Estimating GHG Emissions from Dynamic Global Vegetation Models
1. Dynamic Global Vegetation Models
IPCC event: Estimating GHG Emissions - Reconciling Different Approaches
Dr. Elena Shevliakova ( NOAA/GFDL, USA)
IPCC pavilion - UNFCCC COP27
Sharm el-Sheikh - November 2022
2. What is a Dynamic Global Vegetation Model
• Mathematical equations describing biogeochemical cycling (e.g. Carbon) in vegetation and soils on managed
and unmanaged lands as affected by
• climate change and variability
• changes in atmospheric composition (e.g. CO2 concentration) and fluxes to land (e.g. Nitrogen deposition)
• Land Use and Land Cover Change (via reconstructions and scenarios)
• Vegetation and soil densities of Carbon and other elements are simulated, and not prescribed
• These equations are solved by computer globally on a grid of a certain size (i.e. resolution) for simulations
carried from decades to centuries, from pre-industrial time into the future
• Stand-alone or components of Earth System models
SRCCL, 2019
3. What is under the hood in a typical DGVM?
Parametric and Structural
Uncertainty
• Parametrizations of natural
processes and land use
practices (e.g. Arneth et al
2017, IPCC 2020 ,WG1, Ch. 5)
• Large uncertainty about soil
carbon responses to warming,
drying/wetting, and to
changes in litter quality
(SRCCL, Ch 2. 2019)
• Few DGVMs simulate
vegetation demography
(Argles et al 2022)
Canopy
and
canopy
air
Soil/snow
Atmosphere
Photosynthesis
Plant
and
soil
respiration
Energy
and
moisture
balance
C
&
N
uptake
and
release
t~ minutes to day
fine
roots
Land energy, water, carbon
and nitrogen exchange
leaves
sapwood
labile
wood
C
&
N
allocation
and
growth,
t
~
day
Phenology,
t~
day
month
Mortality,
natural
and
fire
t
~
day
-
year
Biogeography,
t
~
prescribed
or
dynamic
{
Land-use
management,
t
~
day
-
year
Climate statistics
Carbon gain
Plant type
LAI, height,
roots
Land-use
transitions
Vegetation, C&N dynamics
4. • Land-use transition - gross or net
rates from historical reconstructions or
future scenarios, usually based on
IAMs
• Transitioning land fractions carry
vegetation and soil carbon remaining
after vegetation clearing/harvesting
• DGVMs differ substantially in handling
vegetation and soil carbon after
transitions (e.g., Arneth et al 2017,
IPCC 2020 WG1, Ch. 5)
• Conserve C mass
4
example for NOAA/GFDL models LM3 and LM4; Shevliakova et al. 2009, Malyshev et al. 2015
Managed lands
DGVMs differ substantially in representation of managed lands and
implementation of land use scenarios
For each model grid
5. Simulated C Budget in most DGVMs cannot be reported
as the Managed Land Proxy
Δ land (C) =
- Gross LU C emissions from clearing/deforestation
- Gross emissions from wood and ag. harvesting
+ NEP - Fire on second. lands, croplands, & pasture
+ NEP - Fire on natural lands
5
Net
C
flux
on
managed
lands
Biosphere
sinks
and
sources
• Most DGVMs do not separate
soil carbon pools under a
different kind of land use and
are unable to report losses
from soils and Net Ecosystem
Production (NEP) by land-use
category
• Most DGVMs do not separate
secondary and primary
vegetation and cannot report
the regrowth of secondary
vegetation
• Limitations in capturing
regional differences in clearing
and harvesting practices
CMIP6 LUMIP
https://www2.cgd.ucar.edu/research/mips/lumip
6. Land Use Fluxes are typically reported by DGVMs as a difference
between simulations with different assumptions about land use change,
climate change, and atmospheric CO2
The managed land proxy
*negative flux is
into the atmosphere
Ciais et al 2022 ,
Simulations by the
JSBACH DGVM
7. Insights from the NOAA/GFDL land model LM4 with LU fluxes computed
as the Managed Land Proxy
Gt
C/yr
(S3- S2) (S4-S0, pre-industrial climate )
(managed land proxy)
• LUH2 CMIP6 land use scenario, gross
transitions
• Cleared C left on croplands
• Croplands have grass type, but the
growing season is prescribed per
current observations
• As in JSBACH, (S3-S2 approach)
widely used in stand-alone DGVMs
overestimates present-day LU Flux
• The Net Land Use flux computed as
Managed Land Proxy (green) is similar
to the Net LU Flux with constant climate
and CO2 up to the 1990s
8. Insights from the NOAA/GFDL land model LM4 with LU fluxes computed
as the Managed Land Proxy in historical simulation
Gt
C/yr
• Secondary area since 850 per LUH2 CMIP6
• Large secondary vegetation C sink (blue),
depends on the age of secondary lands
• Rangelands (red) and pasturelands (green) are
small C source
• Croplands (with deforestation) are a large,
increasing C source due to both soil respiration
from the legacy fluxes from deforestation AND
from the soil C emissions from stocks
accumulated prior to deforestation
Gt
C/yr
Up to 50 yr
Secondary lands flux
All
• One can’t say how consistent or inconsistent LM4 MLP fluxes with
Inventories without comparing them at the same locations and
for the same C stock changes, particularly soils
includes clearing/deforestation
9. Key Messages
• Comparing global Net LU fluxes from DGVMs and inventories is not informative because
they are estimating fluxes from different areas (e.g., Grassi et al 2018), different C pools, and
under different assumptions about carbon stocks transitions due to LULCC
• Need to understand the LU fluxes on pasturelands, cropland, and secondary lands – the
magnitudes and signs could be different
• It would be helpful to compare DGVMs and inventories in terms of the components of their
budgets where it’s possible (e.g., changes in vegetation stocks) by region and country
• For DGVMs that separate soil C pools on different land use, particularly croplands, we can
start to compare LU fluxes by land use practice, particularly secondary lands, and establish a
methodology for considering their age.
• The modeling community needs to characterize structural uncertainty in DGVMs and
compare land use management representation with those included in the inventories,
including the assumptions about the fate of cleared and harvested carbon as well as the fate
of residue to understand sources of differences in soil C stocks.
• More communication is needed between modelers and the inventory community