Overhaul the subdaily disaggregation logic in initialize_atmos.c

[jhamman]

When we changed the default timestep units to seconds (#188), we introduced a number of bugs into the subdail disaggregation of forcing variables in initialize_atmos.c. Looking at the code, there is a lot of repeated logic that could be abstracted to individual functions. This will greatly reduce the likelihood of bugs coming from slightly different implementations with the same goal.

I'll take point on this issue and will split the work into three steps:

• Create functions for the various forms of subdaily disaggregation supported in initialize_atmos.
• Replace repeated sections of code with these functions
• Test, test, test.

xref #42 and #188

cc @tbohn: I may need a bit of guidance as I go through this.

[bartnijssen]

Can you point to some examples? The MTCLIM was wrapped with as few changes to that code as possible and I would like to keep it that way. Any VIC code we can of course change.

[tbohn]

I think Joe is referring to the logic in initialize_atmos.c for
transferring data from the MTCLIM data structures and the forcing data
structures into the atmos data structure.

Ted

On Thu, Aug 6, 2015 at 12:32 PM, Bart Nijssen notifications@github.com
wrote:

Can you point to some examples? The MTCLIM was wrapped with as few changes
to that code as possible and I would like to keep it that way. Any VIC code
we can of course change.

—
Reply to this email directly or view it on GitHub
#240 (comment).

[tbohn]

OK, as per @jhamman 's suggestion, I'll take a look at getting this subroutinized.

[jhamman]

Alright, I think I've fixed some of these issues shown below but will have to rerun this test when I get back to Seattle. A summary of what where I see the subdaily disaggregation (check these boxes when we are satisfied that things are working):

• Precipitation: looks right, need to plot subdaily subset
• Air Temp: looks right, need to plot subdaily subset
• Shortwave: Units problem, needs work
• Longwave: Units and indexing problem, needs work
• Density: broken
• Pressure: broken
• VP: broken
• Wind: looks right, need to plot subdaily subset

Figure: Comparison of input forcings for one grid cell (48.1875, -120.6875) in the Stehekin test dataset with the disaggregated hourly forcings for VIC 4.2.b and the develop branch.

Figure current as of January 26, 2015.

[tbohn]

Thanks @jhamman. Are you actively working on this? What kind of timeframe did you have in mind for this? I can get to this tomorrow, maybe even later today, but I do have some other things to work on that need to be my top priority today.

[jhamman]

Actually, I also have this plot for a shorter period:

Figure: Same data as above but for shorter time period and only showing VIC 4.2.b (left) and develop branch (right).

[jhamman]

@tbohn -

After talking with @bartnijssen, we have come up with another path forward here. The basic idea is to remove the MTCLIM and subdaily disaggregation entirely from the Classic Driver.

1. Remove MTCLIM
2. Replace initialize_atmos with vic_force, basically a function to pack the atmos_struct
3. Create a standalone forcing generator tool based on initialize_atmos that uses the VIC.4.2 C code to do the MTCLIM/disaggregation

This greatly simplifies the Classic Driver implementation but requires us to develop the stand alone tool in the near future.

@tbohn - how do you see this working out?

[tbohn]

That makes more sense to me. Will the stand-alone tool be based on your python tool? Or will it be in c?

Seems to me that the stand-alone tool will need netcdf i/o (in addition to classic ascii), so maybe python is the way to go?

[bartnijssen]

I think we should do the standalone tool in two steps:

1. The existing VIC code with all (most) VIC functionality stripped out. This tool would be able to reproduce exactly the hourly VIC forcings that we had in VIC 4.2. That way we are not making too many changes at once.
2. Switch to a python-based forcing tool, which may or may not have additional functionality.

Creating (1) would hopefully be straightforward. It would be what VIC produces with the OUTPUT_FORCE option. It'll require a few updates to handle the subhourly timestep.

[jhamman]

closed as wont fix since we removed sub-daily dis-aggregation all together in #288.