S-wave velocity estimation using converted-wave VSP data
Zhang, Minyu 1980-
MetadataShow full item record
One of the key benefits of zero-offset vertical seismic profile (VSP) is to provide accurate velocity information in the vicinity of boreholes. For 3-component VSP, we usually pick the first break times of direct downgoing S-wave to estimate the S-wave velocity. However, the direct downgoing S-wave arrivals may be too weak to pick precisely. To mitigate this disadvantage, I present a new S-wave velocity estimation method using converted (P-to-S) waves in VSP data. Traveltimes of the strongest converted PS-wave are used to build an S-wave velocity model. Two synthetic traveltime datasets are generated and used to test the method. The first one, without noise, is used to investigate the sensitivities of results to different initial models. The inversion results are slightly different: the RMS error between the true velocity and the inverted velocity varies from 1 m/s to 10 m/s for different initial models. The second test, with random noise, explores the sensitivities of results to errors in picked traveltimes. Results show that the RMS error of output velocity varies from 2 m/s to 92 m/s when the RMS error of input traveltimes ranges from 0.1 ms to 8 ms. In addition to the synthetic tests, real 3-component zero-offset VSP and offset VSP datasets from the Huabei oil field, China are examined using the proposed method. The VSP survey used a dynamite source and 3-component receivers, with the receiver depth from 360 m to 2020 m. The inversion results show that, from 360 m to 900 m depths, the S-wave velocity derived from offset VSP fits quite well with that obtained from zero-offset VSP. The Vp/Vs is approximated 2.8 and the lithology is mudstone with sandstone. From 910 m to 2020 m, the S-wave velocity derived from offset VSP is close to that from zero-offset VSP. The Vp/Vs is around 1.7 and the lithology is primarily limestone and dolomite.