Differences in land CO2 fluxes between global models and country GHG inventories: implications for the Global Stocktake
1. Differences in land CO2 fluxes between
global models and GHG inventories:
implications for the Global Stocktake
Giacomo Grassi (JRC)
Detlef Van Vuuren (PBL), Elke Stehfest (PBL), Simone Rossi (JRC), A. Cescatti
(JRC), J. House (Bristol Univ), A. Popp (PIK), J. Rogelj (IIASA)
COP25, IPCC SRCCL, 3rd December 2019
2. 2
Modified from : Creating common purpose:
the integration of science and policy in
Canada’s Public Service, Canadian Centre for
Management Development,2002
Can you tell me where I am?We’re lost.
And you must be a policymaker.
I gave you an accurate answer, but you
don’t understand …
You are at Latitude 50 North and Longitude 4 East, at
100 m above sea level.
You must be a scientist. I asked you a simple
question, you gave me too complex information
and I’m still lost.
3. 3
The Global Carbon Budget
(average 2007-2016 from Global Carbon Project 2017)
46%
Atmosphere
Forests
Oceans
30%
24%
88%
12%
+
Fossil fuel emissions
Land use change
Land Use, Land-Use Change and Forestry (LULUCF)
Energy,
transport,
etc.
The forest sink is
complex to measure
and only partly
anthropogenic
4. 4
Global models (IPCC Reports)
Land-use change
(Bookkeeping, IAMs, DGVMs)
Anthropogenic CO2 flux from land-
use change, regrowth, some other
form of management
Land sink (DGVMs)
Non-anthropogenic CO2 flux,
associated to the response of land to
human-induced environmental
changes (e.g., increasing
atmospheric CO2, nitrogen
deposition, climate change)
Land-related Co2 flux categories
Country GHG inventories
Land use, Land-use change and
Forestry (LULUCF)
“Managed land” adopted as a pragmatic
proxy for anthropogenic CO2 fluxes (IPCC
2006 Guidelines) GHG inventories
REPORT net emissions for managed land
(“what the atmosphere sees”).
Then, when ACCOUNTING the LULUCF
contribution toward countries’ targets (NDC),
special accounting rules may be used to filter
the reporting and reflect better the impact of
mitigation actions.
?
comparable
5. 5
Atmospheric budget
Anthropogenic GHG emissions/removals
Mitigation actions
Global models & IPCC Reports
Global models & IPCC Reports
Country GHG inventories
(reporting)
Country NDC
(accounting)
6. 6
- Grassi et al. 2018)
Global anthropogenic land CO2 fluxes: models vs. countries
For 2010, ≈ 5 GtCO2/y gap
(mainly on the forest sink)
Country land-related
climate pledges
IAMs land mitigation pathways
This gap holds also for the
future: is it a problem?
(based on IPCC SRCCL Fig 2.4)
7. 7
Max weight
allowed is 150 kg IPCC
ARs
you have to slim!
I weigh
12o kg
I weigh
8o kg
GHG
inventories
I will slim
10 kg
I will slim
20 kg1st NDC
(80-10) + (120-
20) = 170 kg ..
not enough!
How the Paris Agreement works
8. 8
The Global Stocktake (GST) will assess (from 2023, every 5 yrs) the countries’
collective progress towards the < 2°C target “in the light of the best available science”.
The GST requires
comparability
Inputs to theGST: a) Aggregated countries’ GHG data
b) IPCCAR6 and other scientific data
compared to assess
the future “gap”
Future
GAP
Historical
gap
Future
GAP
10. 10
1) Coarse & incomplete representation of land-use change / management in global
models (Pongraz et al. 2018) impact difficult to quantify
2) Inaccurate & incomplete estimation of land CO2 flux in country GHG inventories,
especially in developing countries (Grassi et al. 2017) difficult to quantify
3) Conceptual differences in estimating «anthropogenic» land CO2 sink, which make
global models and countries hardly comparable >3 GtCO2/y (Grassi et al. 2018).
While in the long-term improvements are expected in both models and countries, the
Global Stocktake in 2023 requires quick fixes to allow comparing country pledges with
IAMs’ mitigation pathways.
Possible causes of the gap between global model and countries
11. 11
These differences are reflected also in the SPMs of IPCC SR1.5 and SR CCL:
A.3.3. Global models and national GHG inventories use different methods to estimate anthropogenic CO2 emissions and removals for the land sector. Both
produce estimates that are in close agreement for land-use change involving forest (e.g., deforestation, afforestation), and differ for managed forest. Global
models consider as managed forest those lands that were subject to harvest whereas, consistent with IPCC guidelines, national GHG inventories define managed
forest more broadly. On this larger area, inventories can also consider the response of land to human-induced environmental changes as anthropogenic, while the
global model approach treats this response as part of the non-anthropogenic sink. For illustration, from 2005 to 2014, the sum of the national GHG inventories net
emission estimates is 0.1±1.0 GtCO2yr- 1, while the mean of two global bookkeeping models is 5.1±2.6 GtCO2yr-1 (likely range). Consideration of differences in
methods can enhance understanding of land sector net emission estimates and their applications.
Global models and countries use different approaches to estimate anthropogenic land CO2
fluxes, due to differences in:
a) attributing the impact of human-induced environmental changes
b)forest area considered managed
12. 12
a) Differences in attributing the impact of human-induced environmental changes
(IPCC SRCCL Fig. 2.6)
Direct human
induced effects
Land use change
Management (e.g. harvest)
Indirect human
induced effects
Climate change, atm.CO2
increase, N deposition etc.
Natural effects
Interannual variability
Natural disturbances
Most countries include the impact of “indirect effects” (environmental change)
≈ 5 GtCO2/y ≈ -11GtCO2/y ≈ 0 GtCO2/yAverages* for 2005-2014
*Bookkeeping and DGVMs: Le Quere et al., 2018; IAMs: average for 2010 of IAM runs in SSP Database. Country GHG Inventories: Grassi et al., 2018
Managed Unman.Managed Unman.
✔
Bookkeeping/IAMs
Anthropogenic
“Land-use change”
Country inventories
Anthropogenic
“LULUCF”
Managed Unman.
✔
✔
✔
Managed Unman.
✔ ✔
✔ ✔
DGVMs
Natural
“Land Sink”
13. 13
Primary forest
bookkeeping model
natural forest IAMs
Unmanaged forest
GHGIs
0
1000
2000
3000
4000
5000
Mha
Global forest area
Secondary forest
bookkeeping model
managed forest IAMs
Managed forest GHGIsBookkepping IAMs Countries
b) Differences in forest area considered managed
• Total forest area similar (around 4000 Mha)
• ‘Managed’ area in Bookkeeping and IAMs (i.e. subject to harvest) much smaller than
Countries ‘managed’ area (where human practices are applied to perform production,
ecological or social functions)
Globalforestarea(Mha)
IPCC SRCCL
Fig. SPM.1.
Managed
forest
Intact
forest
Global land use in 2015
“60–85% of the total forested
area is used, at different levels
of intensity”
14. 14
This sink in
anthropogenic
No, that’s
natural
Different scopes, two communities different approaches to identify anthropogenic flux.
Countries include the impact of “environmental change” on a larger “managed” area.
And for the future (IAMs scenarios?)
Changing the countries’ approach is difficult.
Models’ results may be post-processed
approach successfully applied to historical period:
How to reconcile these differences for the Global Stocktake?
15. 15
CURRENT Land CO2 flux categories NEW Land CO2 flux categories
Land use (Bookkeeping/IAMs):
CO2 from direct effects in managed land
a) CO2 from direct effects in
managed land
Natural land sink (DGVMs):
response of land to human-induced
environmental changes (indirect + natural
effects), in both managed and
unmanaged land
b) CO2 from indirect + natural
effects in managed land
c) CO2 from indirect + natural
effects in unmanaged lands
e) Natural “Land sink”: b + c
d) LULUCF in country GHGIs: a + b
Possible new table to “translate” global model estimates:
a) Differences in attributing the impact of human-induced environmental changes
consistency
with previous
estimates
consistency with
country GHG
inventories
(Grassi et al., in preparation)
16. 16
b) Differences in managed forest area
ESA-CCI: 300 m global land cover map 1992-2015 by
European SpaceAgency'sClimate Change Initiative
Intact forests
(Popatov et al.)
ESA-CCI (total forest classes)
minus
Intact forests (≈unmanaged)
= ‘managed’ forest
We need a map of managed areas close to countries’ approach
P
b
n
U
G
In
0
1000
2000
3000
4000
5000
Mha
Global forest area
Se
bo
ma
Ma
ES
= m
Houghton IAMs Countries ESA-CCI
(Grassi et al., in preparation)
17. 17
700
700
3200
3200
Managed
Forest area
(Mha, 2010)
Illustrative results for IMAGE, SSP1- baseline
-4
-2
0
2
4
6
8 2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
GlobalanthropogeniclandCO2flux(GtCO2/y)
IMAGE SSP1 BAU - Reconciliation with Country GHG data
Direct effects only,
model's forest area
Countries, unconditional
NDC
Countries, conditional
NDC
-4
-2
0
2
4
6
8 2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
GlobalanthropogeniclandCO2flux(GtCO2/y)
IMAGE SSP1 BAU - Reconciliation with Country GHG data
Direct effects only,
model's forest area
Direct + indirect effects,
model's forest area
Countries, unconditional
NDC
Countries, conditional
NDC
-4
-2
0
2
4
6
8 2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
GlobalanthropogeniclandCO2flux(GtCO2/y)
IMAGE SSP1 BAU - Reconciliation with Country GHG data
Direct effects only,
model's forest area
Direct + indirect effects,
model's forest area
Direct + indirect effects,
countries' forest area
Countries, unconditional
NDC
Countries, conditional
NDC
Countries’ data are from Grassi et al. 2017
18. 18
Conclusions
• Large gap in land CO2 fluxes between global models and countries, mostly on forests.
• Gap mainly due to a conceptual difference in estimating “anthropogenic” CO2 sink:
countries include the impact of “environmental change” on a larger “managed” area.
• Reconciling these differences is needed to assess country collective progress towards
meeting modelled mitigation pathways under the Global Stocktake.
• To this aim, preliminary results of an ex-post processing of IAMs results are promising.
• Next steps: use data from all IAMs
This is how science look at the global C cycle.
You can see already that LULUCF in both sides of the panel: problem and potential solution
And this this how these fluxes are captured in GHG inventories, which are the basis of climate policies.
Global models tipically dintighuish Land-related Co2 flux categories in 2 categories: «land-use change» and «land sink».
The countries GHG inventories use the term LULUCF.
Since the categories «land-use change» and «LULUCF» are described as «anthropogenic», they are assumed to be comparable. The following slides will challenge this asumption
Next slides will focus of global models/IPCC report and country GHG reporting
When comparing the global-level net land anthropogenic CO2 fluxes, a large difference of about 5GtCO2 is observed for recent years between Bookkeping models and country GHG inventories. The reason for this discrepancy will be elaborated in the following slides – the short version of it is that we are comparing apples with oranges.
This gap holds also for the future, between IAMs scenario and country climate pledges. Is it a problem?
To answer this question ,let’s see how the PA and its Global Stocktake (GST) work...
These two men want to fly in the Balloon. This is their objective, like staying well below 2C for the countries.
150 kg is the <2C trajectory.
The graph in the left is the ideal condition. In the historical period, scientific and country data match.
However, the current situation is better expressed in the graph on the right, where a GAP exist in the historical period between scientific and countries.
In the context of the GST, any significant GAP between model scenario and countries’ data for the HISTORICAL PERIOD will make the assessment of FUTURE “emission gap” more difficult and uncertain.
In other words, it’s important that the men and the scientists use the same balance
The focus of this presentation will be on these conceptual difference in estimating «anthropogenic» land CO2 sink
Let’s see the two reasons better in the next slides
And now let’s see how the MODELS and the COUNTRY GHG inventories include these effects in their estimates
The bookkeeping models, by definition, include only DIRECT effects. The same for the IAMs (IMAGE is an exception in including also indirect effects in managed lands)
The «natural Land sink» estimated by the DGVM includes the indirect + natural effects on all land
Finally, country GHG inventories include both direct effects and, in MOST CASES, also indirect+natural effects occurring on the land that they considered «managed»
Therefore, we can now understand the reason of the discrepancy in estimates: Most GHG inventories include the impact of “indirect effects”, on a much larger managed forest area than global models.
FIRST REASON: Differences in managed forest area
t’s worth to note that the IPCC SRCCL provides estimate of «used forest» very close to country GHGI
We propose a new table to translate the conceptual differences in estimates between countries and global C models:
The CURRENT categories include include only LAND USE and the NATURAL LAND SINK.
In the proposed NEW structure, LAND USE would remaning the same, while the NATURAL LAND SINK would be split in two: the flux occurring in managed land and the one in Unmanaged land.
Conceptually, summing a+ b would increase consistency with country GHG inventories
Let’s start with reconciling the difference in AREA: how to get a a map closer to countries’ approach to managed forests?
To this aim, we combined ESA-CCI and intact forest map. The Intact forest map is a well-know dataset (from Hansen group) used for many purposes.
Subtracting the intact forest map to the total forest map, we obtain a map of non-intact forest
It can be seen that most of the difference is due to the large difference in area considered managed between ORANGE and GREEN.
We should NOT expect a perfect match (impossible to achieve!), but it’s clear that the proposed approach makes the original models results much more comparable with country GHGI.
