The complexity of geological phenomena determines that bifurcation and confluence are common in river sediments, so the common river model is much more complicated than the above single river model. For the bifurcation and merger of rivers, this paper only takes simple but common geological phenomena such as bifurcation of main channel and tributary channel merging into main channel or tributary channel as the main research object, and discusses the method of establishing its skeleton model. Due to the limitation of research time and actual geological conditions, the new complicated situation of bifurcation on the bifurcated river channel is not considered for the time being. In the process of research, we must first distinguish between main channels and sub-channels; Secondly, considering the position relationship between the main channel and the branch channel, determine which main channel the branch channel belongs to; Finally, the center lines of the main channel and the sub-channels are predicted, so as to establish the channel skeleton model including the sub-channels.

For the branch channel, it is still assumed that its distribution range can be determined by the search window determined by its fluctuation amplitude. Therefore, for the water system with inlet, the inlet can still be analyzed according to the method of establishing single channel skeleton model. But at this time, the channel condition data in the search window is not necessarily the main channel deposit, but the sub-channel deposit, so it is necessary to carefully analyze the condition data in the search window. Because of the particularity of petroleum industry, dominant sand bodies are often selected for drilling, so the probability of drilling in the main channel is much higher than that in the branch channel. Therefore, it can be decided whether it is a main channel or a branch channel according to the amount of channel condition data falling into the search window. When the river condition data exceeds the specified value, it can be temporarily considered as siltation of the main river. This is because when there are both upstream tributaries and downstream tributaries, the conditional data in the search window may be composed of the conditional data of two tributaries, so it is necessary to analyze the conditional data. If the river condition data are divided into upstream and downstream groups, the upstream and downstream situation data are few, which may be tributary rivers; If the channel data are completely distributed in the upstream or downstream, it can be temporarily classified as tributary deposits because the direction or source of channel extension can not be found. Then consider the distance from the nearest main channel, and judge whether it is the main channel or the branch channel through the distance. If it is within a certain distance, the upper and lower reaches are defined as branch channels, otherwise it is regarded as main channel deposits. When the river condition data falling within the search window does not exceed the specified number, it is also necessary to analyze these river data points. If the distance between these river data points and the center line of the nearest main river exceeds a certain limit, then it can be randomly determined whether it is a tributary or a main river with random numbers. If the distance from the center line of the main channel does not exceed the limited range, it is considered as a branch channel.

Once the branch channel is determined, it is necessary to judge the extension direction of the branch channel and which main channel it belongs to.

First, judge which main channel the branch channel belongs to. According to the method that the distance between the conditional data point and the center line of the main channel is minimum, it is determined that the branch channel belongs to the nearest main channel. Secondly, according to the position of this conditional data point, it is determined whether it is located on the left or right side of the main river channel. Third, according to the condition data and the source direction, determine whether it is located in the upper or lower reaches of the river. If it is located in the upper reaches of the river, then this tributary flows into the main river. At this time, use the search window to search the remaining conditional data points that do not belong to the main channel or the tributary channel along the possible inflow direction, and if they fall within the search range, add the points to the tributary channel; If it is located downstream of the river, then this tributary is caused by the bifurcation of the river. Similarly, use the search window to search the remaining conditional data points that do not belong to the main channel or the branch channel along the possible branch direction, and if they fall within the search range, add the points to the branch channel.

It should be noted that because the tributary channels do not necessarily span the whole study area, the distribution of the generated tributary channels can be defined in advance according to the extension distance of the tributary channels. A tributary channel may encounter other main channels or tributary channels in its extension direction. At this time, it can be considered that the tributary channel merges into the main channel or tributary channel here to form the main channel or tributary channel, and the merged tributary channel extends and terminates.

After determining the main channel and tributary channel, the center line of the channel can be predicted. Using one-dimensional Gaussian function to generate river centerline, the prediction method is similar to that of a single river centerline, so I won't repeat it here. Fig. 5-9 is a channel skeleton model with branched channels generated by the above method.