Forward simulation uncertainty range

[cnicol-gwlogic]

Hi there,

Firstly - thanks very much for a very nice piece of software. Much appreciated.

My query: I might be missing something dead obvious (sorry) , but is there a method for plotting simulation forecast uncertainties (based on the history match)? I can't see how to obtain this for periods beyond the observation record, other than using the emceehammer (ha! which I have done, but would like to keep things simpler if I can).

Many thanks.
Chris.

[dbrakenhoff]

Hi Chris,

I'm not entirely sure if this is what you're looking for but you should be able to pass tmin and tmax to the calculation of the prediction interval. This is the uncertainty as a result of estimated parameter uncertainties and the standard deviation of the residuals (for the calibration period).

For example:

ml  # some model
ml.solve(tmax="2015")
pi = ml.fit.prediction_interval(tmin="2015", tmax="2022")

Maybe that helps?

Davíd

[cnicol-gwlogic]

Brilliant @dbrakenhoff - thanks a lot, that works perfectly (apols, I should have noticed that).

[nelerey]

Maybe it'd be worth emphasizing in the documentation and/or uncertainty examle that tmin and tmax can be passed to pastas.solver.BaseSolver.prediction_interval() in the **kwargs?
(suggesting this because I just found this closed discussion after forgetting how to do this and not finding the answer in the docs.)

[dbrakenhoff]

Done! See #785.