LithoFluid Probability

Description Calculate lithology and fluid probabilities and perform classification using statistical rock physics models.
Module(s) Lithofluids
Requirements Lithofluid model, Volume, Velocity Model
Related Using Trends, Using Predict
Works with 2D, Stacks

The LithoFluid Probability process uses Bayesian prediction to calculate probabilities and perform classification using statistical rock physics models. Two volumes are required with content matching the data in the statistical model (e.g. Acoustic Impedance and Vp/Vs, mu*Rho and lambda*Rho).

The process can be configured to generate a single "maximum likelihood" classification volume, or multiple probability volumes (one for each of the lithofluid combinations).

Each Lithofluid Probability process can also be used in Predict to calculate probabilities and classify well curves, providing valuable QC and calibration.

Create a LithoFluid Probability process

Create a LithoFluid Probability process
  1. In the Control Panel, open the Process tab.
  2. At the tab header, click the Add icon and select New Process.
  3. Double-click LithoFluid Probability.
  4. Type a new name for the process and click OK. Create a Prior Model
  1. In the Details Panel, select the LithoFluid PDF model at PDF Model from the searchable drop down list (see Importing a LithoFluid PDF).
  2. At PDF Attributes, choose which pair of attributes (crossplot domain) to classify on, e.g. AI_vs_VpVs, AI_vs_SI, or lambdaRho_vs_muRho. The options available depend on the PDFs defined in the selected PDF Model, and this choice determines which two volumes you assign further down the panel.
  3. At PDF Interpolation, choose the interpolation method used to interpolate the rock physics PDFs to the volume sampling.
    • Interpolate PDFs — Insight will interpolate between the available points to smoothly map to the volume sampling.
    • Use Nearest — With finely sampled PDFs, you can choose to explicitly use the nearest ellipse set at each depth.
  4. Select a prior model at Priors (see Create a Prior Model). Prior probabilities can be constants, or vary laterally when defined from horizon maps.
  5. At Water Bottom, select whether to use a Constant or a Horizon to define the water bottom. If a horizon is chosen, you can optionally input a constant value in the adjacent box to offset the horizon.
  6. Assign a volume to each attribute chosen at PDF Attributes. Each box's label updates to match your selection, for example AI Volume/VpVs Volume for AI_vs_VpVs, AI Volume/SI Volume for AI_vs_SI, or lambdaRho Volume/muRho Volume for lambdaRho_vs_muRho.

Warning: If the selected prior model uses horizon maps to set lithofluid ratios, use maps on a 1 IL x 1 CL grid. Coarser priors maps can imprint vertical striping in the output probability volumes (see Create a Prior Model).

Configure the output

  1. At Unknown # stddev, specify the range from the nearest PDF beyond which a point may become Unknown. For example, if the ellipses display 2 standard deviations, and the Unknown # stddev is set as 2, points outside the ellipses will become unknown. We recommend using at least 3 std dev for unknown, so the modelled PDFs will capture almost 100% of the observed data.
  2. At Output Volumes, choose whether to generate the Most Likely (LID), Most Likely (LID and Probability), or All Volumes.
  3. At Output Volume Prefix, select User Defined to give the output volumes its own prefix by selecting User Defined, or select Process Name to allow Insight to name the volumes using the name of the process.

Note: Only All Volumes produces a separate probability volume for each lithology/lithofluid. The other two options each produce a single combined volume, not one per lithofluid: Most Likely (LID) gives the winning lithofluid's class index only, and Most Likely (LID and Probability) (the default) encodes both the winning class index and its probability in one value per sample.

Set the velocity model

  1. At Velocity, select the volume providing the velocity model.
  2. At Velocity Type, choose the type describing the selected Velocity volume. The choices depend on whether the volume is sampled in time or depth: Interval Velocity in Time, RMS Velocity, Average Velocity in Time, or Depth values sampled in Time (ZT) for a volume in time; Interval Velocity in Depth, Average Velocity converted to Depth, or Time values sampled in Depth (TZ) for a volume in depth.
  3. At Velocity Units, choose the matching units. The choices depend on the Velocity Type selected: velocity units (m/s, km/s, ft/s) for the interval/RMS/average velocity types, time units (ms, s) for the TZ type, or distance units (m, km, ft) for the ZT type.

The velocity model is used to convert the Water Bottom horizon, the two input volumes, and the LithoFluid Model onto a common vertical domain, when they are not already all in time or all in depth.

Note: If the process is configured on its own, outside of a Workflow, a message reads "Can use session T/D conversion when not in workflow". In that case, the project session's default velocity model is used if Velocity is left unset.

As a result of this process, new volumes of probabilities will be added in the Volume tab for each litho-fluid, for example: