Thank you to the 80+ participants who joined us live on this webinar for the latest release of Supervisor! We were excited to bring to you an update on the latest release along with some background to the design principles guiding our Supervisor development.
Unfortunately, we were unable to access the recording of the webinar, so we’re providing you with some alternative resources.
You can also access the slide deck here.
A feature demo for 8.13 is available here: View YouTube Video
Questions asked during the live Q&A included:
Q: Do you want to apply multivariate simulation using decorrelation workflow such as MAF / PPMT or do you want to rely on cross variograms and LMC cosimulation? Multivariate simulation in Supervisor will be implemented using a decorrelation workflow with several options available including MAF, PPMT and other decorrelation approaches that are being developed
Q: A couple of things I couldn’t locate in the latest release when compared to S7 – There was nowhere to assess the validity of the simulation (gaussian mean and variance). Supervisor back-transforms the simulated results on the fly and the results of the Gaussian simulations are not saved. This has the disadvantage of not being able to check the simulations in Gaussian space as pointed out but contributes to reduce the amount of storage required to run large conditional simulations.
Checking the simulations in Gaussian space is of prime importance particularly for multivariate simulations using a decorrelation workflow. This option will be available for both univariate and multivariate simulations in the next major Supervisor release.
Q: I couldn’t find where we can apply confidence limits to the simulations or re-blocked estimates? The option is not yet available in the current version of Supervisor, but is in development for an upcoming release.
Q: What tool is available to validate the sims beside the QQ plot? (i.e. grid variogram) Standard tools to validate the conditional simulation results are available in Supervisor. While a specific component is not currently available to ease the process, the variograms can be manually copied directly to the simulation results and verified.
Q: Are you going to implement SGS to run in parallel to save time? Supervisor implementation of SGS is currently optimised to utilise parallel computing to reduce the run-time of the technique.
Q: I really like the grade-tonnage curve with the envelope. Would be great to be able to overlay your block model and see where it sits relative to the envelope of simulation grade-tonnage curves. It is already possible to do this in the current version of Supervisor. One only needs to load a model and overlay the grade tonnage curve in the simulation component.
Q: Do you have kriging implemented in Supervisor software? Kriging is not directly available as a stand-alone option in Supervisor. Nonetheless, it can be accessed and used via the Local Kriging Neighbourhood Optimisation component in Supervisor (LKNO).
LKNO is an alternative approach to optimising estimation parameters that is based on Kriging estimates obtained with different Kriging estimation plans. By considering only one set of user-defined estimation parameters it is possible to get a Kriging estimate in Supervisor.
Q: When I exported a set of reblocked simulations, the CSV contained a ‘Probability’ column. I’m not sure what grade cut-off this field related to? This column refers to the proportion of points inside the block used to get the reblocked simulation value.
Q: Is there any plan to implement direct block simulations in supervisor in the new future? The development team is focused in providing users with a friendly tool to run conditional simulations both univariate and multivariate at the moment. Implementation of direct block simulations is being considered.
Q: Are there any plans to allow the user to overlay the drillholes and simulations together in 3D? The current 3D visualisation implementation in Supervisor does allow the user to create 3D viewers that can display multiple loaded datasets such as point and block data; however, the 3D visualisation currently offered by Supervisor is under consideration for improvement.
Q: How do you want to check the reproduction of multivariate correlations in your future multivariate simulation? Standard validation plots are being developed and implemented in Supervisor to allow the user to check the reproduction of multivariate correlations when using multivariate simulation.
Q: I have not used the SGS in Supervisor, how they get the simulation back to real data? Back-transformation of the SGS results is carried out via the anamorphosis function. As mentioned before, the results are back-transformed on the fly and are not currently saved.
Q: Are you using simple kriging in your SGS? And are you using static mean or dynamic mean for your simple Kriging? The Kriging method by default for SGS is simple Kriging. Ordinary Kriging is available but no static or dynamic mean.
With these exciting new developments in Supervisor 8.13, we’d love to invite you to sign up for a license or to download a trial if you’d like to explore the software first. Got more questions? Email us here.