Couple Phenology into veg3/JULES-RED

[mo-arthurargles]

Couple Phenology into veg3/JULES-RED

Overview

This issue links the LAI estimated in the RED/veg3 demography scheme to the pre-existing phenology code in JULES, coupling the phenology routines to the veg3/JULES-RED component. Alongside the coupling, the initialisation and diagnostics have been updated to allow a reasonable scientific comparison between the older veg2/TRIFFID vegetation dynamics scheme and RED, and the RED dynamic/veg3 coupling code has been refactored and tidied to ease future development.

Links

Development branch: jules_red_veg3_couple_phenology

Pull request: MetOffice/jules#160

Evaluation rose suite for JULES-RED phenology coupling: u-ec255@366212, a temperate Fluxnet site in Belgium (BE), run 2006-2015.

Evaluation presentation: JULES_RED_Phenology_evaluation.pptx

Scientific Evaluation

The evaluation suite was designed to answer three questions:

1. Do we get a correct seasonal cycle in terms of leaf turnover and leaf area index (LAI)?
2. Do we get an identical simulated state after doing a dump restart for the final year of simulation?
3. Can we run with independent phenol_period (kept at 1 day) and triffid_period (set to 10 days)?

Key findings

• Seasonal cycle: JULES-RED reproduces realistic seasonal cycles in leaf turnover and LAI, with phenological behaviour broadly consistent with JULES-TRIFFID for both BDT and NET plant functional types.
• Litterfall treatment: The updated litterfall treatment prevents cumulative leaf litter losses from exceeding the available leaf carbon pool, improving physical consistency for deciduous trees (JULES-TRIFFID's litter loss can exceed the leaf carbon pool by 2-4x for BDT).
• Assimilate: For BDT, JULES-RED avoids removing more leaf carbon than is actually present, resulting in less of a carbon "tax" during winter compared to TRIFFID.
• Dump restarts: Dump restart tests show identical values for assimilate, LAI-related variables (LAI, LAI_BAL, LAI_PHEN), and the prognostic plantNumDensity after restart.
• Independent update periods: Simulations run correctly with independent phenol_period (1 day) and triffid_period (10 days) settings, with vegetation dynamics updated on the expected timescale.s.

Previous links/versions

This issue has been split off from the old MOSRS ticket #1357.