How to preserve the continuity in sebsequent GLM simulations

[Mustafa Onur Onen]

Hello everybody!

I am testing whether GLM can reproduce the same simulated water temperature for 2 different simulation setups with the same inputs as explained below. For both setups, daily inflows/outflows and hourly met variables are provided for 12 month period. I use version "glm-aed 3.3.1a2".

1st Simulation setup

In this setup, I run GLM 12 times (like in a moving monthly window) for a total of 12-month-timeframe. The moving monthly window starts from the first hour of the current month to the last hour of the same month.

Before executing each run, I transfer final conditions obtained (from nc) at the end of the previous simulation as the initial condition (in nml). The name of these conditions are given as they are included in nml file as follows.

&init_profiles (lake_depth, num_depths, the_depths, the_temps, the_sals, avg_surf_temp, restart_variables)

I transfer these information for all layers.

After completing 12 simulations, the results contained in different nc files are merged to obtain a comprehensive water temperature time series for the entire 12 month period.

2nd Simulation setup

I execute another simulation without stopping GLM from the beginning to the end of the 12-month period. Here, the aim is to investigate whether the continuity in the 1st simulation setup is preserved as well as it is in the 2nd simulation setup.

Results

I have noticed disparities between the results of these two setups (even though the flows and met variables are the same), with the discrepancies being more pronounced during the warm season. The figure below illustrate the absolute difference between water temperatures obtained by the abovementioned simulation setups. The simulation period starts at January 1 00:00 and ends at December 31 23:00. Vertical black lines represent the first timestep (hour) of the each monthly run which are used in the first setup. In this figure, I would expect to get differences starting with the black lines but it seems they start in the middle of the moving window.

Any ideas on this issue are greatly appreciated.

Best regards,

Mustafa

P.S. There is another conversation in this group addresing the same issue which didn't help much (link).