Near surface velocity adjustments in presence of rugose water bottom A quantitative approach for Canyon Scaling

Transcription

1 10 th Biennial International Conference & Exposition P 406 Near surface velocity adjustments in presence of rugose water bottom A quantitative approach for Canyon Scaling Himanshu Kumar*, Pramod Srivastava*, Ashoka Dubey,* M.K.Balasubramaniam #, Subrata Chakraborty # Summary Velocity model building process tries to correct initial velocity model to an optimally correct model. The error in the initial velocity model plays an important role in controlling the number of tomographic iterations. Smoothing is an integral part of initial velocity building process. Near water bottom zones are prone to a higher degree of velocity errors in the initial model. The rugose water bottom specifically in presence of narrow and deep canyon cuts may come up with initial velocity models showing loss of sub-parallel properties of interval velocity near water bottom. This type of errors may cause spurious imaging effects even to deeper levels; corrective measures may involve huge efforts during tomographic inversion. Here authors have presented a simple but efficient workflow to account for the major part of near water bottom surface velocities. It is done through estimating canyon depths and converting them into interval velocity corrections for shallower zones of the model. The process may be called as canyon scaling of interval velocities and considerably reduces further tomographic efforts. Keywords: Velocity model building, Canyon scaling Introduction Depth Imaging & velocity model building for deep water environments has some explicit challenges like rugose water bottom, canyons, etc. If these effects are not accounted for in velocity model building, then this may lead to artefacts and may result in ambiguous interpretation (Figure 1). These artefacts are caused due to interval velocity variations within the shallow sediments due to variations in seafloor depth because of the canyons & cliffs. Normally the velocity contours should approximated follow the seabed topography (refer to figure 2), in fact uncompacted sediments near edges of canyons cause interval velocities sub-parallel to water bottom. However, smoothing involved during the initial velocity model preparations, the velocity contours no longer follow the seabed topography, leading to the inaccurate velocities in the complex water bottom regimes. We can make the velocity contours follow the seabed topography during the generation of initial velocity model. For this purpose we can flatten the velocity model along the water bottom horizon and then apply the smoothing followed by inverse flattening the smoothened velocity model. This will result in the velocity contours to follow the seabed topography from water bottom to the end of data and this velocity can cause severe artefacts in the seismic image. Using such a model will require a lot of tomographic updates to fix the velocity model (please refer to figure 3). Conventional & Contemporary Solutions The conventional solution for these kinds of near seabed velocity anomalies is to fix the velocity model in the shallow part while smoothing during creation of initial velocity model or via velocity model building. Obviously any error in the velocity model is possible to be corrected through tomographic inversions, but resolving these kinds of near seabed velocity anomalies may require a number of tomographic updates which is very time consuming and expensive. *Petroleum Geo-Services, Mumbai, # Reliance Industries Ltd., Mumbai Petroleum Geo-Services, 202/B, Everest Nivara Infotech Park-I, Plot No. D-3, TTC, MIDC, Turbhe, Navi Mumbai himanshu.kumar@pgs.com

2 There is a lot of discussion & literature available regarding the deep water challenges but it was found out that there is not much technical literature available on how to deal with these kinds of deep water challenges. Sergey Birdus (2008) presented a technique which utilizes complex geomechanical modelling to deal with velocity variations immediately below rugose seafloor. It is based on the fact that stress applied to given sediments changes the seismic velocities. The geomechanical solution provided by Birdus involves: 1. Building the initial interval velocity model using RMS velocities 2. Applying geomechanical correction for areas with varying seafloor depth 3. Using seismic tomography to finalize the model including the shallow part 4. Employing pre-stack depth migration for seismic imaging The geomechanical solution does not mention the relationship between the depth of the anomaly and the velocity difference. In our opinion, it is difficult and hard to use if you don t have the idea of geomechanical principles. Some more insights may be found out from the work done by Stewart et.al. (2007)SPG, Fruehn, J. et al (2008). We have tried to develop a simple & easy quantitative approach for this problem, where we try to derive a relation between the depth of the anomaly and the velocity difference. Below we describe a quantitative approach to take care of most of the deep water canyons for velocity model building. This procedure is known as Canyon Scaling. Canyon Scaling The canyon scaling approach deals with preparation of initial velocity model. Initial velocity model is in general prepared through smoothing of velocity field obtained from PSTM. The smoothing of velocity model is done in Time domain. The following assumptions are made for the canyon scaling approach: The velocity contours follow the water bottom below the canyon The depth of canyon is linearly related to the slow down in velocity required in the model. The velocity effect diminishes at some level, i.e. the iso-velocity contours no longer follow the canyoned waterbottom. This level may range from 200 to 500 m below seabed. In our case we found that this depth is around 500m below the water bottom. Canyon scaling approach involves the following steps: 1. Define the canyon zones within the data 2. Build a canyon scalar based on the canyon depth and the velocity difference up to the required depth below water bottom 3. Smooth the canyon scalar as required 4. Apply the smooth canyon scalar to the velocity model 5. Run a Pre-Stack depth migration to verify the results Step 1: Define the canyon zones The water bottom horizon (WBZ) is heavily smoothened to get a regional water bottom (WBZR) as shown in figure 4 below. The difference between WBZ & WBZR is known as hdiff. A Canyon Zone is defined as the one where hdiff is positive and a No Canyon Zone is defined where hdiff is negative or zero. Canyon scaling is applied to these zones only, where hdiff is positive. The concept of regional water bottom will obviously focus on narrow cuts which are marked by the highest degrees of inaccuracies generated during preparation of initial interval velocity model. Step 2: Build a canyon scalar based on canyon depth & velocity difference For this step we have to determine the depth (below water bottom) up to which the scalar will be generated and applied to the velocity model. In our case we found that the required depth is around 500m below the water bottom. 2

3 Hence, we define a layer L1 from WBZ to WBZ+500m. A global average velocity (Vglobal) is calculated in the no canyon zone (zone where hdiff is negative or zero) within the layer L1. Then a trace average velocity (Vtrace) is calculated in layer L1 in the canyon zone (zone where hdiff is positive). Now a trace-by-trace Raw Scalar is defined in the canyon zones as per equation (i) below: The Raw Scalar is then plotted against hdiff to derive a linear relation between them, over different areas of the velocity volume. Figures 5 show the plots of Raw Scalar vs. hdiff and the derived relationships. Based on the analyzed results the following relation is selected. In this particular case the relationship is as under: Using the above relationship a canyon scalar volume is generated. Step 3: Smooth the canyon scalar The canyon scalar derived in the step 2 above is smoothed using a median filter before applying to the velocity model. Step 4: Apply the smooth canyon scalar to the velocity model The smoothed canyon scalar volume is multiplied to the initial velocity model. This scalar is gradually set to 1 at some depth below water bottom, in this case it is WBZ+500m. Also, for the no canyon zones (i.e. zones where hdiff is negative or zero) the scalar is set to 1. This way the canyon scalar is applied to the canyon zones only. The velocity model so obtained is the initial velocity model after canyon scaling. Conclusion The quantitative canyon scaling approach can be used to build the initial depth interval velocity models for depth imaging. It is quite a simple and straight forward method to deal with the challenges posed by deep water canyons. It will be reducing number of tomographic updates for shallower zones. Acknowledgements We express our gratitude to N. Sinha, Head Expolration, Reliance Industries Ltd. for carrying out this project and for granting the permission to present the results. We would also like to thank Bruno Virlouvet(PGS) for his help & support throughout the project. The initial concepts of the process presented by Ed Levis are also acknowledged. Also we would like to thank Gajendra Joshi(PGS), Ashish Kumar(ONGC) & Manish Pandey(PGS) for their help during this project. References Birdus, S. 2008, Restoring velocity variations below seafloor with complex topography by geomechanical modeling, SEG Las Vegas 2008 Annual Meeting Stewart, P.G. & Jones, I.F., 2007, Solutions for Deep Water Imaging, SPG, Geohorizons Fruehn, J., Jones, I.F., Valler, V., Sangvai, P., Biswal, A. & Mathur, M., 2008, Resolving near-seabed velocity anomalies: Deep water offshore eastern India, Geophysics, 73, VE235 VE241 Step 5: Run a Pre-Stack depth migration to verify the results A pre-stack depth migration is run using the velocity models with and without canyon scaling to verify the results. Figures 6 & 7 below show the results of canyon scaling. 3

4 Figure 1: Presence of canyons in water bottom causes distortions like sagging, pull down, amplitude stripes, etc. in the deeper part of the imaged depth section. (Data courtesy: RIL) Figure 2: The velocity contours should follow the seabed topography 4

5 Figure 3: Smoothing of initial velocity model (Data courtesy: RIL) A) The velocity contours do not follow the seabed topography if normal smoothing is carried out. This causes artefacts in the seismic image. B) To make the velocity contours follow the seabed topography we can first flatten the water bottom and then smooth the velocity and then inverse flatten the water bottom again. In this case the velocity contours follow the seabed topography from top to the end of data. This will cause severe artifacts throughout the seismic image. C) Ideally the velocity contours should follow the seabed topography up to certain depth below the water bottom. Figure 4: Terms & definitions used for canyon scaling 5

6 Figure 5: Plots of raw scalar vs. hdiff for different canyon zones Figure 6: Depth Stack, Velocity Model and Depth Gathers before & after canyon scaling 6

7 Figure 7: Depth slice of initial velocity 1800m before & after canyon scaling 7