Opened 3 months ago

#881 new defect

Unexplained jumps in total carbon biomass

Reported by: mmcgrath Owned by: somebody
Priority: major Milestone: ORCHIDEE 4.2
Component: Biogeochemical processes Version: trunc
Keywords: Cc:

Description

This may be related to multiple stable states in the total biomass noticed by Daniel.

During a global two-degree spinup (FG1, SVN config) with revision 7692 (TAG4.1), there are some unexplained jumps in the total biomass for PFT7. For example, around year 1960 for the pixel located at 35.0N, 63.0E. Unsure how often this happens, but at first glance it could concern 1% of pixels, primarily those with relatively low biomass (400-1000 gC per square meter). The attached plot shows the total carbon biomass (top), the number of individuals (and those recruited; middle), and the RDI (bottom).

Previously FM1 followed self-thinning. We decided to use self-thinning to mimic gap dynamics, which is why the number of individuals drops even though forest management is deactived. This is seen in the RDI plot.

The first segment of the biomass represents something real: big trees are dying faster than the recruits can take over. Trees are growing, we hit the upper RDI once in while, we remove big trees, we get a lot of light, we add small trees, biomass is slowly decreasing and will result in the end of the stand.

Somewhere around 1959 and 1960 something strange happens. As we are renewing the stand we get more and more but smaller and smaller individuals. The target RDIs are calculated as a function of diameter. So it looks that around 1959-1960 diameter dropped sufficiently to make the upper RDI decrease. Between 1960 and 1965 we need more details. The process is probably hidden in the aggregated outputs. The wiggles in RDI are suspicious because the seem not to reach the targets. RDI should simply increase. The steep decrease in RECRUITS_IND suggest that suddenly the canopy is closed (could be the outcome of adding many individuals).

We have no ideas of any intended process/hack/line of code that should result in the increase in biomass.

We will likely need daily output starting from around 1955 to diagnose the issue further.

It should be checked in the code whether we have a statement that truncates the total number of individuals. We have changed so much to get FM1 and FM2 in the same framework that such a line may have entered the code (but passed through my memory).


Attachments (1)

PFT7_problem_timeseries.png (122.4 KB) - added by mmcgrath 3 months ago.
An example of a problem pixel in FG1.r6792.OD for PFT7

Download all attachments as: .zip

Change History (1)

Changed 3 months ago by mmcgrath

An example of a problem pixel in FG1.r6792.OD for PFT7

Note: See TracTickets for help on using tickets.