Transient PICO Assignment

lawrence-bird

Hi folks:

I have a quick question about using transient PICO models for basal forcings.

I see that there is a requirement for md.basalforcings.farocean_salinity to have all values > 0. This seems to include the time row. I have included 2 examples here:

Example 1:

md.basalforcings.farocean_salinity = [34.1, 34.2, 34.4; 0, 1, 2];

This fails, with a warning:

model not consistent: field 'basalforcings.farocean_salinity' should have values above 0

Example 2:

md.basalforcings.farocean_salinity = [34.1, 34.2, 34.4; 0.083, 1, 2];

This runs happily.

My question is: In Example 2 here, what salinity value is used for the initial timestep (t = 0)? In addition, if I only define one salinity value (to be held constant in time), does the time value matter, or is it simply ignored by ISSM?

md.basalforcings.farocean_salinity = [34.0; 0.083];

I'm not sure if there is a particular reason that ISSM requires the timestep for salinity values to be > 0 (this is not the case for Temperature values), but I just want to make sure the behaviour here doesn't have too much of an impact.

Thanks.

Lawrence

mathieumorlighem

Hi @lawrence-bird

this is a bug. I looked at it yesterday and fixing it is not as easy as I would have liked (basically the rule '>0' is applied to the whole matrix instead of the first rows). I will try to fix it later this week. In the meantime, you can just deactivate the consistency check:

md =solve(...., 'checkconsistency','no');

Cheers
Mathieu

lawrence-bird

Hi mathieumorlighem

Thanks for this update.

Can you confirm the behaviour/impact of this bug? In this case, when the first time specified is 0.083, what would ISSM use for t = 0? I am running with a monthly timestep (0.083). When specifying constant values, such as:

md.basalforcings.farocean_salinity = [34.73; 0.083];
md.basalforcings.farocean_temperature = [272.3; 0.083];

A melt field is returned for each timestep in md.results, as expected, with the first results available at timestep 0.083. Would the melt rate for time = 0 simply be 0? When specifying fields like this:

md.basalforcings.farocean_salinity = [34.73; 0.083];
md.basalforcings.farocean_temperature = [272.3; 0];

A melt field is also returned for each timestep in md.results, as expected, with the first results available at timestep 0.083. Would ISSM use a salinity of 0 for t = 0 and the specified temperature?

I've got some runs that use both of these setups and I'm wondering if I'll need to re-run them once the bug is fixed, or if this bug actually has no/negligible impact on the way basal melt is applied.

Thanks.

Lawrence

mathieumorlighem

This bug has no impact on the simulation, it is just the check that is overzealous. ISSM uses a linear interpolation between the times that are prescribed by the field. If the time step falls before the first provided time, then it will use the first column. Note that ISSM will not calculate the model state at time = md.timestepping.start_time, these are provided by initial conditions.
I hope this makes sense!
Mathieu

lawrence-bird

Hi mathieumorlighem

Thanks for this. To clarify, am I correct to think then that the following definitions would return identical results, and basal melt would start being applied at the very first timestep (because nothing would be used at md.timestepping.start_time)?

md.timestepping.start_time = 0;
md.timestepping.time_step = 1/12;
md.timestepping.final_time = 100;

Example 1:

md.basalforcings.farocean_temperature = [272.30; 0.083]
md.basalforcings.farocean_salinity = [34.73; 0.083]

Example 2:

md.basalforcings.farocean_temperature = [272.30; 0]
md.basalforcings.farocean_salinity = [34.73; 0.083]

Cheers,

Lawrence

mathieumorlighem

yes, because the first relevant time in the calculation with be 0+1/12 = 0.083.