RokDoc 2026.3 delivers a set of additions to the reservoir characterisation workflow that span well-tie QC, inversion environment improvements, frequency-domain analysis, 4D time shift processing, probabilistic classification, and project navigation. The common thread across these changes is precision - tools that let you observe more detail, iterate more efficiently, and compare results more directly within the same session. Here is what is new and how to work with it.
Inversion Multi-Filters in Ji-Fi (requires Ji-Fi or Advanced Ji-Fi License) allows different bandpass filters to be applied and compared within a single Ji-Fi invert-at-well session. Filters are defined directly in the session - each with its own low-cut and high-cut start and end frequencies - and inverted around the wells together, so the results appear side by side rather than requiring separate sessions to compare. Previously, comparing inversion results at different frequency configurations meant running each configuration in isolation, which made direct comparison impractical.
This is most relevant when you are examining the sensitivity of the inversion result to the low-frequency model. By running multiple filter configurations within the same session, you can directly observe how much of the impedance structure at a given depth is driven by the low-frequency model and how much by the seismic data. A useful starting point is to run at least three configurations: low-frequency model only, seismic band only, and the combined result. Where the three outputs agree, the inversion is well constrained by the data; where they diverge, the result is leaning on the model. This distinction matters most in areas where the seismic signal-to-noise ratio is lower or where the low-frequency model is less well constrained by well control — exactly the areas where the difference between model-driven and data-driven results is most consequential for interpretation.
In Coloured Inversion (Reservoir Characterisation License, requires Seismic Inversion module), the relative track now displays the band-pass filtered absolute inverted log. This is a change to what is shown in the well track at the well location: previously the relative result was shown; now the absolute log filtered to the inversion band is shown instead. This makes the well track more directly comparable to what the inversion is producing in the seismic-derived volume.
The practical implication is that calibration and QC at the well is now done against the same quantity being examined in the seismic volume. The bandpass-filtered absolute impedance, rather than the relative quantity. This is most useful during the QC step before you finalise the inversion model, where verifying that the inverted result at the well is consistent with the bandpass-filtered log gives you greater confidence that the inversion is performing correctly in the calibrated zone. A useful starting point is to compare the new track display to the log-derived absolute impedance bandpass-filtered to the same frequency range and check for consistency across the full well. Any systematic discrepancy at this stage is a signal worth investigating before you commit to the final inversion output.
Two improvements to 4D time shift workflows are included in this release, both under the Advanced SDC License, and both developed in direct response to requests from the services team.
Time Shift Estimation now runs approximately 2x faster than the previous implementation. For iterative workflows where the estimation is run multiple times to refine parameters or test different configurations, the cycle time for each iteration is directly halved. A useful starting point is to put that speed toward a parameter sensitivity test that may previously have been impractical on production-sized data - varying the estimation window or the reference trace configuration across several runs to verify that the chosen parameters are stable before committing to the full dataset.
The SDC Seismic Shift recipe is now fully supported for export via the MPI chain, covering individual time shift quantities across the full chain. This completes MPI coverage of the Advanced SDC conditioning workflow: estimate, filter, and apply time shifts can now be calibrated as a recipe and exported in a single operation, alongside the other conditioning steps, rather than requiring manual export for some components. The full-chain coverage means time shift results can be exported at scale through the same MPI workflow used for every other SDC quantity, without exception. A practical approach is to configure the time shift estimation on a pilot area first, verify output quality against a known well, then deploy the MPI chain export for the full survey. The faster estimation makes the pilot step a lower time commitment, which in practice means it is more likely to be done carefully before the full export is committed.
A Spectral Attribute type is now available in the attribute framework, displaying the per-trace frequency spectrum of a seismic quantity between two user-defined frequency bounds. It is created and managed in the same way as other attribute types and is available across the Rock Physics, Reservoir Characterisation, and Attrimod licenses.
From a reservoir characterisation perspective, the Spectral Attribute is most useful for identifying zones where the frequency character of the seismic differs systematically from the surrounding area - whether that reflects a processing artefact, an acquisition footprint, or a genuine geological signal. A useful starting point is to display the Spectral Attribute at your target level and examine whether the frequency character is spatially consistent. Where it varies in a way that correlates with structure or amplitude anomalies, the spectral character adds a dimension of information to the interpretation that can help distinguish tuning effects from genuine impedance variation. Where it varies without apparent geological cause, it may point to a processing or acquisition issue worth addressing before interpretation proceeds.
Bayesian Classification (available in both Rock Physics and Reservoir Characterisation Licenses) is now substantially faster in both RD2D and RD3D: approximately 2x faster for PDF exports and 2.5x faster for DTA exports, with no change to the underlying outputs. From a reservoir characterisation perspective, this matters most for workflows where Bayesian classification generates probability volumes for lithofacies or fluid discrimination at the seismic survey scale. The speed improvement makes it more practical to run multiple scenarios: varying priors, modifying training data, or comparing classification results at different seismic conditioning states, all within a single working session. If your workflow involves running the full classification on each iteration of the inversion output, the improved speed directly reduces the elapsed time between inversion and classification QC.
Two Platform-license updates make large reservoir characterisation projects easier to navigate and present consistently.
Volume filtering is now available in 3D sessions. Volumes can be filtered by log type or by string search, so locating a specific quantity in a project with hundreds of volumes - a particular impedance result, a single PCP angle stack, a named facies volume - becomes a quick navigation step rather than a scroll through the full tree. For RC projects that accumulate many derived volumes across inversion, classification, and conditioning passes, this reduces the cognitive overhead of working in a large session and makes it easier to bring exactly the volumes you want into a display.
A directory of images can also now be designated as a custom lithology pattern library, loaded automatically on startup and available alongside the default patterns in the pattern picker. For teams that work to a company or partner standard, this means lithology and facies patterns can be applied consistently across users and projects, rather than each interpreter reaching for whatever default is nearest. The patterns carry through to well sections and facies displays, so classification and interpretation results are presented in a shared visual language from the outset.
These additions reflect a consistent direction: improving the precision and efficiency of the workflows that reservoir characterisers run most often, particularly well-tie calibration, inversion sensitivity analysis, probabilistic classification, and now the navigation and presentation of large survey-scale projects. If there are specific workflows you would like us to cover in a dedicated how-to or deep-dive - Ji-Fi multi-filter configuration, Bayesian Classification prior design for reservoir targets, or time shift QC workflows - let us know in the comments or reach out directly. Your suggestions are what shape the technical content of future posts.