Application of Seismic Attribute Analysis Technology Guided by Model Forward Modeling in L29 Area

: Taking the SaErTU and Putaohua oil layers in the L29 well area of LHP oilfield in the northern Songliao Basin as the research object, In response to the difficulty of interference between thin interbedded sand and mudstone and strong reflection between strata, which have a significant impact on fine prediction of sand bodies, a stratigraphic model is established based on the geological characteristics of the target layer to eliminate the impact of stratigraphic reflection ； Then add a thin layer of sand body to its interior, establish a thin interlayer model, and obtain a geological model that is more in line with the actual situation of the target layer. Using forward simulation, analyze the seismic response characteristics of sand bodies, extract 30 seismic attributes from 4 categories: amplitude statistics, composite seismic trace statistics, sequence and frequency spectrum statistics, and calculate the correlation between cumulative sandstone thickness and seismic attributes, and select a sensitive seismic attribute set;By combining sedimentary and drilling data, the seismic attribute with the highest sensitivity to the target layer is selected. The multi-level dimensionality reduction and gradual improvement of seismic attribute selection methods using "model forward modeling, attribute analysis, relevant optimization, and drilling implementation" can effectively improve the prediction accuracy of thin interbedded sand bodies and greatly reduce the risk of oilfield exploration and development.


Introduction
Seismic attributes are special measurements of geometric, kinematic, dynamic, or statistical characteristics derived from pre stack or post stack seismic data through mathematical transformations.The research and application of seismic attribute analysis technology play an important role in structural interpretation [1][2], reservoir prediction [3][4][5], and fluid detection [6][7].But there are many types of seismic attributes, and different seismic attributes represent different meanings and play different roles in different fields.Some seismic attributes have clear physical meanings and are directly related to reservoir parameters or geological features, such as coherence, curvature, and ant body attributes, which mostly reflect fault characteristics [8].However, most seismic attributes do not have a clear physical meaning, only mathematical operations on seismic amplitude, frequency, and phase.It is necessary to analyze the geological significance represented by seismic attributes based on actual geological conditions.Optimizing seismic attributes or combinations of seismic attributes is a key issue in achieving reservoir prediction [9].Many scholars at home and abroad have conducted research on the prediction of thin interbedded sand bodies, and have achieved certain results.An Peng et al. [10] On the basis of analyzing the tuning effect, time-frequency analysis technology is used to optimize the sensitive frequency, and the data volume is reconstructed based on the sensitive frequency to eliminate similar single sand body thicknesses but significant differences in seismic waveform response characteristics.The sensitive seismic attributes are optimized to predict the distribution of sand bodies.Zhang Junhua et al. [11] used 90 ° phase shift technology to improve the well seismic relationship and effectively improve the accuracy of reservoir prediction for thin interbedded reservoirs.Peng Zuolei et al. [12] used principal component analysis to perform principal component analysis on various seismic attributes, establish the connection between multiple seismic attributes and thin reservoirs, reduce the multiplicity of single seismic attribute analysis, and use support vector machine method to predict thin reservoirs.Liu Wei et al. [13] established a typical thin interbedded sand and mudstone model, focusing on analyzing the variation characteristics of 8 seismic attributes, and summarized their variation patterns with the thickness of thin interbedded sandstone, providing a certain theoretical basis for optimizing the prediction attributes of thin interbedded sand and mudstone reservoirs.
From the literature, scholars in the past mainly focused on the study of the interference effect between thin interbedded sand bodies, and there was little research and practical analysis on the impact of strong reflection between formations on reservoir prediction.This article takes the SaErTU and Putaohua oil layers in the L29 well area of L oilfield as examples.Based on the geological characteristics of the target layer and the development of the sand body, the influence of stratigraphic reflection is first considered to establish a stratigraphic model, and then a thin layer of sand body is added to its interior to obtain a more practical geological model.Afterwards, seismic forward modeling was conducted to calculate the correlation between cumulative sandstone thickness and seismic attributes, and a sensitive seismic attribute set was selected.By combining sedimentary and drilling data, the seismic attributes with the highest sensitivity to the target layer are selected to improve the prediction accuracy of thin interbedded reservoirs.

Geological Overview of the Work Area
The L29 area is located on the eastern slope of the L anticline in the northern part of the Songliao Basin.The Sa and Pu oil layers in this area are controlled by the Qiqihar sedimentary system in the northwest and the Nehe Yi'an sedimentary system in the north.The delta front channel sand bodies and sheet sand bodies are vertically staggered and severely heterogeneous, making them typical thin interbedded reservoirs (Figure 1).The interference between sand bodies has a significant impact, making it difficult to predict advantageous sand bodies.

Optimization of forward modeling attributes for earthquake models 3.1 Wedge model
As shown in Figure 2, a set of wedge-shaped body models is designed, with the thickness of the wedge-shaped body gradually increasing from 0m to 40m from left to right.
The proportion of internal sandstone is 25%, 50%, 75%, and 100%, respectively.In the model, yellow represents sandstone, with a velocity of 4650m/s and a density of 2. 5g/cm3; White represents the surrounding rock, with a velocity of 3800m/s and a density of 2. 4g/cm3.The seismic reflection record is obtained by using a 45 Hz main frequency Reyker wavelet and a self excited and self received ray tracing method.

Comparison of Sand Bodies in PuTaoHua Reservoir
Comparison of Sand Bodies in Sartu Reservoir Extract the amplitude and frequency of each wave peak on the top surface of the wedge-shaped body model, and draw the amplitude and frequency tuning curve (Figure 3).The amplitude tuning curve indicates that regardless of how the sandstone ratio changes, the amplitude reaches its maximum near the tuning thickness.But as the sandstone content decreases, the tuning thickness decreases, the amplitude weakens, and the intensity changes become more complex.The frequency tuning curve shows that the thickness of the wedge-shaped body is below 27m, and the frequency decreases with the increase of thickness.
The change in sandstone content has little effect on the frequency attribute; When the thickness is greater than 27m, the frequency begins to increase.The sandstone content has a significant impact on the frequency attribute.
The smaller the sandstone content, the greater the rate of frequency increase.The forward modeling of the thin interbedded sand and mudstone model (Figure 2 and Figure 3) shows that both changes in sandstone content and thickness can cause significant changes in seismic amplitude.The amplitude attribute can be used to predict the development of sandstone [14-16] .However, in practical applications, it has been found that the thin interbedded model did not take into account the impact of strong reflections between strata on the reflection shielding of sand bodies.As a result, starting from this forward model, it can be found that multiple seismic attributes have good sensitivity to sand bodies and high consistency.However, the predictive ability of the seismic attribute sand bodies extracted in practice varies greatly.It is difficult to optimize the seismic attributes that are most sensitive to sand bodies using this type of model.

Forward modeling of thin interbedded sand and mudstone based on stratigraphic models
Thin interbedding refers to a set of strata composed of multiple thin layers that cannot be distinguished from each single layer by seismic reflection.Thin interbedded reflection wave is a composite wave formed by the interference of multiple thin layer reflections, which cannot distinguish each single layer.Therefore, the thin interbedded reflection wave is the overall seismic response of the thin interbedded layer [17].The impact of strong reflections between superimposed strata on the reflection shielding of thin interbedded sand bodies has brought significant difficulties to the prediction of thin interbedded sand bodies.The accuracy of seismic attribute prediction results is often low and the multiplicity of solutions is strong.Conventional thin interlayer models are difficult to characterize the seismic

Sandstone ratio 25%
Sandstone ratio 50% Sandstone ratio 100% Sandstone ratio 75% response characteristics of the target layer and cannot be used for seismic optimization.In order to improve prediction accuracy and reduce multiplicity, a geological model is first established to eliminate the strong reflection shielding effect between strata, and then sand bodies are added to it to establish a thin interbedded model of sand and mudstone.The results obtained using this geological model will be more targeted and applicable.Analyze the drilling data in the area, determine the velocity and density parameters of surrounding rock, sand, and mudstone in each formation (Table 1), and design stratigraphic models for the Sartu and Putaohua oil layers based on geological characteristics.Obtain forward seismic reflection records using the self excited and self collected ray tracing method.As shown in Figure 4, the forward modeling results of the formation are as follows: strong amplitude peak reflection on the top surface of the S II oil layer group, medium strong amplitude peak reflection on the S I top, S III top, and P top surfaces, while medium strong amplitude valley reflection is formed at the bottom boundary of P due to contact with the underlying low-speed formation.The seismic forward model is consistent with the reflection characteristics of the actual seismic profile, indicating that the model parameters are relatively reasonable.Ultimately, the sand mudstone thin interbedded model can effectively eliminate the influence of interlayer reflection.After determining the stratigraphic model, sand bodies are added layer by layer in each oil layer group to establish a thin interbedded model of sand and mudstone.As shown in Figure 5, taking the S Ⅰ oil reservoir group as an example, the thickness of a single layer of sandstone is 2 meters, with an interval of 1 meter.The width of each single row of sand bodies is 300 meters, and the number of sandstone layers in the model gradually increases from 1 to 7 from left to right.Forward seismic reflection records were obtained using the self excited and self received ray tracing method using 45Hz Reke wavelet.From the seismic forward modeling, it can be seen that as the thickness of sandstone increases, the amplitude shows a significant increasing trend.Although the number and total thickness of thin interbedded layers of sand and mudstone in these 7 columns are different, their seismic reflection characteristics are roughly the same, that is, the reflection waveforms are all composite waves with a single waveform, and there is no increase in the intermediate reflection interface compared to the stratigraphic reflection.

Seismic attribute optimization
For the seismic forward modeling of thin interbedded sand and mudstone in the S Ⅰ oil layer group, a total of 30 seismic attributes were extracted from the top and bottom interfaces of the oil layer group, including amplitude statistics, composite seismic trace statistics, sequence and spectrum statistics, and a graph of the variation of seismic attributes with cumulative sandstone thickness was drawn.
As shown in Figure 6  For example, the maximum peak amplitude value in Figure 6 increases with the increase of cumulative sandstone thickness, which has a good correlation with sandstone thickness.It reaches its maximum value at a thickness of 10m, and then the maximum amplitude value slightly decreases.Due to the average development of sandstone in the S Ⅰ oil reservoir group, the accumulated maximum thickness is only about 10m after drilling, so the maximum peak amplitude can be used to predict the development of sandstone.Among the amplitude statistical attributes, the root mean square amplitude attribute has a good correlation with sandstone thickness and can also be used to predict the development of sandstone.As shown in Figure 7, in the statistical category of composite seismic traces, there is a negative correlation between the average instantaneous frequency attribute and the cumulative sandstone thickness.As the sandstone thickness increases, the attribute values monotonically decrease.The two have a good correlation and can be used to predict the development of sandstone.
In the sequence and spectrum statistics category (Figure 8), the correlation between various seismic attributes and cumulative sandstone thickness is generally slightly worse than that of amplitude statistics and composite seismic trace statistics, and their seismic attributes are not suitable for predicting sandstone in the S Ⅰ oil layer group.Therefore, based on the statistical analysis of the thin interbed forward model of the stratigraphic model, a seismic attribute set with maximum peak amplitude, root mean square amplitude, and average instantaneous frequency is preliminarily selected.Based on the preliminary determination of sensitive attributes, further attribute optimization is carried out based on regional sedimentary characteristics and drilling data.As shown in Figure 9, based on the development of sandstone in the S1 oil layer group, it is believed that a cumulative sandstone thickness of over 3m is the advantageous reservoir for this oil layer group.Therefore, the seismic attribute prediction of the oil layer group is adjusted to yellow and red for areas with a cumulative sandstone thickness of more than 3m, and to blue for areas with a cumulative sandstone thickness of less than 3m.
Calculate the prediction accuracy of each sensitive seismic attribute, with a maximum peak amplitude prediction accuracy of 81.3%, a root mean square amplitude prediction accuracy of 81.3%, and the average instantaneous frequency prediction accuracy of 87.5%.
During the sedimentation period of the S Ⅰ oil layer formation, as the lake level rises, the delta gradually retreats northward, with a large area of inter flow bay facies distributed in the southern part, and shore-shallow lake facies visible in the southeast.The average instantaneous frequency attribute can better characterize this sedimentary feature.Therefore, the average instantaneous frequency attribute is ultimately selected to characterize the development of sandstone in the S Ⅰ oil reservoir group.The above methods can be used to optimize the seismic attributes that are most sensitive to sandstone and conform to sedimentary characteristics.

Application Effect Analysis
Adopting the forward seismic attribute optimization method based on the stratigraphic model for the thin interbed model, the seismic attribute optimization was carried out for the three oil layer groups of the Sartu oil layer and the three sandstone groups of the Putaohua oil layer.During the sedimentation period of the Sartu oil layer, it was mainly composed of sheet-like sand deposits, with sand bodies distributed in a sheet-like manner, good continuity, and relatively large scale.Figure 10 shows the seismic attribute maps selected for the three oil layer groups in Sartu, with an average instantaneous frequency prediction accuracy of 87. 5% for the S-Ⅰ oil layer group, an average amplitude attribute prediction accuracy of 84.4% for the S-Ⅱ oil layer group, and an reflection intensity slope prediction accuracy of 81.3% for the S-Ⅲ oil layer group.The attribute prediction shows the development of sandstone in each oil layer group, with a flaky distribution and good continuity, which can better characterize the sedimentary characteristics of each oil layer group in the area and effectively predict the sandstone development area.During the sedimentation period of the Putaohua oil layer, there was a sedimentary feature of water intrusion from bottom to top.The underwater distributary channels in the delta front are frequently diverted and vertically distributed in staggered and continuous layers.On the plane, the lateral changes of the reservoir are rapid, and the sand bodies are mainly distributed in strip or lens shapes, with short extension distance and small development scale.Adopting the forward seismic attribute optimization method based on the stratigraphic model, the root mean square amplitude attribute of the upper sand formation of the Putaohua oil layer is optimized, with a prediction accuracy of 81.3%.The minimum amplitude attribute is selected for the sand group in the grape blossom, with a prediction accuracy of 84.4%.The optimal average absolute amplitude attribute for the lower sand formation of the Putaohua oil layer is selected, with a prediction accuracy of 84.4%.The selected seismic attribute predictions have a coincidence rate of over 80% and can better characterize the sedimentary characteristics of various sandstone formations in the Putaohua oil layer.

Conclusion and Understanding
1) The forward modeling of the thin interbedded wedgeshaped body model shows that both changes in sandstone content and thickness can cause significant changes in seismic amplitude.The amplitude attribute can be used to predict the development of sandstone.However, in practical applications, it has been found that the thin interbedded model does not take into account the impact of strong reflections between strata on the reflection shielding of sand bodies, making it difficult to select the most sensitive seismic attributes for sand bodies.The multi-level dimensionality reduction and gradual improvement of seismic attribute selection methods using "model forward modeling, attribute analysis, relevant optimization, and drilling implementation" can effectively improve the prediction accuracy of thin interbedded sand bodies.This greatly reduces the problem of multiple solutions in predicting seismic attributes of thin interbedded sand bodies, and provides an effective method for optimizing seismic attributes of thin interbedded reservoirs.

Figure 1 .
Figure 1.L29 area Comparison of Sand Bodies in the Development Zone of Sartu and Putaohua Oil Layers

Figure 3 .
Figure 3. Wedge body model amplitude and frequency tuning curve diagram

Figure 4 .
Figure 4. Analysis of Forward Modeling Results of Stratigraphic Models

Figure 5
Figure 5 Geological model and seismic forward modeling of thin interbedded layers in S-I oil reservoir group , the horizontal axis in the figure represents the forward seismic channel number, the right vertical axis represents the cumulative thickness of sandstone, and the black line represents the thickness values of sandstone for different channels.It can be seen that as the channel number increases, the cumulative sandstone thickness increases, and the size of the channel number can represent the size of the cumulative sandstone thickness.The left vertical axis represents the normalized seismic attribute values, and the colored lines represent the variation curves of various seismic attributes with the cumulative thickness of sandstone.If a certain seismic attribute value monotonically increases or decreases as the channel number increases, it indicates a good correlation with the cumulative sandstone thickness and can be used to predict the development of sandstone.

Figure 6
Figure 6 Amplitude statistical seismic attributes versus cumulative sandstone thickness variation

Figure 7 .Figure 8 .
Figure 7. Changes in cumulative sandstone thickness due to statistical seismic attributes of composite seismic

Figure 9 .
Figure 9. Preliminary Selection of Seismic Attribute Set for S-I

Figure 10 .
Figure 10.Optimization Results of Properties of Various Oil Layer Groups in Sartu Reservoir

Figure 11 Optimization
Figure 11 Optimization Results of Properties of Each Sandstone Formation in Putaohua Reservoir

Table 1
Forward Modeling Parameters of Each Layer Group