| As a critical target for the hydrocarbon resource, the deep water canyon’s sand body is paid more and more
attention by the academia and industry. Its reservoir modeling has become a hotspot and challenging issues in current
research, especially for areas covered with sparse well pattern and low seismic resolution. This study takes the Lingshui
section of the Central Canyon in Qiongdongnan Basin as an example. The pseudo-well is setted by integrating seismic
attributes, P-impedance data and sedimentary forward simulation result. Then, with the layered modeling strategy being
carried out, the reservoir model is built according to the evolution process of the canyon. We select corresponding
modeling strategies for different stages of canyon morphology and sedimentation process:(1) The initial formation stage of
the canyon: being in a non restricted sedimentary background, and the turbid water channel inside the canyon is prone to
oscillation, making it difficult to determine the boundary of the sand body. Only a large range of composite channel zones
and landslide bodies can be identified. So we use well seismic data to fit virtual well data, and use sequential Gaussian
simulation for modeling. (2) Stable development stage of the canyon: the upstream is a non restricted sedimentary
background, which can be identified a large number of landslide bodies. The downstream is a restricted sedimentary
background, with the development of multi-stage stacked turbidite channels with relatively straight morphology, which
accumulate at the bottom of the canyon. So we select the identified composite waterway zone and use truncated Gaussian
simulation for modeling. (3) The later reconstruction stage of canyon: the restrictive canyon is infilled by the lateral and
vertical accretion channels, which could be identified more obviously from the seismic profiles. For isolated waterways in
the upstream area, firstly, the size of the sand body is statistically analyzed, then a quantitative three-dimensional training
image is established, and finally, multi-point geostatistics methods are used for modeling. For downstream connected
waterways, truncated Gaussian simulation is still used for modeling. By selecting targeted modeling strategies for different
sedimentary filling processes, the reliability of predicting the distribution of sand bodies in each stage can be maximized. |
