How to calculate the δ18O of soil water in the water isotope-enabled LME simulations

[forestry2009]

I have found several soil-related variables in the link Last Millennium Ensemble Project Data Sets

For example,

SOILLIQICE_10CM_H218O, SOILLIQICE_10CM_H2OTR

H2OSOI_H218O, H2OSOI_H2OTR

QSOIL_H218O, QSOIL_H2OTR

If I want to calculate the SOILLIQICE_10CM, the formula is as follows:

δ18O of SOILLIQICE_10CM = (SOILLIQICE_10CM_H218O/ SOILLIQICE_10CM_H2OTR-1) * 1000.

Is this correct or not?

Thanks a lot

 

[nusbaume]

Hi Ru Huang,

That formula should work for calculating the δ18O of soil water. The only thing you should watch out for is grid cells where the bulk water variable (in this case "SOILLIQICE_10CM_H2OTR") is very small, which can cause numerical issues and thus un-realistic δ18O values. To prevent this, I would recommend setting a threshold for bulk soil water, say 10^-10, and only look at grid cells above this threshold.

If for some reason you are still unable to produce reasonable-looking soil water δ18O from the LME runs please let me know.

Thanks, and good luck!

Jesse

 

[forestry2009]

Dear Jesse,

Thanks so much for your help.

Based on your valuable suggestions, I get some prelimary results. Then I will conduct further analysis. I will report the results to you later on.

Best wishes

Ru Huang

 

[Nitesh]

Hey Jesse,

Still, I am getting a range of values between 1200 to -400 permil for d18O but dD looks reasonable. What could be the reason?

 

[nusbaume]

Hi Nitesh,

Apologies for the month-late response! Did you apply the equation and bulk water check described above? If so, can you attach a plot of the soil d18O that shows the bad d18O values, along with the plotting script you used to generate the plot? Any scripting language should be fine. I should also note that if doing a time-average to make sure that the values are mass-weighted by the soil water, or else spurious, low-mass delta-values can artificially generate these bad numbers (especially for 18O).

Finally, if the bad values are only in a few sparse grid-cells, then as a temporary fix you can just ignore any d18O values above and below a certain threshold (for positive values I usually have a cut-off at +50). Although this won't help determine the root cause of the problem, it could allow you to continue your analyses while maintaining the general spatial and temporal soil water d18O patterns. There also should be no concern that these values will propagate within the model (if everything is working correctly), as there are multiple mass checks throughout the model code which should remove the errors if they get too large.

Anyways, I hope that helps, and looking forward to seeing the error example!

Thanks, and have a great day!

Jesse