Inspect the benchmark data
==========================

Goal
----

Understand the benchmark-format files created from the OPM Flow result before
interpreting the comparison figures.

Configuration
-------------

This lesson uses the output generated in :doc:`run-simulation`. No configuration
changes are required.

Command
-------

List the generated CSV files:

.. code-block:: console

   find YOURS -name "*.csv" -maxdepth 3 -print

How it works
------------

The main benchmark products are:

``time_series.csv``
   Uniformly sampled pressure, mobile, immobile, dissolved, seal, and migration
   quantities used by the time-series comparisons.

``spatial_map_Xh.csv``
   Saturation and dissolved-CO2 concentration mapped to the FluidFlower
   reporting grid at time ``X`` hours.

``sparse_data.csv``
   Sparse benchmark quantities derived from the time series, including maxima,
   event times, and values evaluated at 72 hours.

``sim_metrics_0.txt`` and ``func``
   Per-observation Wasserstein metrics and the combined objective used by ERT or
   Everest job workflows when those files are generated.

Result
------

Confirm that the requested spatial maps exist for 24, 48, 72, 96, and 120 hours.
The CSV files may also be generated by another simulator, provided that they
follow the same benchmark layouts and coordinate ranges.

.. note::

   You can generate ``sparse_data.csv``, ``time_series.csv``, and
   ``spatial_map_Xh.csv`` files with another simulator and use pofff to compare
   those results with the experimental data, forecast, CSIRO, MIT_M1, and CSSR
   data.

   Place the files in a dedicated output folder, such as ``MY_SIMULATOR``, and
   run:

   .. code-block:: console

      pofff -t 24,48,72,96,120 -m none -o MY_SIMULATOR -f all

   The spatial-map CSV files must cover the FluidFlower reporting domain from
   0 to 2.8 m in the x direction and from 0 to 1.2 m in the z direction. The
   files must otherwise follow the benchmark layouts expected by pofff.

Next
----

Continue with :doc:`generate-comparisons`.