Example Gallery¶
Worked examples of the INTEGRATE workflow, from a first inversion through to
full probabilistic raw-material assessment. Each example is a plain Python
script: use the download links at the bottom of a page to get it as a .py
file or as a Jupyter notebook.
Examples showing figures were executed when these pages were built. The remainder depend on data, hardware, or credentials that are not available at build time, and are shown as source only.
Getting Started¶
Start here if you are new to INTEGRATE. The first example runs the complete chain from data to posterior; the two variants show how to skip the forward modelling step, and how to work from an existing posterior.
Getting started with INTEGRATE - with no forward code
Getting started with INTEGRATE - posterior analysis only
The Complete Workflow¶
The complete INTEGRATE workflow in a single example, from prior construction through forward modelling and inversion to analysis of the posterior.
The complete INTEGRATE workflow: from data to posterior
Data¶
Joint inversion is supported by using multiple data types. The first example splits a single tTEM data set (with both low and high moment) into two separate data sets, which can then be inverted separately or jointly. Borehole data can also be used, either on its own or in combination with tTEM data.
Borehole data in INTEGRATE – representation and integration
Noise¶
INTEGRATE allows for using both correlated and uncorrelated Gaussian noise models.
Synthetic Case Study example with different noise models (uncorrelated/correlated)
Hypothesis Testing¶
Competing geological scenarios can be encoded as separate priors and compared against the same data, so that the data decide which scenario is supported.
Daugaard Case Study with three lithology-resistivity prior models.
Daugaard Case Study with geology-resistivity-category prior models.
Querying the Posterior¶
Ask questions of the posterior ensemble – probabilities that a condition holds, or percentiles of a derived quantity – at every survey location.
Synthetic Case¶
A fully synthetic case where the true model is known, so the posterior can be checked against the answer it is supposed to recover.
Plotting¶
Visualising results: profiles and cross-sections extracted from a posterior ensemble.
Raw Material Assessment¶
End-to-end probabilistic raw-material (sand/gravel) assessment, including volume estimates with uncertainty, and comparison against earlier deterministic assessments of the same areas.
Daugaard: probabilistic raw-material assessment with INTEGRATE
Sdr. Felding: probabilistic raw-material assessment with INTEGRATE
Other Examples¶
Smaller, focused examples covering individual pieces of the workflow: prior construction, merging priors and data sets, parameterisation choices, and timing.
Merging multiple data files from the same survey (ESBJERG)
INTEGRATE - Demonstration of merge_prior() function
INTEGRATE -Generic Prior Model Generation Examples
