Mf6 Budget output: How to save cummulative budget for large number of simulation steps ? Issue with memory and saverecord parameter.

[valDallalba]

Hello everyone,

I'm currently working with a modflow mf6 model, using the flopy package (version 3.3.6) and I'm struggling to work efficiently with the budget outputs fp.mf6.ModflowGwfoc package and the saverecord option.

I'm running a daily 20 years model, that is composed of 2 period, the first one for initializing the model in steady state and the second one for running the transient simulation. The first period is composed of 1 simulation step and the second period of more than 7000 simulation steps (one for each day). Once the model is done running, my aim is to calculate zonal budget between different zones (using the ZONEBUDGET exe of the USG) and this for different time steps (monthly will be great for example).

The first issue that I have, is that I want to calculate cumulative volume exchange between different zones at different time steps (let's say for every month of my model or for every 30 days/time steps), since the budget outputs are saved as snapshot of the define time step to be saved, I need to save all the budget for every step between the monthly exchange that I'm interested and then calculate the cumulative exchange. However, saving all the steps between period of interested is too demanding in terms of memory for my specific model (366 budget steps in binary format are corresponding to a 45 GO .cbc file for example), and while I could reduce the number of steps being saved and "interpolate" the volume exchanged between the missing one, this cost me important imprecision when calculating the final cumulative budget over long period of time.

Moreover, assuming that I would be able to correctly interpolate the volume exchange between time steps, the second issue that I have encountered is that when I'm manually specifying the number of time steps to be saved (["BUDGET", "STEPS", 1, 5, 10, 15...]) at one point the model stop saving them (around when the .cbc. file is around 10GO, corresponding to 74 saved steps). I do not think that this is due to a memory issue because when specifying the saverecord as ["BUDGET", "ALL"] the .cbc file is 45 GO and is able to save 366 steps.

Here is a part of my code where I initialize the ModflowGwfoc package:

#define which heads and budget to save
save_head   = ["HEADS", "STEPS", 1, 5, 10, 15...]
save_budget = ["BUDGET", "STEPS", 1, 3, 6, 9, ...]

#create dictionnary record for each model period
print_rec = {0:[["HEAD", "LAST"], ["BUDGET", "LAST"]],
                1:[["HEAD", "LAST"], ["BUDGET", "ALL"]]}
save_rec = {0:[["HEAD", "LAST"], ["BUDGET", "LAST"]],
                1:[save_head, save_budget]}

#define oc package
oc   = fp.mf6.ModflowGwfoc(gwf,budget_filerecord='daily.cbc', head_filerecord='daily.hds', saverecord=save_rec, printrecord=print_rec)

I was wondering if :

• I was parametrizing something wrong with the fp.mf6.ModflowGwfoc package or the saverecord option ? Is there a way to reduce the size of it even if the output format is already a binary file?

• Is it possible in modflow and flopy to directly save the cumulative budget instead of the step budget, reducing the number of steps that would require to be saved ?

• If anyone knows a workaround to the saverecord file issues, with for example a way to manage to save the budget of each X number of steps in different budget files?

Thank you for your help and comments ! I'm happy to precise any point that is not clear :)

Best,

Valentin