negative PRECL in CESM1 output?

[lesley_l_smith@noaa_gov]

Hi,
I'm not 100% sure this question is in the correct forum. If another forum would be better, please let me know.
We downloaded the
CESM1 CAM5 BGC 30-Member Large Ensemble Atmosphere Post Processed Data, Monthly Averages
data and I started looking at the precip. I'm seeing some PRECL values < 0.
I don't understand this.
Ideas?
Thanks for any insight,
-Lesley

 

[strandwg]

I have verified that there are negative values for PRECL; however, they are very very small - order 10^-22 to 10^-28.They also comprise a very small fraction of all points - the most is 104 out of 66,355,200 values, the fewest is 7 out of 7,299,072.I suspect that they are roundoff-level values.