Opened 5 months ago

Last modified 5 months ago

#2831 accepted project-task

enabling irrigation in FLUXNET/JULES suite u-al752

Reported by: pmcguire Owned by: pmcguire
Component: JULES Keywords: JULES, FLUXNET, irrigation,Rose/Cylc
Cc: mtodt, kwilliams Platform: JASMIN
UM Version:

Description

Can you help with enabling irrigation in the FLUXNET/JULES suite u-al752?

Change History (9)

comment:1 Changed 5 months ago by pmcguire

  • Status changed from new to accepted

comment:2 Changed 5 months ago by pmcguire

Hi Patrick

I've tried to run u-al752 for Mead as Karina said, but JULES fails because it can't find a forcing file:
/group_workspaces/jasmin2/jules/pmcguire/fluxnet/kwilliam/suite_data/vn1.0/prescribed_sthuf/prescribed_sthuf_US-Ne1.txt

That file exists in
/group_workspaces/jasmin2/jules/pmcguire/fluxnet/kwilliam/data_for_suite_u-al752/vn0.5/old/peg/vn0.3/prescribed_sthuf/

but the simulation uses forcing data from vn1.0:
/group_workspaces/jasmin2/jules/pmcguire/fluxnet/kwilliam/suite_data/vn1.0/fluxnet/US_Ne1-met.dat

Can I just copy prescribed_sthuf/ from data_for_suite_u-al752/vn0.5/old/peg/vn0.3/ to suite_data/vn1.0/ ?

Cheers
Markus

comment:3 Changed 5 months ago by pmcguire

Hi Markus
The advice that we got from Anna Harper in Jan 2018 is that "You don't need the prescribed soil moisture or LAI, we are still improving this data before making it widely available. As long as you run with PRESCRIBE_DATASETS=0 (in the rose-suite.conf file) it won't look for those two data files."
So, if you can run with PRESCRIBE_DATASETS=0 instead of PRESCRIBE_DATASETS=2, then maybe that is the best solution for now.
Patrick

comment:4 Changed 5 months ago by pmcguire

  • Cc mtodt added; m.todt@… removed

comment:5 Changed 5 months ago by mtodt

That worked indeed.

However, I encountered issues when enabling irrigation for simulations of Chinese FLUXNET sites (for the PORCELAIN project).
1) Irrigation has an impact on all Chinese FLUXNET sites even if ONLY grasses are irrigated, although three of these sites are forests.
2) Simulations fail for three of the Chinese grassland sites (CN_Du3, CN_Ham, CN_Sw2) during spin-up because fsmc ⇐ 0 in physiol.F90. That is quite strange considering irrigation is prescribed as unlimited, which means tiles are "irrigated to critical point from an unconstrained water supply".

comment:6 Changed 5 months ago by mtodt

Regarding 1):

CN-Cng, CN-Dan, CN-Du2 - These are sites categorised as grassland and assigned tile fractions of 0.7 for C3 grass, 0.15 for C4 grass, and 0.15 for soil via /ancil/tilefracs/GRA_fracs9.dat. There are no discernible differences in SMC between irrigating all tiles and irrigating only grass tiles, which makes sense.

CN-Ha2 - This site is similar to the previous three apart from its classification as wetland, which means tile fractions are 0.8 C3 grass and 0.2 lakes.

CN-Cha - This site is categorised as a mixed forest and /ancil/tilefracs/MF_fracs9.dat assigns tile fractions of 0.35 for broadleaf deciduous trees, 0.35 for needleleaf evergreen trees, 0.2 for C3 grasses, and 0.1 for soil. That leads to an affect on SMC depending on whether only grasses or all tiles are irrigated, and those simulations indeed differ. However, SMC is higher when only grasses are irrigated.

CN-Din, CN-Qia - These sites are categorised as broadleaf evergreen forests and needleleaf evergreen forests, respectively, and tile fractions of 1 for tropical BETs and NETs, respectively, are assigned based on /ancil/tilefracs/BET-Tr_fracs9.dat and /ancil/tilefracs/ENF_fracs9.dat, respectively. There shouldn't be any impact when irrigating grass tiles but not only is there an impact on SMC, SMC is also higher when only grass tiles are irrigated compared to when all tiles are irrigated.

This is fairly confusing.

Activating irrigation for different tiles doesn't seem to be the problem, though, because error logs show that those three sites failing do so for tile 1 (trees) when all tiles are irrigated and for tile 6 (C3 grass) when only grasses are irrigated.

comment:7 Changed 5 months ago by pmcguire

Hi Markus

That is indeed confusing that (i) three of the six Chinese grassland sites fail with fsmc < 0 in physiol.F90, despite 'unlimited' irrigation, and (ii) irrigating the 2 Chinese forest-sites only in their zero'd grassland tiles has an effect.
If you want to talk about it, just let me know.

Patrick

comment:8 Changed 5 months ago by mtodt

Regarding 2)

As a test, I've run the subset 'all' with irrigation activated for all tiles. Out of 72 sites (one was still queuing) the following 13 failed:
CN-Du3, CN-HaM, CN-Sw2 — All three are grasslands, HaM is labelled "alpine Tibet".
DK-Nuf, DK-Zaf — Permanent wetlands ("low Arctic tundra fen" and "high Arctic tundra fen").
NO-Adv — Permanent wetland.
RU-Che, RU-SkP — Permanent wetland and deciduous needleleaf forest.
SE-St1 — Permanent wetland. ("Between shallow lake and fen-type wetland.")
US-Atq, US-Ivo, US-PFa, US-UMB — Permanent wetlands, mixed forest, and deciduous broadleaf forest.

Failing sites are either Arctic wetlands, boreal/temperate forests (RU-Skp, US-Pfa, US-Umb), or Chinese grasslands. Elevations for the Chinese grasslands are not given, but the Tibetan site obviously has some elevation. CN-Du3 is located close to the region Inner Mongolia north of Beijing, and CN-Sw2 is located within Inner Mongolia west of CN-Du3. All of the sites that fail likely feature temperatures falling substantially below the freezing point, and my suspicion is that JULES fails for these sites due to freezing soil water. (irr_crop is set to 0, so theoretically vegetation should get irrigated all year. Those three sites still fail when setting irr_crop to 1, and setting irr_crop to 2 requires the crop model to be active.)
It has to be noted, though, that JULES doesn't fail at some other cold sites, e.g. FI-Hyy, FI-Lom, US-Los, US-Syv, US-WCr.

comment:9 Changed 5 months ago by pmcguire

  • Cc kwilliams added
Note: See TracTickets for help on using tickets.