A bug report on bare soil albedo specification in the newly released 4.2.0

[pengbinpeluo]

The newly released VIC 4.2.0 would output a negative biased surface albedo when bare soil type exists(total vegetation cover fraction < 100%) in the model cell. There probably is a bug on the albedo specification for different land cover types in the subroutine full_energy.c. If BARE_SOIL_ALBEDO is not prescribed in the following format, another column for bare soil albedo in the forcing file can be added to fix this issue.

The code goes at about line 313 in full_energy.c like:

      /** Compute Bare (free of snow) Albedo **/
      /* modified by Bin Peng to fix the bug */
      if (iveg!=Nveg)
        bare_albedo = veg_var[iveg][0].albedo;
      else
        bare_albedo = BARE_SOIL_ALBEDO;

Contact me(Bin Peng) at pengbin@radi.ac.cn or 112233pengbin@163.com for any further information about this bug report.

[jhamman]

@pengbinpeluo, thank you for the bug report.

Briefly looking at the code in full_energy.c and in the initialization routines of the model, I think this indeed is a bug where bare_albedo = 0.0 for the iveg==Nveg case. In fact, BARE_SOIL_ALBEDO, is not used current code.

Rather than changing the code in full_energy.c as proposed by @pengbinpeluo, I'd suggest correcting the initialization of the bare soil vegetation tile.

As a side note, it looks like this bug was introduced in VIC.4.1.1 in this change:

[jhamman]

I just pulled this up in the debugger and was able to confirm that the bare_albedo for the bare soil tile remains zero throughout the model run. This should be an easy fix and one I'll pull together today.

I will issue a PR for to the hotfix/4.2.1 branch and will manually port to the VIC.5 develop branch.

[bartnijssen]

Is this specific to 4.2. or did this basically always happen (even in older versions of VIC).

On Dec 19, 2014, at 11:13 AM, Joe Hamman notifications@github.com wrote:

I just pulled this up in the debugger and was able to confirm that the bare_albedo for the bare soil tile remains zero throughout the model run. This should be an easy fix and one I'll pull together today.

I will issue a PR for to the hotfix/4.2.1 branch and will manually port to the VIC.5 develop branch.

—
Reply to this email directly or view it on GitHub.

[jhamman]

It looks like this has been happening since VIC.4.1.1.

[tbohn]

Thanks for fixing this, Joe. This is most likely my fault - sorry about
that!
Cheers,
Ted

On Fri, Dec 19, 2014 at 11:54 AM, Joe Hamman notifications@github.com
wrote:

It looks like this has been happening since VIC.4.1.1.

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

[jhamman]

closed via #181.

[pengbinpeluo]

Thank you all for this bug fixing. Cheers.