradual subsidence.).

Tectonic subsidence affects the accommodation space. When the basin subsided, the water depth increased (Fig. 13a), and when the basin was uplifted, the water depth decreased (Fig. 13b). Therefore, the influence of tectonic subsidence on the model is the same as that of the initial water depth, but tectonic subsidence did not occur at one time; it proceeded slowly during the deposition of the H1 layer. Compared with the results of the simulation in Fig. 11, the difference between the two simulation results for the sedimentary architecture is even smaller. However, tectonic subsidence will affect the depth of the sedimentary strata.

The simulation results show that the effect of this parameter on sedimentation is similar to that of the initial water depth, and the effect is not significant. Thus, tectonic subsidence is a low-sensitivity para- meter.

刘萝卜锅

4.1.4. Supply of source sediments

Regarding the source parameters, the total amount of source ma- terial is relatively known and was thus not adjusted. However, the sand/mud ratio affects the spatial distribution of the sedimentary re- servoirs, and because few wells were drilled in the study area, the statistics of the sand ratio in the well may not correspond to the overall geological conditions of the study area. Therefore, based on the original sand ratio (64.5%), the sand ratio is increased by 25% (80.63%) and decreased by 25% (48.38%). Then, the sand ratios of 80.63%, 64.5% and 48.38% are simulated three times. The results are shown in Fig. 14.

Fig. 14. Simulation results after changing the sand ratio (The sand ratio is the initial value reduced by 25% in (a), the initial value in (b) and the initial value increased by 25% in (c). From a to c, the sand ratio increases, and the thickness of the sand body also increases.).

When a lower sand ratio is set as the input parameter (sand ratio is reduced by 25%), the sand ratio of the simulation results is lower (Fig. 14a); when a higher sand ratio is set as the input parameter (sand ratio is increased by 25%), the overall sand ratio of the simulation re- sults is higher (Fig. 14c).

In addition to the obvious change in the sand ratio, with an in- creasing sand ratio from Fig. 14a–c, the overall thickness of the sand body in the high sand ratio model is larger (mainly near the source). This result is because with the increase in the sand ratio and the un- changed water transport capacity, the proportion of sediments trans- ported away from the source decreases, resulting in the majority of sediments being deposited near the source.

Based on the above analysis, the sand ratio of the provenance area is a high-sensitivity parameter.

4.1.5. Sediment transport

Regarding the sediment transport parameters, the sediment particle size is based on an experimental analysis of the collected data and was thus not adjusted. The density of continental river sediments varies greatly; hence, in the process of adjustment, the main purpose is to adjust the river transport capacity. The initial river transport capacity is 0.065 g/L. The river transport capacities of this test are therefore 0.1*0.065 g/L, 0.065 g/L, and 10*0.065 g/L, and the results are shown in Fig. 15.

Fig. 15. Simulation results after changing the river transport capacity (The river transport capacity is the initial value reduced 10 times in (a), the initial value in (b) and the initial value increased 10 times in (c). From a to c, the sedimentary topography changes from a steep slope to gentle slope.).

The difference between these three models is very obvious and is mainly reflected in the difference in sedimentary architecture. The se- dimentary bodies in Fig. 15a are steep, while those in Fig. 15c have relatively gentle slopes, and those in Fig. 15b have moderate slopes between those in Fig. 15a and c. This result is mainly due to the in- crease in the river transport capacity; the river will erode the sediments near the source and continue to move towards the center of the basin, resulting in the reduction in the thickness of sediments near the source direction while forming relatively flat terrain.

Through the above analysis and comparison, sediment transport is considered a high-sensitivity parameter.

4.1.6. Discussion

In the parameter sensitivity analysis, the sand ratio and river transport capacity were determined to be high-sensitivity parameters by testing, while the maximum initial water depth, lake level change amplitude and tectonic subsidence were determined to be low-sensi- tivity parameters. The high-sensitivity parameters are well understood, and different proportions of lithology and different transport capacities will certainly lead to differences in the spatial distribution of lithology.

However, the three low-sensitivity parameters we defined control the changes in the basin accommodation space; for example, with lake level changes, the accommodation space also changes, resulting in differences in the spatial distribution of lithologies. Simulations by several scholars confirm that lake level changes and tectonic subsidence are high-sensitivity parameters (Robustelli et al., 2014; Cloetingh et al., 2015; De et al., 2017; Cooper et al., 2018).

The deposition rate of depression basins with stable deposits is generally 10–20 m/Ma, while the continental faulted basins in China generally have an adequate supply of source sediments. For example, the Cenozoic maximum sedimentary thickness of the Xihu Depression was greater than 10 000 m and that of the Oligocene Huagang Formation was approximately 1000 m. However, the deposition rate of the layer studied in this paper was approximately 60 m/Ma, and in fans or deltas with a sufficient terrigenous clastic supply, sedimentary basins can have deposition rates of up to 400 m/Ma (Yu, 2008). Therefore, under the condition of a relatively sufficient supply of source sediments, the three factors—maximum initial water depth, amplitude of lake level changes and tectonic subsidence—are relatively low-sensitivity factors. Therefore, the variation in sediments and the mode of transport are the main factors that affect the spatial distribution of sediments. Yin et al. (2015) also got the same conclusion when studied a block in the Bohai basin.

In addition, the H1 layer of the Huagang Formation was deposited under a temperate climate (Zhou, 1994). The climate was relatively dry, the sediment ratio of the river was relatively high, and the river carrying capacity was relatively high. These factors also explain why the river carrying capacity of the study area should be 0.13 g/L. Be- cause of the stable climate during the H1 deposition period, the impact of climate change was not considered.

4.2. Parameter optimization and simulation results

The classification of parameter sensitivity can not only help us understand the main controlling factors affecting sequence and sand body distribution but also help us effectively carry out parameter ad- justment research. There are many parameters in stratigraphic-sedi- mentological forward modeling. If we cannot understand the influence of different parameters on the simulation results, we cannot quickly obtain the optimized model parameters. When adjusting the para- meters, the parameters with the highest sensitivities (which have greater impacts on the model) need to be adjusted first, and then, the parameters with relatively low sensitivities can be adjusted to obtain the optimal set of parameters.

4.2.1. Model 1

First, we find that the thickness and structure of the succession have changed greatly under the same source supply compared with the re- sults in Fig. 15. To determine the appropriate transport parameters, we refined the uncertainty interval of the high-sensitivity factors; that is, when the default river transport capacity (rtc) was 0.065 g/L, the un- certainty interval was 0.1 × 0.065 g/L-10 × 0.065 g/L. According to multiples of 0.065 g/L, the uncertainty interval was divided into 19 values. These 19 values are rtc*0.1, rtc*0.9, rtc*0.8, rtc*0.7, rtc*0.6, rtc*0.5, rtc*0.4, rtc*0.3, rtc*0.2, rtc, rtc*2, rtc*3, rtc*4, rtc*5, rtc*6, rtc*7, rtc*8, rtc*9, and rtc*10, where rtc = 0.065 g/L.

Comparing the results in Fig. 14, the structure of the successions has minor differences, and the change in thickness is small. However, be- cause changing the sand ratio leads to a change in the sandstone plane distribution, the influence of the sand ratio parameters on the simula- tion results is relatively small compared with the river transport capa- city. Therefore, the sand ratio is divided into 11 values (80.63, 77.4, 74.18, 71, 67.73, 64.5, 61.27, 58, 54.8, 51.58, and 48.35) based on the initial value (64.5%) with a variation range in the sand ratio from +25% to −25% using a 5% change rate.

With the three low-sensitivity parameters set to the initial values, the eleven values of the sand ratio were intersected with the 19 values of the river transport capacity, and 209 models were obtained.

The simulation steps are as follows:

(1)  When the sand ratio is 80.63%, the river transport capacity has 19 values (rtc/10, rtc/9, rtc/8, rtc/7, rtc/6, rtc/5, rtc/4, rtc/3, rtc/2, rtc*2, rtc*3, rtc*4, rtc*5, rtc*6, rtc*7, rtc*8, rtc*9, and rtc*10, where rtc = 0.065 g/L), and the other parameters maintain the initial values. Then, we obtained 19 sets of parameters that changed with the sand ratio and river transport capacity and ran the simu- lation 19 times.

(2)  When the sand ratio is 77.4%, the river carrying capacity has 19 values (as above), 19 sets of parameters are obtained, and the si- mulation was run 19 times again.

(3)  When the sand ratio is 74.18%, the river carrying capacity has 19 values (as above), 19 sets of parameters are obtained, and the si- mulation was run 19 times again.

(11) When the sand ratio is 48.35%, the river carrying capacity has

19 values (as above), and 19 sets of parameters are obtained. Again, the simulation was run 19 times.

By dividing the sand content into 11 values and the river carrying capacity into 19 values, 209 sets of possible parameters were obtained, and the simulation was run 209 times.

Fig. 14 shows that when the sand ratio is fixed, the sedimentary topography becomes smoother and the thickness of the strata changes with increasing river carrying capacity (the center of the basin becomes thicker and the source direction becomes thinner).

In Fig. 15, the carrying capacity of the river is fixed, and the sand ratio is changing. With the increase in the sand ratio, in addition to the change in the overall sand ratio, a higher sand ratio leads to more sand bodies deposited towards the source direction, which affects the thickness of the strata (thicker towards the source direction).

After clarifying the influence of these two parameters on the model, we compared the sedimentary architecture of the simulation results with the seismic profile and compared the H1 thickness and sand ratio of the simulation results with the well data and determined a set of the most suitable parameters. The sand content is increased by 10% or 67.73% on the basis of the initial value. The carrying capacity of the river is two times that of the initial value, that is, 0.13 g/L. The set of parameters corresponding to this model is labeled Model 1 (Figs. 16 and 17). Fig. 16 shows that the simulated sedimentary architecture is in good agreement with the seismic architecture. Fig. 17 shows that the simulated H1 thickness is in good agreement with the well thickness, and the overall trend of the sand ratio is in good agreement.

Fig. 16. Simulation results for Model 1 (The model obtained after 209 sets of parameter tests.).

Fig. 17. Comparison between simulation results and well data from Model 1.

刘萝卜锅

4.2.2. Model 2

The H1 thickness and sand ratio of the model in Fig. 17 agree well with the well data, but there are some differences in depth. As shown in Fig. 17, the simulated H1 layer in Well A is 3–60 m (thickness is 57 m), and the logging curve is 0–57m (thickness is 57m). The simulation result of the Well B is 6–70 m (thickness is 64 m), and the logging curve

shows 0–65 m (thickness is 65 m). The simulated sand ratio of the two wells is consistent with the logging curve, and the simulated H1 thickness is in good agreement with the well data, but there are some deviations in depth, and the simulated H1 depth is shallower than the actual H1 depth.

Therefore, the model needs to be further adjusted. Because these three parameters have less influence on the model than the highly sensitive factors, the number of parameter intervals is also less, as the scheme is based on the parameter sensitivity analysis. That is, the maximum initial water depth was 25 m, 30 m (initial value) and 35 m, and the variation amplitude of the lake level was 2 m, 5 m (initial value) and 8 m, respectively. The tectonic subsidence has 3 values, the initial value, the initial value increased by 5 m, and the initial value decreased by 5 m.

The maximum initial water depth, the amplitude of the lake level change and the range of tectonic subsidence were intersected in the simulations, and 27 sets of simulation parameters were obtained (the value determination method is the same as that in Model 1). Then, the 27 simulation calculations were carried out under the condition that the river transport capacity and sand ratio of Model 1 remained unchanged. Finally, Model 2 (Figs. 18 and 19), which is most suitable for the geo- logical conditions of the work area, was confirmed. The corresponding parameters are as follows: the maximum initial water depth is 25 m, the change in the lake level is 8 m, and the tectonic subsidence is the initial value decreased by 5 m.

Fig. 18. Simulation results for Model 2.(On the basis of Model 1, this model was obtained after 27 sets of parameter tests.)

Fig. 19. Comparison between simulation results from Model 2 and well data.

Through comparison, we find that the sand content of Model 2 has a better relationship with the corresponding well by adjusting the three low-sensitivity parameters (Fig. 19). In addition, by increasing the tectonic subsidence, the range of lake level change is increased, and the relatively low water depth is maintained. The simulated H1 depth is successfully increased, which is consistent with the well data.

Comparing the sand ratio between the simulation and the well data (Fig. 19) shows that the simulation results agree well with the log curve (sand ratio). Compared with Well A, Well B is in the late stage of progradation, and the sand ratio of Well B is higher than that of Well A overall. The simulation results also show these change characteristics.

Thus, the simulation results are not only in good agreement with the well but also reflect the macroscopic sedimentary characteristics.

5. Conceptual model and high-quality reservoir prediction

5.1. Conceptual model

In the model parameter optimization, 236 models were tested ac- cording to the parameters, and only Model 2 duplicated the H1 de- positional and lithologic characteristics observed in the well data. In addition, since there are three-dimensional seismic data for this area, the macrosedimentary architecture of the two strata was mainly com- pared through the seismic profile, and a time-depth conversion of the seismic data was carried out before the comparison. The seismic profile shown in Fig. 20a is the depth domain.

Fig. 20. Contrast diagram of the sedimentary architecture in a cross section of the H1 layer. a Structure based on seismic impedance. b Sedimentary architecture profile of the simulation (sand ratio).

According to the seismic reflection structure in Fig. 20a, due to the downward trend of the water level of the target layer, the seismic re- flection presents an overall progradational architecture. The thickness of the sedimentary bodies gradually increases from left to right, and the area of the sedimentary bodies also gradually increases, reflecting the increasing trend of the sediment supply. Fig. 20b is a simulation result, and the overall depositional architecture is in good agreement with the seismic reflection data. Due to the limited seismic resolution, the seismic data can only reflect the sedimentary architecture, and the si- mulation results are on a smaller scale with more detail. The research results show that the simulation results not only reflect the macroscopic pattern of the sedimentary body but also reflect more details than the seismic data, allowing better prediction of the seismic data.

The sand ratio can reflect the distribution of sedimentary facies. In practice, the sand ratio is often used as a quantitative index to de- termine sedimentary facies (Yu et al., 2013). By analyzing the char- acteristics of the sand ratio in different strata, we can elucidate the sedimentary evolution characteristics of the study area.

Model 2 also confirms the distribution characteristics of high- quality tight sandstone reservoirs (Du, 2016) and is suitable as a con- ceptual sedimentary model for the exploration of tight sandstone in China. High-quality tight sandstone reservoirs need continuous dis- tribution. This rule can be seen from the evolution map of Model 2 in prograding deltas—as the sediments move towards the basin, the de- positional scale becomes increasingly larger, and the sandstone be- comes more connected (Fig. 21). The cross section in Fig. 20b shows that the delta sand bodies that formed later have obviously higher sand ratios, which generally means higher porosity and can be targeted as exploration areas for high-quality reservoirs. This evidence implies that, according to our model, the conceptual model proposed for the Xihu Sag seems to be a coherent model for interpreting high-quality re- servoirs in tight sandstone.

Fig. 21. Sedimentary evolution maps in the study area (As the water depth falls and the sediments move towards the basin, the sediment scale increases, and the sandstone connectivity improves.).

Therefore, through the above analysis, we can understand the dis- tribution and evolution of the sand ratio through Model 2 and use Model 2 as a conceptual model for high-quality reservoir prediction. The simulation parameters of Model 2 are as follows: the maximum initial water depth is 25 m, the amplitude of lake level changes is 8 m, the tectonic subsidence is the value in Fig. 10, the river transport ca- pacity is 0.13 g/L, and the sand ratio is 70.9%.

5.2. High-quality reservoir prediction

High-quality reservoirs were previously defined as high-porosity and high-permeability zones surrounded by low-porosity and low-per- meability rock (Surdam, 1997; Zou et al., 2006; Zhu et al., 2009). With an increasing understanding of tight sandstone reservoirs, the concept of high-quality reservoirs has also been continuously developed and improved. At present, the main characteristics of high-quality sand- stone reservoirs are relatively good physical properties and a large extent (Du, 2016). Therefore, in addition to the physical properties, the thickness of the reservoir should be considered. Only when the thick- ness and physical properties satisfy certain conditions is it possible to produce enough oil and gas to meet certain economic requirements.

In practice, the exploration and development potential of a reservoir is usually evaluated in terms of its reserves. The porosity, thickness and saturation of the reservoir are the main parameters in this calculation. Therefore, by combining the characteristics of high-quality reservoirs, this study primarily uses the product of porosity and thickness to quantitatively evaluate and predict the distribution of high-quality re- servoirs. The thickness parameters of the reservoir can be obtained from the results of depositional simulations. Our analysis suggests that there is a good correlation between the sand ratio characteristics and the porosity in the study area (Fig. 22); that is, areas with a high sand ratio have relatively good physical properties. Reliable porosity parameters can be obtained by the sand ratio, and then the product parameters of porosity and thickness in the study area can be obtained (Fig. 23).

Fig. 22. Sand ratio and porosity diagram of the H1 layer.

Fig. 23. Product of porosity and thickness for the H1 layer.

The prediction of high-quality reservoirs (Fig. 23) accurately reflects the scale and distribution of the actual sedimentary sand bodies. Near the source delta plain, the sand bodies are thin, and the area of the distributed high-quality reservoirs is small. On the slopes of the frontal sedimentary bodies, the terrain slope is relatively steep, the unloading of water bodies is strong, and the thickness of sedimentary sand bodies is relatively large. Additionally, because of the hydrodynamics of the depositional environment, the sediment in the sedimentary sand bodies in this area is well sorted, and the sand bodies have relatively good physical properties, resulting in the development of widely distributed high-quality reservoirs in the frontal slope areas.

刘萝卜锅

6. Conclusion

Stratigraphic-sedimentological forward modeling is based on the sequence stratigraphy theory, which can be used to analyze the effects of tectonic subsidence, lake level change, sediment supply and trans- port mechanism on reservoir sedimentation. The simulation results can be used to study the depositional stages, evolution process and dis- tribution characteristics of sand.

Based on the parameter sensitivity analysis, the river transport ca- pacity and sand ratio are the most important parameters in lacustrine delta-dominated sedimentary basins with fast deposition rates in China and should be the first model parameters to be adjusted. The maximum initial water depth, lake level change and tectonic subsidence are re- latively low-sensitivity parameters. By adjusting the high-sensitivity parameters first and then the low-sensitivity parameters and simulating the parameters many times, the sedimentary model that can best ex- plain the distribution characteristics of high-quality reservoirs can be determined and can be used as a conceptual model for high-quality reservoir prediction.

According to the characteristics of tight sandstone reservoirs, a method is proposed for predicting the distribution of high-quality re- servoirs using forward modeling. In this method, the sand ratio model is obtained by stratigraphic-sedimentological forward modeling, the sand ratio model is then transformed into a porosity model, and finally, product data of the sand content and porosity are obtained. The re- search method proposed in this paper is suitable for less-explored areas with scarce data. This work shows that this method can not only ef- fectively reproduce the spatiotemporal evolution characteristics of se- dimentary bodies but can also predict the parameters related to oilfield production from limited data. Additionally, further exploration in the field will reduce risk.

Increasing investment in less-explored areas will be the focus of tight sandstone reservoir research in the future. The prediction method proposed in this paper for high-quality sandstone reservoirs yields good prediction results for less-explored areas and contributes to the ex- ploitation of tight sandstone reservoir reserves by promoting a break- through in the production of tight sandstone reservoirs and effectively reducing exploration and development costs, thus improving interna- tional competitiveness during periods of low oil prices.

Acknowledgments

The research in this paper has been funded by major national oil and gas projects, including the Enrichment Regularity of Tight Oil & Gas and the Key Technology of Exploration and Development (Nos. 2016ZX05046-002, 2016ZX05047-005).

The author appreciates the Dionisos® software license provided by ESSCA Beijing company (the Chinese agent of Dionisos software) and thanks Dr. Yin Xiangdong for his valuable comments on the paper. In particular, the author thanks the anonymous reviewer who provided very good suggestions on the structure of the paper and improved the overall quality of the paper.

Appendix A. Supplementary data

Supplementary data to this article can be found online at

https:// doi.org/10.1016/j.marpetgeo.2018.11.027.

Supplementary data to this article can be found online at

https:// doi.org/10.1016/j.marpetgeo.2018.11.027.

刘萝卜锅

References

Alzaga-Ruiza, H., Granjeon, D., Lopez, M., Seranne, M., Roure, F., 2009. Gravitational collapse and Neogene sediment transfer across the western margin of the Gulf of Mexico: insights from numerical models. Tectonophysics 470, 21–41.

Berne, S., Rabineau, M., Flores, J.A., Sierro, F., 2004. The impact of quaternary global changes on strata formation: exploration of the shelf edge in the Northwest Mediterranean Sea. Oceanography 17 (4), 92–103.

Bitzer, K., Salas, R., 2002. SIMSAFADIM: Three-dimensional Simulation of Stratigraphic Architecture and Facies Distribution Modeling of Carbonate Sediments, vol. 28. Pergamon Press, Inc., pp. 1177–1192 10.

Bosence, D.W.J., Waltham, D.A., 1990. Computer modeling the internal architecture of carbonate platforms. Geology 18, 26–30.

Burgess, P.M., Lammers, H., Van Oosterhout, C., Granjeon, D., 2006. Multivariate se- quence stratigraphy: tackling complexity and uncertainty with stratigraphic forward modeling, multiple scenarios, and conditional frequency maps. Am. Assoc. Petrol. Geol. Bull. 90, 1883–1901.

Caers, J., 2011. Modeling Uncertainty in the Earth Sciences. Chichester:Wiley Blackwell. Chen, Jie, Liu, Chuanhu, Tan, Mingyou, Hongmei, Wei, 2016. Depositional model of

prograding delta confluences: a case from esm3 members in the paleogene dongying

sag. Acta Sedimentol. Sin. 34 (6), 1187–1197.
Cheng, G., Si, C.S., 2013. Study on sedimentary numerical simulation method of fan delta

sand body. J. Geol. 37 (2), 178–182.
Cloetingh, Ziegler, Beekman, Burov, Garcia-Castellanos, Matenco, 2015. second ed.

Tectonic Models for the Evolution of Sedimentary Basins, Treatise on Geophysics, vol.

6. pp. 513–592.
Cooper, Meireles, Green, Klein, Toldo, 2018. Late Quaternary stratigraphic evolution of

the inner continental shelf in response to sea-level change, Santa Catarina, Brazil.

Mar. Geol. 397, 1–14 2018.
Cross, T.A., Harbaugh, J.W., 1989. Quantitative dynamic stratigrapy: a workshop, a

philosophy, a methodology. In: Cross TA Quantitative Dynamic Stratigraphy.

Prentice Hall, pp. 3–20.
Csato, I., Granjeon, D., Catuneanu, O., Baum, G.R., 2013. A three-dimensional strati-

graphic model for the Messinian crisis in the Pannonian Basin, eastern Hungary.

Basin Res. 25, 121–148.
Dai, Jinxing, Ni, Yunyan, Wu, Xiaoqi, 2012. Tight gas in China and its significance in

exploration and exploitation. Petrol. Explor. Dev. 39 (3), 257–264.
de Paula Faria, Desiree Liechoscki, dos Reis, Antonio Tadeu, de Souza Jr., Olinto Gomes,

2017. Three-dimensional stratigraphic-sedimentological forward modeling of an Aptian carbonate reservoir of the sag stage of Santos Basin, Brazil. Mar. Petrol. Geol. 88, 676–695.

Dong, Gang, He, Youbin, 2010. Study on the restoration of ancient water depth based on formation thickness. J. Yangtze Univ.(Nat Sci Edit). 7 (3), 484–486.

Dong, Ning, Liu, Zhenfeng, Zhou, Xiaoying, 2008. Seismic detection and prediction of tight gas reservoirs in the Ordos Basin. Oil Gas Geol. 29 (5), 668–675.

Du, Jinhu, 2016. Chinese Continental Compact Oil. Petroleum Industry Press ISBN: 9787518312610.

Emery, D., Myers, K., 1996. Sequence Stratigraphy. Blackwell Science Pub, Co., London. Fang, Zhongyu, Wang, Liping, Du, Jiayuan, Liang, Lifeng, 2017. Nonlinear AVO inversion

based on hybrid intelligent optimization algorithm. OGP 52 (4), 797–804.
Fenwick, D., Scheidt, C., Caers, J., 2014. Quantifying asymmetric parameter interactions

in sensitivity analysis :Application to reservoir modeling. Math. Geosci. 46 (4),

493–511.
Granjeon, D., 1996. Modelisation stratigraphique deterministeeconception et applications

d'un modele diffusif 3D multilithologique. Memoires Geosciences Rennes, PhD

Dissertation. Geosciences Rennes, Rennes, France.
Granjeon, D., Joseph, P., 1999. Concepts and applications of a 3D multiple lithology. In:

Diffusive Model in Stratigraphic Modeling, Numerical Experiments in Stratigraphy: Recent Advances in Stratigraphic and Sedimentologic Computer Simulations. SEPM (Society for Sedimentary Geology), Tulsa, pp. 197–210.

Griffths, C.M., Dyt, C., Paraschivoiu, E., Liu, K., 2001. SEDSIM in hydrocarbon explora- tion. In: Merriam, D., Davis, J.C. (Eds.), Geologic Modeling and Simulation, pp. 71–97.

Gu, Wen, Xu, Min, Wang, Duohan, Zheng, Hong, Zhang, Xiong, Zhang, Dongjun, Luojun, 2016. Application of seismic motion inversion technology in thin reservoir predic- tion: a case study of the thin sandstone gas reservoir in the B area of Junggar Basin. Natural gas geoscience 27 (11), 2064–2069.

Harbaugh, J.W., Bonham-carter, G., 1970. Computer Simulation in Geology. John Wiley and Sons, New York.

Hawie, N., Barrois, A., Marfisi, E., Murat, B., Hall, J., El-Wazir, Z., Al-Madani, N., Aillud, G., 2015. Forward stratigraphic modelling, deterministic approach to improve car- bonate heterogeneity prediction; lower cretaceous, Abu Dhabi. In: Abu Dhabi International Petroleum Exhibition and Conference. Society of Petroleum Engineers.

He, Zixin, Fu, Jinhua, Xi, Shengli, Fu, Suotang, Bao, Hongping, 2003. Geological features of reservoir formation of Sulige gas field. Acta petroleisinica 24 (2), 6–12.

Hoy, R.G., Ridgway, K.D., 2003. Sedimentology and sequence stratigraphy of fandelta and river-delta deposystems, Pennsylvanian Minturn Formation, Colorado. AAPG Bull. 87 (7), 1169–1191.

Hu, Mingyi, Ke, Ling, Liang, Jianshe, 2010. The characteristics and pattern of sedimen- tary facies of huagang formation in Xihu Depression. J. Oil Gas Technol. (J. Jianghan Petroleum Inst.) 32 (5), 1–5.

Huang, Xinrui, Huang, Jianping, Sun, Qixing, Cui, Wei, 2016. Progress in seismic pro- cessing method of tight oil. Prog. Geophys. 1, 205–216.

Hutton, E.W.H., Syvitski, J.P.M., 2008. Sedflux 2.0: an advanced process-response model that generates three-dimensional stratigraphy. Comput. Geosci. 34, 1319–1337.

Jia, Chengzao, Zou, Caineng, Li, Jianzhong, Li, Denghua, Zheng, Min, 2012. Assessment criteria, main types, basic features and resource prospects of the tight oil in China. Acta petroleisinica 33 (3), 343–350.

Kenyon, P.M., Turcotte, D.L., 1985. Morphology of a delta prograding by bulk sediment transport. Geol. Soc. Am. Bull. 96, 1457–1465.

Li, Zichao, 2009. Computer simulation and software of filling process in sedimentary basins. Western Exploration Engineering 11, 71–74.

Li, Yonggen, Xu, Shengfeng, 2008. Application of seismic rock physics and forward si- mulation in predicting tight sands tone reservoirs. J. Oil Gas Technol. (J. Jianghan Petroleum Inst.) 30 (6), 61–66.

Li, Xianghui, Zhang, Jie, 1999. A review of sea level and sea level change. Sediment. Facies Palaeogeogr. 19 (4), 61–72.

Li, Shoujun, Zheng, Deshun, Jiang, Zaixing, Hu, Bin, Wang, Jinxiang, Jiao, Yehong, 2005. Water depth of palaeo -lacustrine basin recovered by dominance diversity of os- tracoda:an example from sedimentary period of the member 3 of shahejie formation of paleogene in dongying sag , shandong province. J. Palaeogeogr. 7 (3), 400–403.

刘萝卜锅

Li, Chunpeng, Liu, Zhiguo, Yang, Songling, Yin, Chuan, Yan, Qinghua, Tan, Zhuo, 2017. Study of reservoir lithology prediction method in low exploration block. Prog. Geophys. 32 (4), 1569–1573.

Lin, Changsong, Li, Sitian, Ren, Jianye, 1995. Sequence architecture and depositional systems of erlian lacustrine fault basins, North China. Earth Sci. Front. 2 (3), 124–132.

Lin, Changsong, Xie, Xinong, Zhang, Yanmei, Liu, Jingyan, 1998. Tow-Dimensional de- positional sequence simulation. Acta Sedimentol. Sin. 16 (2), 68–73.

Lin, Changsong, Liu, Jianyan, Hu, Bo, 2010. Computer simulation on the formation of depositional sequences in tectonic active basin: case study on rift and foreland basins. Acta sedmentologica sinica 28 (5), 868–874.

Liu, Zhenfeng, 2014. Key techniques of integrated seismic reservoir characterization for tight oil & gas sand. Prog. Geophys. 29 (1), 182–190.

Liu, Wanjin, Zhou, Hui, Yuan, Sanyi, Liu, Wenling, 2013. Applications of spectral in- version in the seismic attribute interpretation. OGP 48 (3), 423–428.

Lu, Hongbo, Jiang, Zaixing, Tang, Dahai, Yang, Weili, 1998. Application of sporopollen ecos -tratigraphic analyses to sequence stratigraphic study. J. Univ. Pet. (China) 4, 18–20 China.

Martinez, P.A., Harbaugh, J.W., 1993. Simulating nearshore environments. In: Computer Methods in the Geosciences. Pergamon Press, Oxford, pp. 12.

Nordlund, U., 1999. FUZZIM: forward stratigraphic modeling made simple. Comput. Geosci. 25 (4), 449–456.

Ouyang, Qianlin, Liu, Weilin, 2014. Variation characteristics of water level in poyang lake over 50 years. Resour. Environ. Yangtze Basin 23 (11), 1546–1550.

Overeem, I., Syvitski, J.P.M., Hutton, E.W.H., 2005. Three-dimensional numerical mod- eling of deltas. River deltas: concepts, models and examples. In: In: Bhattacharya, J.P., Giosan, L. (Eds.), 83. Soc. Sediment. Geol. Spec. Publ., pp. 13–30.

Pelletier, J.D., 2004. Persistent drainage migration in a numerical landscape evolution model. Geophys. Res. Lett. 31, L20501.

Plint, A.G., McCarthy, P.J., Faccini, U.F., 2001. Nonmarine sequence stratigraphy: updip expression of sequence boundaries and systems tracts in a high resolution framework, Cenomanian Dunvegan formation, Alberta foreland basin,Canada. AAPG Bull. 85 (11), 1967 e2001.

Rabineau, M., Berne, S., Aslanian, D., Olivet, J., Joseph, P., Guillocheau, F., Bourillet, J., Ledrezen, E., Granjeon, D., 2005. Sedimentary sequences in the Gulf of Lion: a record of 100,000 years climatic cycles. Mar. Petrol. Geol. 22, 775–804.

Rahman, A.H.A., Menier, D., Mansor, M.Y., 2014. Sequence stratigraphic modelling and reservoir architecture of the shallow marine successions of Baram field, West Baram Delta, offshore Sarawak, East Malaysia. Mar. Petrol. Geol. 58, 687–703.

Ren, Lai-yi, Fu, Junhui, Lin, Guifang, Zhang, Dongxia, 2001. The application of sporo - pollen fossil information function for palaeoenvironment analysis. J. Northwest Univ. (Natural Sci. Ed.) 31 (6), 506–508.

Robustelli, Ermolli, Petrosino, Jicha, Sardella, Donato, 2014. Tectonic and climatic con- trol on geomorphological and sedimentary evolution of the Mercure basin, southern Apennines, Italy. Geomorphology 214 (2), 423–435.

Seard, C., Borgomano, J., Granjeon, D., Camoin, G., 2013. Impact of environmental parameters on coral reef development and drowning: forward modelling of the last deglacial reefs from Tahiti (French Polynesia; IODP Expedition #310). Sedimentology 60, 1357–1388.

Sømme, T.O., Helland-Hansen, W., Granjeon, D., 2009. Impact of eustatic amplitude variations on shelf morphology, sediment dispersal, and sequence stratigraphic in- terpretation: icehouse versus greenhouse systems. Geology 37 (7), 587–590.

Surdam, 1997. Tight gas reservoirs–an unconventional natural energy source for the fu- ture. AAPG (Am. Assoc. Pet. Geol.) Bull. 88, 1–32.

Syvitski, J.P.M., Hutton, E.W.H., 2001. 2D SEDFLUX1.0: an advanced processresponse numerical model for the fill of marine sedimentary basins. Comput. Geosci. 27, 731–754.

Torra, O.G., Bitzer, K., Cabrera, L., Roca, E., 2009. SIMSAFADIM-CLASTIC: a new ap- proach to mathematical 3D forward simulation modeling for terrigenous and car- bonate marine sedimentation. Geol. Acta 7, 311–322.

Vesely, F.F., Rostirolla, S.P., Appi, C.J., Kraft, R.P., 2007. Late Paleozoic glacially related sandstone reservoirs in the Parana' Basin, Brazil. AAPG Bull. 91 (2), 151–160.

Wang, Guoshou, Zhou, Zuoming, Xiao, Chaohui, Li, Meizhou, Zhou, Weikui, Wang, Lin, 2002. Sedimentary characteristics of eogene Pinghu formation and huagang forma- tion in Chunxiao zone of xihu lake depression. Oil Gas Geol. 23 (3), 257–265.

Wang, Lingli, Dan, Weidong, Xin, Honggang, Ma, Yanli, Zhang, Zhenhong, Bingying, Hao, 2013. The recognition of prograded delta and palaeogeographic evolution of 6th member of yanchang formation in Jiyuan area of Ordos basin. J. Oil Gas Technol. (J. Jianghan Petroleum Inst.) 35 (2), 11–16.

Wang, Kaiyan, Xiao, Zengjia, Jiang, Yan, Pang, Siyu, Gao, Yuhang, 2017a. Seismic inversion method of stochastic stimulation in depth domain and its application. Prog.

Geophys. 32 (4), 1665–1672.
Wang, Zhiqiang, Cao, Siyuan, Chen, Hongling, Sun, Xiaoming, Fan, Ping, 2017b. Wave

impedance inversion based on tv regularization and toeplitz - sparse matrix factor-

ization. OGP 52 (6), 1193–1199.
Warrlich, G.M.D., Waltham, D.A., Bosence, D.W.J., 2002. Quantifying the sequence

stratigraphy and drowning mechanisms of atolls using a new 3-D forward strati-

graphic modelling program (CARBONATE 3D). Basin Res. 14, 379–400.
Xiu, H., Chris, D., Cedric, G., Tristan, S., 2012. Numerical forward modelling of 'flux-

oturbidite' flume experiments using SEDSIM. Mar. Petrol. Geol. 35, 190–200.
Xu, Guosheng, Zhao, Lili, Xu, Fa, Zhang, Tao, Zhao, Xingbin, Zhang, Zhen, Liu, Wei, 2012. Seepage flow characteristics of tight sandstone reservoirs in Huagang Formation in a

certain structure of Xihu depression in East China Sea Basin. J. Chengdu Univ.

Technol. (Sci. Technol. Ed.) 39 (2), 113–121.
Xue, Shuhao, Liu, Wenlin, 1999. Sedimentary Geology and Hydrocarbon Exploration in

Lake Basin. Petroleum Industry Press, pp. 62–65.
Yang, Weili, Wang, Yi, Zhang, Zhongchen, Chen, Liang, Li, Jingchang, 2017. A three-

dimensional numerical sedimentary simulation model for the lower-middle ordovi-

cian of shunnan area, tarim basin. Acta Sedimentol. Sin. 35 (1), 117–123.
Ye, Tairan, Fu, Shun, Lv, Qibiao, Lai, Weirong, 2009. Applicaiton of multiwave joint

inversion to the prediction of relatively high-quality reservoirs - an example from the prediction of deep tight clastic sandstone reservoirs in the western Sichuan Basin, Basin. Oil Gas Geol. 30 (3), 357–362.

Yin, Xiangdong, Huang, Wenhui, Wang, Pengfei, Wang, Qiming, Yan, Deyu, Zhou, Xiaozhou, 2015. Sedimentary evolution of overlapped sand bodies in terrestrial faulted lacustrine basin: insights from 3D stratigraphic forward modeling. Mar. Petrol. Geol. 86, 1–13 2015.

Yin, Xiangdong, Huang, Wenhui, Wang, Pengfei, Wang, Jiayin, Wang, Qiming, YanDeyu, Zhou, Xiaozhou, 2017. Sedimentary evolution of overlapped sand bodies in terrestrial faulted lacustrine basin: insights from 3D stratigraphic forward modeling. Mar. Petrol. Geol. 86, 1431–1443.

Yu, Xinghe, 2008. Sedimentology of Reservoirs in Clastic Rocks. Petroleum Industry Press ISBN: 9787502163990.

Yu, Xinghe, Li, Shengli, Li, Shunli, 2013. Texture- genetic classifications and mapping methods for deltaic deposits. Acta Sedimentol. Sin. 31 (5), 782–797.

Zhang, Jianmin, Huang, Kai, Liao, Xinwu, Yue, Honglin, Zhang, Lan, 2015. Reservoir  prediction and application of progradation reservoir, E2S3M southern Bohai water.