Benchmarking in subsurface hydrological inversion: high-fidelity reference solution and EnKF replication data

<p><strong>Description:</strong></p>
<p>Dataset published with the paper "Towards a community-wide effort for benchmarking in subsurface hydrological inversion: benchmarking cases, high-fidelity reference solutions, procedure and a first comparison". You can use these data to generate the comparisons between the EnKF and the MCMC solution as seen in the <a href="https://doi.org/10.5194/hess-2024-60">paper</a>.</p>
<p><strong>Folder structure & Nomenclature:</strong></p>
<ul>
<li>Each folder with <strong>reference</strong> data starts with "<strong>ref_</strong>", followed by the scenario identifier like for S0, "ref_S0". See the scenario description below for further explanation.
<ul>
<li>In each of the reference folders, the several MCMC chains are stored. Please refer to the publication for further information about the file.</li>
</ul>
</li>
<li>Each folder with solutions that are not references, but just <strong>replications</strong>, start with "<strong>rep_</strong>", like for the S0 steady state EnKF, "rep_EnKF_S01".
<ul>
<li>In these folders, the datafiles including the solutions are stored. </li>
<li>For your own references, we recommend creating your own folder called "rep_YourFolderName_ScenarioSpecifier".</li>
</ul>
</li>
</ul>
<p><strong>Scenario description:</strong></p>
<ul>
<li>S0 is the base case. It features a relatively strong degree of heterogeneity with σθ = 2, relatively accurate measurement data with σe = 0.05 [L], irregularly placed observations, and steady-state groundwater flow.</li>
<li>S1 features the regular grid of observations instead of the random one. While irregular monitoring 375 networks are more realistic, the very close spacing of a few monitoring wells may pose a problem to some methods due to their high autocorrelation. Therefore, S1 is a fallback scenario.</li>
<li>S2 is again like S0, but reduces the strength of heterogeneity from σθ = 2 to σθ = 1. While σθ = 2 is a more realistic degree of heterogeneity, it may already be challenging for methods that are explicitly or implicitly linearization-based. Therefore, S2 is a fallback scenario.</li>
<li>S3 is again like S0, but increases the assumed level of observational errors from σe = 0.05 [L] to σe =0.1 [L]. Given the overall head difference of 20 [L] across the domain by the boundary conditions, these values can be classified as high and medium accuracy, respectively. Iterative or sampling-based methods may have problems with the accuracy requirement posed by the large accuracy in S0. Once again, S3 is a fallback solution. However, as posterior uncertainties will remain larger for smaller measurement accuracy, S3 may also trigger stronger non-linearities across the larger remaining postcalibration uncertainty ranges.</li>
<li>S4 changes S0 to feature transient (instead of steady-state) groundwater flow. This is relevant for<br />EnKF-type methods that work via transient data assimilation and that do not iterate.</li>
</ul>
<p><strong>Download and benchmarking:</strong></p>
<a href="https://github.com/LS3-university-of-stuttgart/hydrological-inversion-benchmarking">Github Package</a>