GStatSim interface

[mmaelicke]

@GatorGlaciology, this might be interesting for you.
This pull request introduces an preliminary interface to GStatSim. There are 3 main options for interfacing:

• generate a GStatSim compatible prediction_grid from a Variogram instance
• generate the input to a skrige_sgs or okrige_sgs simulation from a Variogram
• directly run skrige_sgs or okrige_sgs in parallel, from a Variogram instance

additionally, the interface defines a Grid class, which generates the prediction_grid but keeps track of the shape needed for the target field, to transform the result. @GatorGlaciology, maybe this class is also helpful for you? I could open a pull request to add it to GStatSim instead of SciKit-GStat, if you are interested.

interfaces

prediction grid

You can generate a prediction_grid like:

coords, vals = skg.data.pancake(N=300).get('sample')
vario = skg.Variogram(coords, vals, maxlag=0.6, n_lags=25)

# use like
grid = vario.gstatsim_prediction_grid(resolution=5)

# this will find rows, cols as close to 400x400 as possible to have the same resolution in both directions
grid = pred_grid = vario.gstatsim_prediction_grid(rows=400, cols=400)  
print(grid).   # prints <Grid with 401 rows and 405 cols at 1.235 resolution>

# then convert to GStatSim:
grid.prediction_grid

simulation parameters

The simulation params can be accessed via the interface

from skgstat.interfaces import gstatsim_mod

# you can pass a grid, or a resolution or a (rows, cols) tuple as grid
grid, cond_df, variogram_params = gstatsim_mod.simulation_params(vario, 5)

all in one

And finally use the simulation directly

fields = vario.simulate(grid, method='simple', n_jobs=8, size=8)

This creates a list of 8 simulated fields. If 8 cores are available, this is as fast as using GStatSim only (or n_jobs=1) as each simulation is run in parallel.

ToDos

@GatorGlaciology, I will add a bit of docs and tests. Meanwhile you are more than welcome to comment on this interface.
I also need to define, which Variogram instances actually can be exported to GStatSim.

[GatorGlaciology]

Very cool! This looks incredibly useful. Let us know if we can help in any way

[codecov]

Codecov Report

Attention: 22 lines in your changes are missing coverage. Please review.

Comparison is base (a59da91) 90.74% compared to head (4c9090a) 90.48%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #155      +/-   ##
==========================================
- Coverage   90.74%   90.48%   -0.26%     
==========================================
  Files          23       24       +1     
  Lines        2475     2639     +164     
==========================================
+ Hits         2246     2388     +142     
- Misses        229      251      +22     

Files	Coverage Δ
skgstat/Variogram.py	96.41% <100.00%> (+0.04%)	⬆️
skgstat/interfaces/gstatsim_mod.py	85.71% <85.71%> (ø)

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.