Orientering om ny skovfremskrivning, fra klima-, energi- og forsyningsministeren

Side 1/3
Kontor
Kontor for Klimafremskrivning
Dato
22. april 2024
J nr. 2024- 1671
/ RVKCH
Orientering om skovfremskrivningsmodel til brug for KF24
Til brug for Klimastatus og -fremskrivning 2024 (KF24) har Institut for Geoviden-
skab og Naturforvaltning (IGN) på Københavns Universitet udviklet en ny skovfrem-
skrivningsmodel med henblik på at forbedre nøjagtigheden.
Arbejdet med en ny fremskrivningsmodel blev igangsat, idet den tidligere skovfrem-
skrivningsmodel havde vanskeligt ved at forklare skovenes CO2-optag på kort sigt.
Blandt andet underestimerede den tidligere model skovenes optag med ca. 0,7
mio. ton CO2e i 2021 og ca. 1,6 mio. ton CO2e i 2022. Der blev derfor indgået aftale
med IGN om, at IGN skulle udarbejde en mere retvisende model til KF24.
Danske skove er siden 1990 vokset både i areal og tæthed og ved tilvæksten har
skovenes træer optaget CO2 fra atmosfæren. Det skønnes i den nyeste Skovstati-
stik 2022, at skovenes samlede kulstoflager er steget kontinuerligt og er nu øget
med ca. 59 pct. ift. 1990. Det skønnes, at der er bundet ca. 202 mio. ton CO2e i
skovenes biomasse og ca. 23 mio. ton CO2e i træprodukter.
Den nye model fra Københavns Universitet er mere finkornet og bygger på et bre-
dere datagrundlag. IGN estimerer skovenes kulstofpuljer i den nye skovfremskriv-
ningsmodel på et individuelt træniveau ved at indføre observerede stammedia-
metre og arter fra opmålinger til den danske skovstatistik. Herefter simuleres udvik-
lingen i kulstofpuljerne bl.a. ud fra vækstmodeller baseret på europæiske data samt
mortalitets- og hugstsandsynligheder baseret på skovstatistikken. På den baggrund
giver den nye model alt andet lige et mere realistisk bud på, hvad der kommer til at
ske i de danske skove.
IGN’s nye skovfremskrivningsmodel skønner ligesom den tidligere fremskrivnings-
model en forestående reduktion i skovenes CO2e-optag i de fem første frem-
skrevne år, jf. figur 1. Herefter stabiliseres fremskrivningen på et optagsniveau pr.
år på ca. 3 mio. ton CO2e fra 2035 og frem, hvor KF23 forventede et stabilt fremad-
rettet optag på ca. 0,5 mio. ton CO2e.
Offentligt
(Omtryk - 25-04-2024 - Klausulering ophævet) KEF Alm.del - Bilag 266
Klima-, Energi- og Forsyningsudvalget 2023-24
Side 2/3
Figur 1
Skovenes udledninger og optag i KF23, KF24 og Skovstatistik 2022, mio. ton CO2e
Anm.: De negative værdier indikerer CO2e-optag
Kilde: DCE og IGN
Reduktion i optag frem til omkring 2028 tilskrives som tidligere en forventning om
stigende hugst. IGN forklarer, at dette skyldes, at en større mængde af især bøg,
rød- og sitkagran er hugstmodent, hvilket betyder, at de med de beregnede hugst-
sandsynligheder forventes fældet i de kommende år. Sammenhængen mellem et
træs diameter og sandsynligheden for, at det bliver fældet, bygger på IGN’s histori-
ske observationer af, hvilken diameter træer tidligere er blevet fældet ved. Derved
bygger den fremskrevne reduktion på en forventning om, at skovens nuværende
træer fældes ved cirka samme diameter som tidligere.
Det nye stabile optagsniveau fra 2035 og frem på ca. 3 mio. ton CO2e tilskrives
grundlæggende ændringer i metoden bag fremskrivningerne, som giver anledning
til såvel ændret tilvækst og hugst. IGN forventer, at den nye model bedre afspejler
den faktiske virkelighed ude i skovene. I den nye model estimeres sandsynligheden
for, om enkelte træer fældes på baggrund af deres størrelse, mens det tidligere
blev estimeret om et område blev ryddet for alle træer, uagtet træernes individuelle
diameter. Således forudsiges især hugsten af bøgetræer bedre, hvor praksis nor-
malt er at fælde enkelte træer, når de er vokset til en passende størrelse. Derud-
over har det europæiske datasæt fra EFISCEN-modellen forbedret det statistiske
grundlag for estimeringen af træers vækst, da der nu er tilføjet væsentligt flere må-
linger af træers udvikling på tværs af Europa.
Grundet denne mere finkornede modellering af skovenes udvikling, forventes sko-
venes kulstofoptag i KF24 at være ca. 2,3 mio. ton CO2e større i 2025 og ca. 1,4
mio. ton CO2e større i 2030 sammenlignet med KF23, jf. figur 2.
Side 3/3
Figur 2
Øget CO2-optag i skov i KF24 vs KF23, mio. ton CO2e
Kilde: DCE og IGN
Usikkerheder
Grundlæggende vurderes opgørelsen og fremskrivningen af udledninger og optag
fra skov og høstede træprodukter samlet set at være forbundet med en større usik-
kerhed end for de fleste andre sektorer. Det skyldes, at nettoudledninger og -optag
er et resultat af små ændringer i store kulstofpuljer. Konkret vurderer IGN en årlig
usikkerhed på ca. 1,5 mio. ton CO2e i skovenes udledninger og optag. Usikkerhe-
derne forbundet med fremskrivningen må forventes at være større. Anvendelsen af
den nye skovfremskrivningsmodel ændrer ikke på denne usikkerhed
Den nye skovfremskrivningsmodel forventes alt andet lige at reducere usikkerhe-
den forbundet med at forudsige omfanget af trætilvækst samt træfældning, der vil
foregå i de enkelte år. Denne usikkerhed opstår, fordi der er mange aktører involve-
ret i forvaltningen af skovarealet. Den faktiske forvaltning af skovarealet i de kom-
mende år afhænger udover træernes alder og diameter af mange andre faktorer
såsom økonomi, priser og efterspørgsel. Udviklingen i skovens kulstofpulje er der-
for behæftet med væsentlig usikkerhed, og forskydninger i hugst vil kunne påvirke
det faktiske forløb i årene, der kommer.
Grundet den store årlige usikkerhed anvender IGN en udjævningsmetode til at re-
ducere de årlige udsving i skovenes historiske og fremskrevne optag og udlednin-
ger. Midling udføres over en femårig periode. Således er det midlede optag i sko-
vene for 2024 udregnet ved at trække skovenes optag i 2019 fra skovenes optag i
2024 delt med fem.
2,3
1,4
0
1
2
3
2025 2030


Side 1/2
Ministeren
Dato
24. april 2024
J nr. 2024-1671
Klima-, Energi- og
Forsyningsministeriet
Holmens Kanal 20
1060 København K
T: +45 3392 2800
E: kefm@kefm.dk
www.kefm.dk
Klima-, Energi- og Forsyningsudvalget
Christiansborg
1240 København K
Orientering om ny skovfremskrivning
Kære alle
Institut for Geovidenskab og Naturforvaltning (IGN) på Københavns Universitet
udgiver deres nye skovfremskrivning i dag den 25. april kl. 9. Til jeres oriente-
ring fremsendes hermed den nye skovfremskrivning, som er fortrolig indtil kl. 9.
IGN udarbejdede også den tidligere skovfremskrivning, der blev brugt i forbin-
delse med Klimastatus og -fremskrivning 2022 og 2023. Grundet flere års afvi-
gelser mellem fremskrevne og historiske data indgik Klima-, Energi- og Forsy-
ningsministeriet i november 2023 en aftale med IGN om, at IGN skulle udvikle
en ny og mere retvisende skovfremskrivningsmodel til brug for Klimastatus og -
fremskrivning 2024 (KF24). Den nye model er mere finkornet og bygger på et
bredere datagrundlag, der går på tværs af Europa.
På baggrund af IGN’s nye skovfremskrivningsmodel forventes et øget kulstofop-
tag fra skovene sammenlignet med KF23 svarende til en partiel mankoreduktion
i KF24 på ca. 2,3 mio. ton CO2e i 2025 og ca. 1,4 mio. ton CO2e i 2030.
Der er fortsat stor usikkerhed forbundet med at fremskrive kulstofoptaget i sko-
vene. Dette skyldes skovens store kulstofpulje, hvor selv små statistiske usik-
kerheder kan have stor betydning for fremskrivningen af årlige ændringer.
Udover IGN's rapport vedhæftes et notat, der beskriver de væsentligste elemen-
ter i den nye model og dens resultater.
KEF-udvalget er inviteret til en teknisk briefing om den nye skovfremskrivning i
Klima-, Energi- og Forsyningsministeriet den 25. april kl. 15.30. Den nye skov-
fremskrivning vil også indgå i den tekniske briefing om den samlede Klimastatus
og -fremskrivning 2024, som KEF-udvalget er inviteret til den 30. april.
Offentligt
(Omtryk - 25-04-2024 - Klausulering ophævet) KEF Alm.del - Bilag 266
Klima-, Energi- og Forsyningsudvalget 2023-24
Side 2/2
Den nye skovfremskrivning indgår som en del af den samlede Klimastatus og -
Fremskrivning 2024, som udgives den 30. april. Der vil i klimafremskrivningen
være forhold, der medvirker til at reducere mankoen for vores klimamål, mens
andre forhold trækker i den anden retning. Først når vi har den samlede klima-
fremskrivning er der tilstrækkeligt grundlag for at vurdere den samlede status for
indfrielse af vores klimamål.
Med venlig hilsen
Lars Aagaard


–
Forest Carbon Pool Projections 2024
Thomas Nord-Larsen, Prescott Huntley Brownell II, and Vivian Kvist Johannsen
IGN Report
April 2024
university of copenhagen
department of geosciences and
natural resource management
Offentligt
(Omtryk - 25-04-2024 - Klausulering ophævet) KEF Alm.del - Bilag 266
Klima-, Energi- og Forsyningsudvalget 2023-24
Title
Forest Carbon Pool Projections 2024
Authors
Thomas Nord-Larsen, Prescott Huntley Brownell II, and
Vivian Kvist Johannsen
Citation
Nord-Larsen, Thomas, Brownell II, Prescott Huntley and Johannsen,
Vivian Kvist (2024): Forest Carbon Pool Projections 2024, IGN Report,
April 2024. Department of Geosciences and Natural Resource
Management, University of Copenhagen, Frederiksberg. 59 p. ill.
Publisher
Department of Geosciences and Natural Resource Management
University of Copenhagen
Rolighedsvej 23
DK-1958 Frederiksberg C
www.ign.ku.dk
Responsible under press law
Vivian Kvist Johannsen
External/internal review
Professor Henrik Meilby, University of Copenhagen
Associate professor Niclas Scott Bentsen, University of Copenhagen
ISBN
978-87-7903-927-8 (web)
Cover layout
Jette Alsing Larsen
Cover photo
Thomas Nord-Larsen
Published
This report is only published at www.ign.ku.dk
Citation allowed with clear source indication
Written permission is required if you wish to use the name of the
institute and/or part of this report for sales and advertising purposes
3
Preface
The large carbon pools of the forests have a relatively significant importance for the Land Use,
Land-Use Change, and Forestry (LULUCF) segment of the Danish emissions inventories and
thereby the overall climate accounting. In order to navigate towards the objectives of the Danish
climate goals, it is therefore necessary to know what emissions are expected from the forests.
Department of Geoscience and Natural Resource Management at University of Copenhagen has
previously made various projections of the forest carbon pools in different contexts and with
different assumptions to provide estimates of forest greenhouse gas emissions.
As a result of the diversity in the data used and the underlying assumptions, the projections have led
to different results. The models have moreover been aimed at a long-term projection but have had
difficulties in the short term in describing the actual development. There is thus a need for a
renewed projection, which shows the expected development in the carbon pools up to 2025 and
2030 and reflects the recent years' development. Consequently, this project was initiated on an
assumption that a simplified projection of forest carbon pools focusing on 2025 and 2030 and
linking to previous projections for the years up to 2040 would produce sufficiently accurate
estimates. However, during the project it proved difficult to simplify calculations while still
incorporating known changes in forest area, age and species distribution, and in forest management
practices on areas set aside for biodiversity protection.
Because of issues arising from simplification of the projections it was decided to take on a, for
Denmark, novel projection tool called EFISCEN-space. The EFISCEN (European Forest
Information SCENario) model, specifically the EFISCEN-space variant, is a spatially explicit forest
model developed to assess the future development of forests at regional to European scales. It
simulates forest growth and dynamics based on inventory data and user-defined management rules,
allowing for the analysis of different forest management and policy scenarios. The model accounts
for various factors such as age class distribution, volume, increment, and forest management
practices, making it a useful tool for predicting forest growth, timber production, and carbon
sequestration under various scenarios. The "space" component in EFISCEN-Space enhances the
model by incorporating spatially explicit information (i.e. plot locations), enabling more detailed
analyses of spatial patterns and processes in forest ecosystems. The foundation for setting up the
model was made on a study visit to Wageningen in November 2023.
4
Similarities and deviations from previous projections are described and justified in this report, and it
is explained why the chosen projection method is expected to provide a more accurate projection
towards and beyond 2030. The deliverable also includes a brief description of alternative projection
models, based on preliminary work with a larger project about forecasts for the forests'
contributions to climate and climate accounts.
The results of this study / project are partly based on the EFISCEN-Space model. We acknowledge
the use of the EFISCEN-Space model as developed by Stichting Wageningen Research, Wageningen
Environmental Research and Wageningen University, Department of Environmental Sciences since
2013.
5
Content
1 INTRODUCTION ..........................................................................................................7
1.1 Aim ......................................................................................................................................................................8
2 CARBON POOL PROJECTION...................................................................................9
2.1 Previous projections of Danish forest resources and carbon pools................................................................9
2.2 International projection models......................................................................................................................11
2.3 EFISCEN-Space model....................................................................................................................................13
3 MATERIALS...............................................................................................................15
3.1 The Danish National Forest Inventory ...........................................................................................................15
3.2 National Inventory data 2018-2022.................................................................................................................16
3.3 Projections of Danish forest carbon pools and emissions with the EFISCEN-Space model system .........17
3.3.1 Growth model................................................................................................................................................18
3.3.2 Harvest probability ........................................................................................................................................19
3.3.3 Designation of areas for nature protection.....................................................................................................20
3.3.4 Mortality........................................................................................................................................................25
3.3.5 Afforestation..................................................................................................................................................26
3.3.6 Ingrowth and reforestation.............................................................................................................................27
3.3.7 Dead wood.....................................................................................................................................................28
3.3.8 Litter ..............................................................................................................................................................28
3.3.9 Harvested wood products (HWP)..................................................................................................................28
4 RESULTS...................................................................................................................29
4.1 Effects of forest management and programme settings ................................................................................31
5 DISCUSSION .............................................................................................................35
5.1 Forest carbon projections................................................................................................................................35
6
5.1.1 Forest growth models ....................................................................................................................................38
5.1.2 Uncertainties..................................................................................................................................................38
5.2 Connection with greenhouse gas reporting ....................................................................................................41
5.2.1 Comparison to previous projections ..............................................................................................................43
5.3 Forest carbon projection methods ..................................................................................................................46
5.3.1 Simple projections .........................................................................................................................................46
5.4 Future development .........................................................................................................................................47
5.5 Assessing actual climate effects .......................................................................................................................49
5.6 Concerns Related to the Discontinuation of the National Forest Inventory................................................50
REFERENCES ..................................................................................................................53
6 APPENDIX .................................................................................................................55
7
1 Introduction
Forests play a pivotal role in the global climate system, acting as significant carbon sinks that
absorb and store carbon dioxide (CO2) from the atmosphere, thus mitigating the impacts of climate
change. The ability of forests to sequester carbon makes them invaluable in the fight against rising
global temperatures and the associated adverse effects on ecosystems and human societies.
However, this capacity is not static and it is influenced by a myriad of factors including forest
management practices, land-use changes, and natural disturbances, all of which can transform
forests from carbon sinks to sources of emissions.
In this context, projections of forest carbon stocks and emissions become crucial. They offer a
window into the future, enabling us to anticipate how different scenarios – ranging from business-
as-usual to wide spread management for nature conservation – might play out in terms of forest
health, biomass, and carbon sequestration capabilities. Such foresight is invaluable for policymakers
and environmental planners as it provides a scientifically grounded basis for formulating strategies
and policies aimed at climate change mitigation.
By understanding potential future states of forest ecosystems, decision-makers can better design
policies that not only protect these critical natural resources but also optimize their role in
sequestering carbon. This is essential for meeting international climate targets, such as those set by
the Paris Agreement, and for developing national strategies that align with sustainable development
goals. Thus, forest carbon stock and emissions projections are not just academic exercises; they are
essential tools for guiding global efforts towards a more sustainable and climate-resilient future.
This report presents a projection of forest carbon stocks and related emissions, offering insights into
the dynamic interplay between forest ecosystems and atmospheric carbon levels. Our findings aim
to inform policymakers, environmental scientists, and forest managers, providing a scientific basis
for sustainable forest management practices and climate change mitigation strategies. Through
comprehensive data analysis and modelling, this report underscores the role of forests in global
carbon cycling and the importance of informed decision-making in preserving these natural
resources for future generations.
8
1.1 Aim
The aim of this report is to make forest carbon projections to provide estimates of CO2 emissions
from Danish forests, including the five principal forest carbon pools as well as harvested wood
products. The aim is further perform a first evaluation of model results in comparison to observed
historical emissions.
9
2 Carbon pool projection
2.1 Previous projections of Danish forest resources and carbon pools
Previous projections of forest resources, carbon pools, and greenhouse gas (GHG) emissions were
based on Markov chain models. Markov chain models are a type of stochastic model that can be
used to project changes in the age class distribution over time. These models involve transition and
conversion probabilities that describe the likelihood of the forest moving from one condition (e.g.,
age class or species composition) to another in a given time period. The probability that the forest
area is transferred to the subsequent age class after a given period is termed the transition
probability whereas the net flow to or from the species classes is termed the conversion probability.
Early projections [1-3] were based the observed species and age-class distribution from
questionnaire surveys conducted by Statistics Denmark. In these studies, 10-year transition
probabilities were derived from the changes in species and age-class distributions observed from
consecutive surveys. For each species class, the aggregated probability that the forest area was
harvested at any given point in time was modelled from the observed area transition and the area
weighted site class in each county, using a logistic function. When applying the model, areas
transferred to the subsequent age-class were estimated as the conditional probability of surviving
into the next age-class (the transition probability) while areas transferred to the youngest age-class
was estimated as one minus the transition probability. The conversion probability was assumed to
be 0. Afforestation was assumed to always enter into the youngest age-class. The development of
forest growing stocks and harvest volumes were modelled from a mathematical formulation of
existing yield tables for the most common Danish forest tree species.
In later projection models, the transition probabilities were modelled from transfers between age-
classes observed on the permanent plots of the Danish National Forest Inventory (DNFI) [4-7]. In
the most recent projection of forest carbon emissions, the survival probability model was estimated
from data collected between 2002-2020 [4], reflecting the management of Danish forest land during
this period. The model used forest age, forest type (deciduous, coniferous, or Christmas
trees/ornamental greenery), and region (Jutland or the Islands) to predict the likelihood of a forest
area progressing to the subsequent age class. Here, growing stocks and harvested volumes were
estimated from observed growing stocks of the sample plots, rather than being modelled from yield
10
tables. The model furthermore included explicit modelling of afforestation and changes to forest
management resulting from the setting aside areas for biodiversity protection.
The Danish Forest Inventory data pose some challenges to the modelling using Markov chain
models as the data is much more detailed and comprehensive than the data from the earlier
assessments, reflecting the actual state of the forest to a much larger degree. Consequently, albeit
being used for more than three decades, modelling forest development from transition probabilities
has some inbuilt shortcomings in relation to contemporary forest management practices:
1) According to the DNFI data, large part of the forest is being managed according to other
principles than the clear-cutting system prescribed in the model, where harvested areas are
always transferred to the first age-class. Although the model could principally be built to
represent transfers to other age-classes, the complexity of the model increases dramatically.
2) To be operational, in the Markov-chain model each forest plot is represented by one
dominant species with one stand age in the model. However, a large proportion of the forest
area includes mixtures of tree species having different ages. Those forests are not
represented particularly well by the species and age-class specific model.
3) The transition probabilities have traditionally been modelled using different forms of
logistic models with a log-linear combination of parameters. Seemingly, the actual patterns
deviate from the model with a shape that is difficult to reproduce. Moreover, transitions at
stand level are rare occurrences and the data available is insufficient to capture the actual
system behaviour.
4) Age used in the models to determine harvesting probability is not the principal harvesting
criteria, which is rather being determined by tree size that is closer related to the resulting
forest products.
To summarize, the previous approach to forest carbon projections has become increasingly
inadequate owing to changes in contemporary forest management and forest structure. To alleviate
the problems with the Markov chain models, a shift in modelling approach towards projections
based on the growth and transition probabilities related to the individual tree, rather than the
regional or national species and age class distribution, is required. When the original models were
developed, this approach was not possible, as no model yet existed that could make such individual
tree projections on a national scale, and long-term data of sufficient resolution was not available.
However, the accumulation of Danish NFI data over two decades now represents a rich time series
11
of data making it possible to estimate these individual tree probabilities, and continued
advancements in computing power have enabled a new generation of models that is capable of
handling such tasks. To select an appropriate model for these projections, a Europe-wide analysis
was conducted with several criteria to identify an appropriate platform for the Danish forest carbon
projections.
2.2 International projection models
Forest projection tools are essential for understanding the dynamics of forest ecosystems, predicting
future changes, and aiding in sustainable forest management and policy making. Consequently, a
wealth of mostly local or national forest projection models have been developed but they are rarely
distributed outside their region of origin. However, a number of forest projection models have been
designed for and used at larger scales and across different climatic and geographical regions. Three
commonly used tools in this domain are EFISCEN, EFISCEN-space, EUREKA, and CBM. Each of
these models has unique features and applications.
EFISCEN (European Forest Information SCENario model) and EFISCEN-space
EFISCEN [8, 9] is a large-scale forest model that projects forest development at regional to
European scales based on national forest inventory data and scenarios of forest management. It
focuses on the volume and biomass of forest stands, considering different tree species, age classes,
and management regimes.
The first versions of the EFISCEN model [8] were built much like the earlier Danish projection
models relying on Markov chains to model age-class distribution development. More recently, the
modelling concept has been changed to rely on single tree observations in the novel EFISCEN-
Space model [9]. This model relies on individual tree observations from forest inventory data
typically from the sample plots of national forest inventories to project forest growth, thinning, and
felling. By making changes to the individual tree harvesting probabilities or climate attributes, the
model can analyse various scenarios related to forest management, climate change, and policy
impacts. Such changes may be applied locally, allowing differentiated treatments according to local
conditions. Outputs include timber volume, biomass, carbon storage, and potential wood supply.
12
EUREKA (European Forest Ecosystem Research Network)
EUREKA is not a single model but rather a network that facilitates the use and development of
various forest models and tools across Europe. It aims to support research, policy-making, and
sustainable forest management by providing a platform for sharing knowledge and methodologies.
The collaborative platform integrates different forest modelling approaches and tools and supports a
wide range of research themes from forest growth and yield to biodiversity and other ecosystem
services. The emphasis on collaboration across different research spheres facilitates data exchange,
methodological standardization, and the application of best practices in forest modelling.
EUREKA mainly supports research and academic studies on forest ecosystems and their
management. The model network has however been used for policy support and decision-making
through the integration of various modelling tools and approaches.
CBM (Carbon Budget Model of the Canadian Forest Sector)
The CBM is a forest carbon accounting model developed by the Canadian Forest Service. It is used
to estimate carbon stocks and stock changes in forest biomass, dead organic matter, and soil carbon
pools under different land-use scenarios and management practices.
The CBM offers detailed accounting of carbon fluxes in forest ecosystems, including emissions
from disturbances like fires, harvesting, and natural disturbances. The model can be applied at
various scales from stand-level to national inventories and supports scenario analysis for forest
management, land-use change, and climate change impacts.
The CBM is tailored for national and sub-national greenhouse gas reporting and carbon accounting.
Furthermore, the model offers potential for making research on forest carbon dynamics and the
impact of management practices on carbon sequestration.
Each of the three tools presented above, offers detailed projections based on inventory data and
contribute to a comprehensive understanding of forest carbon dynamics hereby enabling valid
projections for forest carbon stocks and emissions. However, when selecting the most appropriate
model, we set up a set of criteria including that the candidate model should:
1) have been successfully tested under European conditions and initialised using data from
more than one country,
2) be freely available. This excluded proprietary models designed for country-specific
circumstances or data sources as they would likely require further effort to implement,
13
3) be able to project forest development under different management scenarios, enabling the
consideration of planned management changes in the state forests,
4) be capable of handling mixed stands and growth and development of unmanaged forests.
5) make use of the multi-decade dataset from the Danish NFI, and
6) enable the use of Danish-specific volume and biomass functions.
Based on these criteria and the time available for this study, we chose to use the EFISCEN-Space
model.
2.3 EFISCEN-Space model
The EFISCEN-Space model [9] represents a state-of-the-art, spatially explicit model designed for
comprehensive simulations of forest dynamics, management interventions, and policy scenarios.
Developed collaboratively by European forestry research institutions, EFISCEN-Space integrates
advanced ecological, economic, and social components to provide a nuanced understanding of the
intricate interplay between forests and anthropogenic influences.
At its core, EFISCEN-Space utilizes a dynamic, individual-tree-based approach to simulate forest
stand development over time from national forest inventory sample plots (Figure 2.1). The model
captures the growth and mortality of individual trees, considering factors such as tree species, age,
and environmental conditions. By employing a spatially explicit grid, EFISCEN-Space enables
detailed assessments of forest dynamics at regional and national scales, allowing for a more
accurate representation of diverse ecosystems.
14
Figure 2.1. The EFISCEN-Space matrix model (from [10]). The modelling of tree increment includes a large set of
geographically explicit factors such as climate and soil conditions that enables adaptation to local growing conditions.
EFISCEN-Space's temporal dynamics are driven by a combination of ecological processes,
including natural disturbances such as wildfires, storms, and insect outbreaks. The model
furthermore enables the integration of climate data to project the impact of future climate scenarios
on forest growth and composition. Additionally, land-use changes, reflecting both societal and
economic influences, are incorporated to assess the consequences of evolving human interventions
on forest landscapes.
One of EFISCEN-space's notable features is its ability to simulate various forest management
scenarios. The model incorporates parameters related to thinning, harvesting, and regeneration,
allowing for the exploration of different management strategies and their implications on forest
structure and composition. This functionality is crucial for evaluating trade-offs between competing
objectives, such as maximizing timber yield while maintaining ecological integrity.
EFISCEN-Space's versatility extends to its capacity for simulating multiple ecosystem services.
Beyond timber production, the model enables assessment of biodiversity, carbon sequestration, and
water regulation, providing a comprehensive perspective on the multifaceted contributions of
15
forests to society. This holistic approach facilitates the development of policies that prioritize
sustainability and balance diverse societal needs.
In its inaugural years, EFISCEN-Space has demonstrated its utility in informing forest management
practices and policy decisions. Ongoing technical refinements, calibration efforts, and validation
exercises continue to enhance the model's accuracy and reliability. As EFISCEN-Space evolves, it
stands as a powerful tool for addressing the complex challenges associated with sustainable forest
management, contributing to the advancement of resilient and adaptive strategies for European
forests in the face of changing environmental conditions.
3 Materials
3.1 The Danish National Forest Inventory
The EFISCEN-Space model at its core is developed and initiated from national forest inventory
data. The Danish National Forest Inventory is based on a nationwide 2 x 2 km grid [11]. In each of
the grid cells, a cluster consisting of four sample plots is placed in the corners of a 200 x 200-meter
square. All clusters are measured over a five-year period, with one-fifth of the sample plots evenly
distributed across the country being measured each year. One-third of the groups are permanent and
are located in the southwest corner of the grid cells. These are re-measured for each five-year
rotation of the forest statistics measurements. Two-thirds of the groups are temporary and are
randomly moved within the respective 2 x 2 km cell in the grid for each repetition of the five-year
rotation. With particular reference to the present study, the permanent plots may be used for
assessing growth, probabilities of natural mortality and ingrowth, and management activities related
to harvest of trees and planting of trees.
16
Figure 3.1. Design of the Danish NFI [11].
The sample plots in the forest statistics are circular and have a radius of 15 meters. In total, there are
approximately 43,000 sample plots in the network, with only forest-covered sample plots being
measured over a five-year period. The forest-covered sample plots are identified before each
measurement season based on the latest aerial photos (typically less than one year old). In the forest,
each individual sample plot is located with high geographical precision, allowing for accurate
remeasurement of permanent sample plots as well as linkage with other geographical registry
information. In each sample plot, measurements of many variables are taken, including
measurements of tree size, age, and species, quantity of deadwood, and thickness of the litter layer
(forest floor branches, leaves, etc.), which are among the key factors for estimating forest carbon
pools.
3.2 National Inventory data 2018-2022
The National Forest Inventory data from the latest rotation of measurements is of particular
importance to the carbon pool projections as it forms the baseline and starting point [12]. The
measurements totalled 9.693 sample plots within clusters for which at least one of the sample plots
had forest cover (Table 3.1). Of the total amount of sample plots, 33 % were within permanent
clusters and in most cases remeasured from earlier rotations.
17
Table 3.1. Number of measured clusters and sample plots in the five-year rotation 2018-2022.
Year Clusters Plots
Total Forest Total Forest
2018 2.191 903 8.586 2.018
2019 2.186 844 8.597 1.896
2020 2.190 887 8.569 1.886
2021 2.175 883 8.528 1.951
2022 2.207 879 8.643 1.942
Total 10.949 4.396 42.923 9.693
During the measurements in the 2018-2022 cycle, 114,426 trees were measured for diameter in the
NFI. The diameter distributions for broadleaves follows a log-linear pattern in which there are many
more small trees than large. This pattern is expected but to some extent influenced by the sample
plot design, where larger trees have a higher probability of being measured.
Figure 3.2. Diameter distribution for broadleaves and conifers based on the 2018-2022 rotation of measurements with
the Dansh NFI. Note the log-scale on the y-axis.
3.3 Projections of Danish forest carbon pools and emissions with the EFISCEN-
Space model system
In EFISCEN-space, the plot specific diameter measurements and species registrations are expanded
to a per hectare diameter distribution used to initialize the projection. Owing to the design of the
Danish National Forest Inventory, where trees with a diameter at breast height of less than 10 cm
are only measured in the inner 3.5 m radius circle, including plots with only partial forest cover
would result in diameter distributions lacking smaller trees. Therefore, we only included plots
where the plot centre had forest cover in the simulations.
18
While the National Forest Inventory data was used for initializing the EFISCEN-Space model, it
was also used for training the individual modelling components including harvest and natural
mortality probability models.
Although we adopted the core EFISCEN-Space model unchanged, the model was executed through
a SAS (Statistical Analysis Software) script that enabled the model to run repeatedly in 5-year
cycles (Figure 3.3). This enabled us to add forest area to the model runs every 5 years to account for
afforestation, as well as apply changing management scenarios over time as described for the state
forest below. The SAS script further allows the application of Danish volume and biomass
calculations to the raw output of plot development from the EFISCEN-Space model. A brief
description of key model components and specific setup for the projections in this report follows.
Figure 3.3. Brief description of the application of the EFISCEN-Space model for carbon pool projections in Denmark.
3.3.1 Growth model
EFISCEN-Space utilizes an individual-tree-based growth model to capture the dynamic
development of forest stands over time. EFISCEN-Space uses a Gompertz model, describing a
sigmoid growth pattern [13]. The Gompertz model is defined as:
𝐷𝐷(𝑡𝑡) = 𝐴𝐴 ∙ 𝑒𝑒−𝑏𝑏∙𝑒𝑒−𝑐𝑐∙𝑡𝑡
19
Where D(t) represents the diameter of a tree at time t, A is the upper asymptote, which represents
the maximum achievable diameter, b and c are parameters that influence the shape of the curve, and
e is the base of the natural logarithm. The Gompertz model and its derivatives have been widely
applied in forestry and ecology to understand and predict tree growth. The model is flexible and can
be adjusted to fit different species and environmental conditions.
The derivative of the Gompertz model describes the rate of change in diameter (or diameter growth)
at any given point in time. The model is estimated from repeated NFI tree measurements of 2.3
million trees across Europe and considers factors such as tree species, age, competition among
trees, and local environmental conditions such as temperature and precipitation to simulate the
annual tree diameter growth [13]. At the time of the present study, the growth model had not been
fitted including Danish data and the simulations relied on the breadth of data collected across
Europe.
Figure 3.4. Examples of forest growth curves with the derivative of the Gompertz model. Note that the modelled growth
is specific for each plot location, species, stand conditions, and diameter class. In this case, modelling is made for a
stand basal area of 20 sq. m and for a tree that resembles the stand quadratic diameter tree. Also note that the growth
pattern her for Betula sp. and Quercus robur are very similar and that the two growth curves cannot be distinguished in
the graph.
3.3.2 Harvest probability
EFISCEN-Space integrates a detailed harvest probability model to simulate the impact of forest
management interventions. To this end, a matrix specifies annual species and diameter class-
specific harvest probabilities. These may be user specified, modelled, or simply extracted from
repeated NFI measurements to reflect observed patterns (Figure 3.7). Harvest probabilities may be
specified for individual plots reflecting e.g. geographical differences or may be generic across all
parts of the country.
20
Harvest probability is commonly affected by tree species, size, and age as well as by overall
management objectives and intensity. The harvest model incorporates the impact of these activities
on individual trees, assessing their susceptibility to removal based on size, species, and management
intensity. The harvest probabilities may be defined based on management rules, estimated based on
statistical analysis or may simply be included as a species and size class specific harvest probability
observed from repeated NFI measurements.
For this report, we extracted national harvesting probabilities from the repeated measurements in
the Danish NFI (2002-2022) to represent the historical harvesting probabilities for each species
(Figure 3.5). The harvesting probabilities show some erratic and much fluctuating patterns,
reflecting the often-few observations especially in large diameter classes. In this study, we opted to
use the observed probabilities directly when making the projections. This is something we could
have modelled to produce smooth curves, but we opted to keep the projections as close to actual
data as possible.
Figure 3.5. Observed annual harvest probability curves of the most common broadleaf and conifer species and species
groups in the Danish forests. Harvest probabilities for large diameter classes, depending on species, are based on
expert judgement owing to the lack of observations in the data.
3.3.3 Designation of areas for nature protection
Different probability matrices may be applied to individual plots to account for different
management scenarios. In this way, the model may account for differences in ownership, regulatory
constraints, and conservation objectives, which affect the likelihood of harvesting in specific areas.
For the relatively short period of the projections, the future harvest will largely be determined by
the current forest structure and trees already present in the forest, and it is unlikely that overall
management priorities will change in the private sector. When projecting future forest
21
management, we therefore applied the assumption that management and hence harvesting
probabilities in privately-owned forests will remain similar over the projection period.
However, the case is different for the state-owned forest, where under current policy there will be
significant changes in the management of much of the forest area. The plans to designate large parts
of the forest to be unmanaged (or managed without wood production) should be accounted for in
the model, as the conversion of these areas will occur during the period of our projections. This
includes areas planned to be set-aside as Unmanaged Forest (Urørt skov) as well as areas designated
to be Nature National Parks (Naturnationalparker). After consultation with the Danish Nature
Agency, a future management matrix was developed for the designated areas within the state
forests. This enabled us to specify harvesting probabilities for designated areas for the next 25 years
while accounting for the differing timelines for the transition of the designated areas between the
western and eastern part of the country (Figure 3.6).
First, areas owned by the Nature Agency were identified on a map and each NFI plot part of the
projection was assigned a category according to the specific designation. For this projection, we
identified five different categories including 1) Managed forest (i.e. not designated for nature
protection), 2) Nature National Park, east, 3) Nature National Park, west, 4) Unmanaged Forest,
east, and 5) Unmanaged Forest, west. Differences in harvest probabilities between east and west
Denmark are driven by different lengths of the transition period in the two parts of the country.
22
Figure 3.6. Regions identified by the Nature Agency to designate differing timelines for the transition of the state forests
to Nature National Parks and Unmanaged Forest between the western (blue, typically 25 years transition) and eastern
(green, typically 6 years transition).
For the entire projection period, we assumed that areas of the state forest that have not been
designated for nature protection will continue with the national historical harvesting probabilities as
earlier described. This differs from the projections made in relation to Climate projections 2023 [4]
in which a 20 pct. decrease in harvesting levels in the state forests was assumed as part the frozen
policy scenario. However, part if the effect of reduced harvesting in the state forests will be
observable in the harvesting probabilities. For the designated areas, we have implemented a matrix
as described below (Table 3.2).
In the first five-year period of the projection (from 2022 to 2027), we assumed that exotic species,
largely understood as species exotic to Denmark and Northern Europe, of 40 cm or more in
diameter at breast height (DBH) will be harvested in all set aside areas. Otherwise, we assumed
harvesting according to the historical probabilities for both the exotic conifers and all other species,
taking into account that some harvest will be made to create a desired forest structure prior to the
setting aside (e.g. gap creation, removal of undesired species, or altering understorey light
conditions).
23
In the eastern part of the country (Figure 3.6), we assume harvesting of all exotics irrespective of
their diameter in the following five-year period (2028 to 2032) and onwards. In this region, we
furthermore assume that all other harvesting ceases.
In the western part of the country, we assume a similar pattern, but the conversion period is 25
rather than 5 years. Hence, in the following five-year period (2028 to 2032) and onwards, we here
maintain a 40 cm target diameter for the exotic species and otherwise assumes continued harvesting
according to the historical probabilities. This regime continues until the final run beginning in 2043,
when all non-native conifers of all diameters are also removed in the western part of the country.
Harvest of native tree species (including Norway spruce, larch, mountain pine, and silver fir native
to northern Europe) is expected to gradually cease in the initiation phase. We therefore set the
harvesting probabilities to one quarter of the probabilities observed in the normal harvesting regime.
The harvesting ceases completely after the conversion is completed (after year 2026 in eastern
Denmark and after year 2032 in western Denmark). We are aware that in some areas, actual harvest
strategies may much different from the above described e.g. aiming at creating gaps in the forest
canopy or removing undesired tree species in specific areas. However, such specific modelling was
not possible within the current project although the EFISCEN-Space model would in principle
allow for it.
Deforestation within areas designated for nature protection
Setting-aside forest for both Nature National Parks and Unmanaged Forest has modelling
implications beyond merely the effect on harvesting probabilities described above. The
management required to prepare forest areas to be set-aside for biodiversity protection can be
extensive and commonly involves a variety of actions such as restoring natural hydrological
conditions or historical landscapes, conducting harvests to create a favourable forest structure,
veteranization of trees, and the introduction of grazing by larger animals such as cows and horses.
These measures likely heavily impact the forest carbon pools but are equally difficult to describe in
a modelling context. There is also a general lack of data regarding the potential future development
of these forest types in Denmark.
In this case, we opted to simulate the loss of forest owing to restoration of hydrological conditions
by converting 20 pct. of the Norway spruce dominated forest plots (in the state forest areas to be
set-aside) to non-forest during the conversion period (i.e. 5 years in the eastern part of the country
and 25 years in the western part of the country).
24
We furthermore assumed that 50 pct. of Pinus mugo and Abies alba dominated forest will be
converted to Atlantic heathland as part of the conversion (Table 3.2).
In the simulation, plots corresponding to the 20 and 50 pct. of the forest area respectively were
removed at random from the simulations and the trees on the plots were considered harvested,
simulating that the area was clearcut prior to flooding or other restoration of the landscape.
Table 3.2. Modelling the transition of areas designated for biodiversity conservation as Nature National parks and
Unmanaged Forests within areas owned by the Nature Agency. Areas not designated for nature conservation are
assumed to be managed according to the observed harvesting probabilities in the NFI and similar to forest not owned
by the Nature Agency. The table has been developed in close dialogue with the Danish Nature Agency. Grey shaded
areas show species where harvesting probability is reduced to ¼ of normal harvest probabilities during the conversion
period.
Eastern
Denmark*
Western
Denmark*
Deforestation
Conversion
(years)
Harvesting
probability
after
conversion
(%)
Target
diameter prior
to conversion
(cm)
Conversion
(years)
Harvesting
probability
after
conversion
(%)
Target
diameter
prior to
conversion
(cm)
%
1 Abies. sp. 5 0 40 25 0 40 50
2 Larix sp. 5 0 - 25 0 - 20
3 Picea abies 5 0 - 25 0 - 20
4 Picea
sitchensis
5 100 40 25 100 40 0
5 Pseudotsuga
menziesii
5 100 40 25 100 40 0
6 Pinus
sylvestris
5 0 - 5 0 - 0
7 Pinus nigra
and mugo
5 0 - 5 0 - 50
8 Other Pinus 5 100 40 25 100 40 0
9 Other
conifers
5 100 40 25 100 40 0
10 Betula sp. 5 0 - 5 0 - 0
11 Castanea
sativa
5 100 40 25 100 40 0
13 Fagus
sylvatica
5 0 - 5 0 - 0
14 Robinia
pseudoacacia
5 100 40 25 100 40 0
16 Quercus
robur and
petraea
5 0 - 5 0 - 0
19 Long-lived
broadleaves
5 0 - 5 0 - 0
20 Short-lived
broadleaves
5 0 - 5 0 - 0
* According to the map in Figure 3.6.
25
3.3.4 Mortality
The mortality component of EFISCEN-Space accounts for natural causes of tree death. In a setup
reflecting current management practices, a matrix specifies species and diameter class-specific
mortality probabilities much in the same way as the harvest probabilities. These may be user
specified, modelled, or simply extracted from repeated NFI measurements to reflect observed
patterns (Figure 3.7). Mortality matrices may be specified for individual plots reflecting e.g.
geographical differences or may be generic across all parts of the country.
To simulate the development of the Danish forests, natural mortalities were extracted from repeated
measurements in the Danish NFI (2002-2022) and used to derive historical annual mortalities for
each species (Figure 3.7). These mortalities reflect current forest practices and can be considered to
reflect minor changes in abiotic factors likely occurring for a relatively short projection period. In
initial runs of the model, we opted to use the observed mortalities from the Danish NFI as the basis
for our projections. These were manually adjusted by expert opinion where there were a limited
number of observations for mortalities of a given species and diameter.
Figure 3.7. Observed mortality curves of the most common broadleaf and conifer species and species groups in the
Danish forests. Mortalities for large diameter classes, depending on species, are based on expert judgement owing to
the lack of observations in the data.
Later adjustments of the underlying assumptions of forest management on set aside areas for
biodiversity protection in the areas owned by the Nature Agency is likely to alter the mortality
owing to low or absent harvest affecting stocking severely within the projection period. To
accommodate the likely increase in competition resulting from reduce harvesting, we opted to use
the inbuild dynamic mortality functions in the EFISCEN-Space model [14] in the final simulations.
These models include density parameters and the result of one-sided competition from trees larger
26
than the subject trees and may therefore better accommodate simulations with altered basic
conditions.
3.3.5 Afforestation
In the EFISCEN-Space model, afforestation can be simulated implicitly in the model, using the
included ingrowth module to produce growth on empty plots. However, to accommodate a more
realistic and data driven representation of afforestation, we simulated the afforestation through
imputation of sample plots corresponding to the desired afforestation area at the end of every 5-year
rotation of the model. Imputed plots were selected at random from a sample of 388 reference
afforestation plots less than 10 years old identified from the NFI data.
The selection of imputation samples is conducted via unrestricted simple random sampling with
replacement from the reference afforestation plots. The number of sampled plots correspond to the
anticipated afforestation divided by the area represented by each NFI sample plot (~100 ha). The
imputation involves allocating these selected plots onto non-forested NFI sample plots of the 2018-
2022 rotation of measurements. Imputation was only allowed on plot locations where afforestation
is desired according to municipality plans. The imputation process accounts for both spatial
allocation and, if relevant, temporal dynamics, ensuring a robust representation of afforestation
scenarios in the model.
Historical afforestation levels were determined as the sum of afforestation (both regular forest and
Christmas trees) subtracted the annual deforestation reported in the national inventory report [15].
Historical afforestation from the national greenhouse gas inventory differs from then estimates
obtained in the National Forest Inventory [12] owing to statistical uncertainty as well as due to a
need in the inventory reporting to match the forest area with other land-use types. Hence, the
national greenhouse gas inventory to a larger degree relies on cadastral records rather than actual
observations leading to slightly different forest definitions in the two inventories. For simulations
onwards, we used a frozen policy scenario for the afforestation supplied by the Environmental
Protection Agency (Table 3.3).
27
Table 3.3. Historical and projected afforestation levels based on a frozen policy approach. For the historical
afforestation, we collected all private and public afforestation in the column “Private/Total” as the figures are derived
from the land-use matrix underlying the emissions reporting in which the ownership is unknown.
Year Public Private/Total Climate
forest fund
Year Public Private Climate
forest fund
ha
2002 4110 2025 270 2570 700
2003 4110 2026 270 2570 660
2004 4110 2027 270 2570 870
2005 3372 2028 210 2570 820
2006 3372 2029 210 2570 1000
2007 3372 2030 0 880 1200
2008 3372 2031 0 880 0
2009 3372 2032 0 880 0
2010 3372 2033 0 0 0
2011 3372 2034 0 0 0
2012 1537 2035 0 0 0
2013 4641 2036 0 0 0
2014 364 2037 0 0 0
2015 2115 2038 0 0 0
2016 678 2039 0 0 0
2017 1107 2040 0 0 0
2018 1130 2041 0 0 0
2019 1248 2042 0 0 0
2020 1800 2043 0 0 0
2021 4887 2044 0 0 0
2022 1485 2045 0 0 0
2023 280 2000 100 2046 0 0 0
2024 300 2000 480 2047 0 0 0
3.3.6 Ingrowth and reforestation
The data from the NFI includes areas that have been recently harvested and are temporarily
unstocked which are likely to become re-stocked with trees. Furthermore, EFISCEN-Space
produces empty plots when the simulation results in all trees on a plot becoming dead or harvested.
As EFISCEN-Space utilizes an observed diameter distribution to produce the projection, plots with
no trees cannot be projected into the future unless the plot is populated with new trees.
The problem is similar to introducing afforestation as afforested plots initially have no trees to be
projected into the future. An option would be to use a similar approach as for afforestation,
populating the plots using imputation from known reforested plots, or otherwise specifying the
28
number of replanted trees of a certain species. As it is difficult to make realistic assumptions on
future species composition reflecting local conditions, we opted for the default EFISCEN-Space
process which regrows the species most recently present on the sample plot. In cases where there
are no trees present on the plot at the initialization of the model, the model re-populates the plot
with a “short-lived broadleaves” species group (such as rowan, birch, and aspen).
3.3.7 Dead wood
The EFISCEN-space model outputs species and diameter distributions of trees dying in the
projection period. Although it would be possible to estimate the inflow of dead wood to the carbon
pool, the outflow in terms of degrading wood is largely unknown. As the dead wood pool is
relatively minor to the other forest biomass pools, we considered it outside the scope of this project
to attempt advanced modelling of this aspect of forest carbon dynamics. We therefore assumed an
unchanged level of dead wood, well aware that the activities in the set aside forests will likely
increase the amount of dead wood locally, but also expecting that the overall effect on the forest
carbon pool will be relatively minor.
3.3.8 Litter
The YASSO-model tailored for modelling soil carbon pool development may run within EFISCEN-
space providing estimates of forest litter carbon pool development. However, we found that the time
was too limited to test the results for Danish conditions. Instead, we opted to use a constant litter
pool owing to the short projection horizon.
3.3.9 Harvested wood products (HWP)
The EFISCEN-space model outputs species and diameter distribution of harvested trees. In this
projection, we estimated the biomass in harvested trees in the same way as estimating carbon stocks
in live biomass using species specific d/h-functions and national biomass functions [16].
Recognizing that contemporary forest management often involves harvesting of the entire above
ground biomass (including branches), we estimated biomass in harvested timber from the projected
above ground biomass and the share of timber (46.8 pct. for conifers, 14.1 pct. for broadleaves) in
the national harvest statistics reported by Statistics Denmark. The inflow of biomass in HWP was
subsequently calculated from the cutting yield observed in Danish sawmills (42.3 pct. for conifers,
42.4 pct. for broadleaves). As for previous projections, we did not make assumptions on exported or
imported quantities of round-wood, which are not reported in the sourcing country while the sawn
29
volumes are part of the HWP pool. For the period 2020-2023, imported and exported amounts were
reported at 224,000 and 137,000 tons, respectively. Given that these figures are reported in kg’s by
the importing party with an unknown moisture content, the uncertainty involved in making more
detailed assumptions was evaluated to be prohibitively high.
Emissions from HWP were estimated using the methodology developed for the national greenhouse
gas reporting using the inflow of biomass from the EFISCEN-space model and previously
determined product half-lives to determine the outflow from this pool.
4 Results
The model predictions of forest carbon stocks in above-ground biomass, below-ground biomass,
and total biomass showed a continuous uptake of CO2 in the forests during the entire projection
(Figure 4.1). Mortality started slightly higher than in the later parts of the projections but declined in
the first simulation period to around 1.5 mi. tons CO2-eqv.
Harvesting levels peaked at ~6 mi. tons CO2-eqv. during the first five-year simulation cycle and
hereafter declined to an ultimate low at ~4 mi. tons CO2-eqv. During the remainder of the
simulations, harvest levels are projected to increase, reaching a level similar to the levels observed
in the first cycle of simulations. The harvest fluctuates across the years, likely owing to spikes in the
harvest probabilities as well as in the diameter distribution. However, the projections show a cyclic
pattern which is largely caused by the way the EFISCEN-Space is set up. As the model does not
allow harvesting of the same plot twice in each 5-year run, a random parameter assigns a time for
harvesting each plot. This time (from the beginning of each cycle) is repeated on each rotation of
the model causing a cyclic pattern. This pattern is further exaggerated by the five-year rotation of
the model, where afforestation is added to the forest area and deforestation is simulated by
harvesting and removing plots at the beginning of each rotation.
The resulting projected emissions indicated a continuous uptake in the forest, that was relatively
low in the first 5-year period owing to the larger harvests projected in this period. Later, emissions
projections total an average CO2-uptake ranging between 3 and 3.5 mi. CO2-eqv. slightly increasing
as a result of the increasing forest area.
30
Figure 4.1. Carbon stocks, mortality, harvest, and emissions expressed in CO2-eq. Projections are made from NFI data
collected in 2018-2022 forming the 2022 base line.
In general, the broadleaves delivered a net-uptake resulting in negative emissions with the largest
uptake in beech and long-lived broadleaves such as sycamore maple and cherry (Figure 4.2).
Oppositely, many of the conifer species had near zero emissions and in some notable cases even
substantial emissions for Sitka spruce and Norway spruce.
Figure 4.2. CO2 emissions distributed to species and species groups Left: broadleaves, Right: conifers.
31
The levels and fluctuations of CO2 emissions are comparable with historically reported emissions
[15] replicating the current peak in emissions and subsequently declining to a credible net uptake of
1.5-3.0 mi. tons CO2 in above and below ground biomass (Figure 4.3). The overall fluctuations are
at a similar magnitude to what is observed in the reported figures, but the annual fluctuations seem
somewhat more erratic owing to the previously described harvesting patterns in EFISCEN-space.
However, it should be noted that in the emissions reporting, figures are smoothened as 5-year
averages and reporting of individual years, such as in the projected numbers in (Figure 4.3) would
be expected to show more variation.
Figure 4.3. Reported ([15], full line) and projected (dashed line) emissions from above-ground (green), below-ground
(blue), and total biomass (red). Left: Projection from a NFI 2022 baseline, Right: Projection from a NFI 2012 baseline.
4.1 Effects of forest management and programme settings
To illustrate the effect of changes to the model input, we changed a number of parameters in the
model settings to illustrate 1) the choice of static vs. dynamic mortalities in the model, 2) the effect
of setting aside forest for biodiversity protection, and 3) the effect of afforestation.
Static vs. dynamic mortalities
In the basic setting, mortalities were projected using the dynamic functions included in the
EFISCEN-Space model and estimated on a pan-European dataset. Realizing that the underlying
model was estimated absent of Danish data, we intended to analyse how this choice affected the
results, by making the projections with a static mortality observed directly from Danish NFI data
(Figure 3.7). Our analyses show that the static and dynamic mortalities produce quite different
mortality levels (Figure 4.4) reflecting that the static model does not adjust to altered forest
conditions e.g. when forest is designated for biodiversity conservation or standing stocks are altered
32
owing to projected harvesting. However, differences in mortalities arising from the choice of model
had little effect on overall emissions EFISCEN-Space (Figure 4.4), reflecting the relative minor
importance of dead wood in the overall carbon budget.
Figure 4.4. Carbon stocks, harvest, mortality, and associated emissions for a scenario using a static mortality function
(dashed lines) with the reference scenario using the dynamic mortality function (solid lines) (Figure 4.1).
Setting aside forest for biodiversity protection
When simulating the effect of setting aside forest for biodiversity protection, we made a
counterfactual setting of the model to harvest trees in accordance with the historical probabilities,
although realizing that these observations to some extent includes ongoing conversion of set aside
forests. The simulations indicate a very minor change in carbon pool development and associated
emissions from biomass carbon pools compared to the basic settings (Figure 4.5, Table 4.1). In the
first, 5-year period, harvesting levels were similar for the counterfactual and the reference scenario.
This is likely the result of contrasting effects. On the one hand, reduced harvesting of particularly
native broadleaves increased carbon pools on parts of the forest area, while deforestation and
removal of exotic conifers as the result of nature restoration leads to reduction of carbon pools in
other parts. The setting aside of forest for biodiversity, however, significantly altered the harvesting
33
levels in later rotations of the simulations reflecting that no harvesting is conducted on the ~75.000
hectares set aside.
Figure 4.5. Projected forest carbon pools, harvesting, mortality, and associated emissions from a scenario assuming no
designation of forest for biodiversity conservation (dashed lines) compared with the reference scenario (solid lines)
(Figure 4.1).
Afforestation
To isolate the effect afforestation, we maintained all model settings to be similar to the basic setting
(Figure 4.2) including the setting aside forest for biodiversity protection. The simulations resulted in
a total forest loss of 4,566 ha during the simulations owing to the deforestation occurring on the set
aside forest. Compared to the standard settings, the resulting forest area at the end of the simulations
was 28,565 ha or 4.3 pct. lower. Considering that the afforestation will have comparably low
biomass, it is no real surprise that the no afforestation scenario only had slightly lower carbon pools
and hence also slightly higher emissions than the standard scenario (Figure 4.5, Table 4.1).
34
Figure 4.6. Carbon stocks, harvests, mortality, and associated emissions for a scenario with no afforestation (dashed
lines) and a comparison with the reference scenario (solid lines) (Figure 4.1).
Table 4.1. Forest area, carbon stocks, and emissions for the reference scenario and for the two scenarios where no
areas is set aside for biodiversity conservation and where no afforestation is carried out. Figures are provided for 5-
year averages with the initial year (2022) as the overall reference.
Forest area Biomass carbon stocks Carbon emissions Harvest Mortality
Scenario year
Above
ground
Below
ground
Above
ground
Below
ground
Above
ground
Below
ground
Above
ground
Below
ground
ha 1,000 tCO2-eq
Reference 2022 642,976 138,339 30,441 - - - - - -
2023-2027 642,976 141,330 31,005 -913 -170 4,354 1,000 1,419 361
2028-2032 656,222 149,438 32,703 -2,220 -481 3,484 805 1,250 299
2033-2037 667,556 162,200 35,501 -2,636 -576 3,670 853 1,243 287
2038-2042 667,825 175,630 38,425 -2,720 -590 3,899 909 1,271 286
2043-2047 666,975 188,995 41,300 -2,597 -555 4,144 967 1,304 287
No setting aside for
biodiversity
2022 642,976 138,339 30,441 - - - - - -
2023-2027 642,976 141,251 30,997 -881 -167 4,384 1,004 1,420 361
2028-2032 658,876 149,139 32,667 -2,093 -455 3,732 860 1,251 300
2033-2037 671,060 160,614 35,184 -2,361 -519 3,974 918 1,231 285
2038-2042 671,754 172,544 37,785 -2,394 -520 4,206 977 1,245 281
2043-2047 671,754 184,339 40,326 -2,304 -490 4,420 1,030 1,265 279
No afforestation 2022 642,976 138,339 30,441 - - - - - -
2023-2027 642,976 141,330 31,005 -913 -170 4,354 1,000 1,419 361
2028-2032 640,640 149,404 32,706 -2,191 -476 3,482 805 1,245 298
2033-2037 640,003 161,474 35,350 -2,492 -545 3,645 849 1,227 283
2038-2042 639,153 174,173 38,106 -2,532 -551 3,829 893 1,244 279
2043-2047 638,410 186,732 40,825 -2,432 -519 4,055 949 1,269 279
35
5 Discussion
5.1 Forest carbon projections
The projection indicated in increasing carbon pools corresponding to the levels observed in recent
years [12, 17, 18]. The increase in forest carbon pools was less in the beginning of the projections,
corresponding to an initial peak in emissions in the first five-year period, likely as a result of several
different factors.
Firstly, as evidenced in national reporting on forest statistics [12, 17, 18], the age- and diameter
distribution of trees in Danish forests is skewed (Figure 5.1) with large quantities of mature trees
with a high probability of being harvested according to the historical harvesting probabilities used
in the projection (3.3.2 Harvest probability). This is particularly pronounced for species such as
beech, where prices have been low for several decades, resulting in a build-up of the volume of
mature trees. Consequently, according to national forest statistics (recalculation of figures presented
in [12]), more than 1/3 of the CO2-eq in beech is found in trees with a diameter (measured at breast
height) of more than 60 cm, which would generally be considered mature. Similarly, about 1/4 of
the CO2-eq in Sitka spruce is found in mature trees with a breast height diameter of more than 40
cm.
Recent increase in prices of both broadleaf and conifer timber as well as a favourable market for
forest fuels have resulted in increased harvest levels reflected both in our projections and in the
reported harvesting levels from Statistics Denmark [19]. Also, the conversion of areas set aside for
biodiversity protection in the state forests, which involves clearing of exotic tree species as
simulated in the projection, albeit to a minor degree affects overall harvesting levels and carbon
pools (Figure 4.5).
36
Figure 5.1. Carbon pools in above and below ground biomass for broadleaves and conifers. The red vertical line
indicates approximate maturity for most species in the two categories.
In an attempt to compare the harvesting levels reported from Statistics Denmark [19] to our
projected harvesting, we converted the volumes reported from Statistics Denmark to CO2-
equivalents, using a basic density of 0.55 ton biomass/m3
for broadleaves and 0.38 ton biomass/m3
or conifers, a carbon density of 0.47 gC/g, and a conversion from carbon to CO2 of 44/12. Realizing
that we have no knowledge on the extracted fraction of the total harvest reported to Statistics
Denmark, we compared the reported harvesting expressed in CO2-equivalents to our projected
above-ground biomass. Our results indicated that the projected harvesting is similar to that reported
by Statistics Denmark in recent years, indicating a similar trend (Figure 5.2) albeit a slightly higher
peak in the harvests in the coming years. It should be noted that we expect the projected harvests to
be systematically higher than the values reported by Statistics Danmark, since the projected values
include all above ground parts of the tree, whereas only parts of these are expected to be extracted
and reported to Statistics Denmark.
Whether the projected peak in harvest levels and therefore also in emissions will be observed in the
coming years depends largely on the future price structure of wood products and bioenergy, which
is not reflected in the model. A special challenge to this end is that the NFI data used as the baseline
for the projections was collected during 2018-2022, meaning that some of the trees measured in
2018 and 2019 may have been harvested during the increased harvest in the last years of this period.
Hence, some of the peak observed may in fact reflect harvesting that has already commenced. To
what extent is not possible to say.
37
Figure 5.2. Comparison of projected and reported (Statistics Denmark) harvesting levels expressed in CO2-equivalents.
The recalculating of reported volumes to CO2-equivalents is highly uncertain and the ratio of extracted (and sold)
volumes to total harvest volumes is unknown. Therefore, the direct comparison on projected and reported volumes is
uncertain.
Owing to the large amount of mature beech the model projects increased harvest levels in beech and
hence less uptake of CO2 in the coming years. Another important finding is that we project a
significant net uptake of carbon in oak. This is presumably due to the extensive use of oak in
afforestation projects in recent decades [12], ensuring a large net uptake in the young forests a long
way from maturity and final felling.
Interestingly the model projects only limited net uptakes in conifers and even periodic emissions
from Sitka spruce and Norway spruce, owing to annual harvests and mortality exceeding increment.
This may have several reasons including good prices on softwood timber in recent years, increasing
the harvest probabilities and therefore also the predicted harvest levels. However, increasing health
problems for these two species in particular caused by extended periods of drought during the
summer, pest such as bark beetles and aphids, and windthrow may also have impacted both the
mortality of the species but also the forest owner’s decision to harvest the two species earlier.
In general, the model provided credible projections of forest carbon pool development and
associated emissions. In particular, the model projected similar patterns of emissions as have been
observed from the national estimates based on forest inventory data.
38
5.1.1 Forest growth models
In the current project, the EFISCEN-space model was for the first time parameterized to Danish
conditions, including the setting of harvest and mortality probabilities, and adjusting to local
conditions such as expected afforestation and specific management of forest designated to
protection of biodiversity. During the project, we further attempted to reparametrize the underlying
growth model with observations of tree growth from the Danish National Forest Inventory.
However, we found that despite of the rich data available (about 70,000 trees with repeated
measurements were included), the modelling seemed less robust compared to the in-build growth
model relying on 2.3 mi. trees with repeated measurements observed from a wide range of
geographical conditions across Europe [13].
To enhance the accuracy of EFISCEN-Space model predictions for Danish forest biomass pools,
reparametrizing the model with Danish data is essential. However, as a simple fitting of the models
with Danish data proved insufficient, this entails a full recalibration of the growth models
underlying EFISCEN-Space with the pan-European and Danish data. Such an effort includes also
validation of the models using independent datasets and sensitivity analysis to ensure the reliability
of the reparametrized model. Such an effort was not possible within the current project.
When reviewing the growth patterns simulated in the current version of the EFISCEN-Space model,
we found that some of the models produces simulations inconsistent with current knowledge of tree
growth, such as unlikely late peaking growth or even growth not peaking at all and excessive
growth levels under low or high competition. A likely reason is that although the underlying data
collected from National Forest Inventories across most of Europe has an impressive breadth and
depth, the vast majority originates from forests managed according to some similar standards. This
results in well behaved functional forms under standard conditions but less so when conditions
deviate from the normal. We speculate that data from forest experiments, typically including
deviating forest management and long time series could substantially improve the growth functions
in EFISCEN-Space.
5.1.2 Uncertainties
The projection of forest carbon pools entails a wealth of uncertainties of which we may here only
describe a few.
39
Input data
The input data is measured with a very low uncertainty but represents a sample of the Danish forest
area. Earlier studies have demonstrated that the uncertainty of forest carbon pool estimates are small
(0.9 pct.) but also demonstrated that even a proportionally small uncertainty may have a large
effects when applied to large pools. With more than 160 mi. tons CO2-eq stored in the biomass, the
uncertainty expressed as the standard error is around 1.5 mi. tons CO2-eq. As the uncertainty of
projections presented here will always be larger than the direct estimates from actual measurements
a numerically substantial uncertainty should be expected. In particular when considering that
emissions are calculated as the difference between two subsequent and uncertain estimates of forest
carbon pools.
Natural catastrophes
The model used in this study assumes harvesting probabilities and natural mortalities to follow
some previously observed patterns. However, climate change is projected to result in warmer
summers with more frequent droughts, winters with more precipitation and more frequent flooding,
as well as more frequent and heavier storms. It is thus likely that mortality patterns will change
during the projection period as it has been observed in southern and central Europe.
Specifically for Denmark and building on historical observations, it is far from unlikely that we will
see catastrophic windthrow one or more times during the projection period. In the largest-ever
windthrow observed in Denmark, 3.6 million cubic meters of wood were windthrown
corresponding to a similar number expressed in ton CO2-eq. Such a windthrow would significantly
alter the reported emissions from the LULUCF-sector depending on assement method for reporting.
Changed growing conditions and climate change
As stated in the methods section, we opted to use the currently available growth models obtained
from repeated measurements of trees on National Forest Inventory sample plots. However, climate
change is currently altering the conditions for forest growth in Europe [20] and may impact also the
growth of Danish forests. Effects of climate change are expected to be more elaborate in extreme
latitudes and altitudes. Some tree species may increase vitality and growth at higher boreal latitudes
or higher altitudes and the opposite at lower dry and warm locations [21, 22]. Regional growth
trends are less clear in areas currently better suited for tree growth and recent studies report overall
increasing growth trends for European trees [23] and forests [20]. However, despite this general
40
pattern, severe drought events and generally changed precipitation patterns in some regions result in
declining vitality and growth of some of the most abundant European forest tree species [24-27].
The cumulative effect of changed temperature and precipitation patterns in Denmark is unknown,
but the EFISCEN-Space model holds the possibility to alter the underlying growth drivers by
changing model factors to simulate growing conditions currently native to other parts of Europe.
However, with the relatively short projection scope used in this study (25 years), we found that
climate change and its effect on forest growth during this period would likely be moderate and
opted to use current climate conditions in the model.
Human behaviour
In the projections, we have assumed that human behaviour related to harvesting of trees follows
historical patterns. This is a far from likely assumption when realizing that societal changes may
heavily affect the way we use the forests. As an example, a change in the Chinese market for beech
wood in combination with heavy windthrow in central Europe caused an abrupt decline in demand
and more than halved the price of beech wood around year 2000. The prices have so far not
recovered entirely and the change in prices has for more than 20 years reduced the harvesting of
beech substantially. As such the harvest is currently around half of what was reported to Statistics
Denmark in 1990-1999 and only on third of the reported figures in 1960-1970. Oppositely, the
breakout of the war in Ukraine in February 2022 caused a massive increase in energy prices causing
an enormous demand for firewood. Although the price change was only temporal and therefore had
limited impact, the harvesting of firewood went up in the forest likely having an impact on forest
carbon pools.
The forests have many other functions aside from their climate change mitigation potential. Aside
from the changes in price and market dynamics, their treatment may be affected by other desires
regarding the services they provide. A notable example is the setting aside forest for biodiversity
protection. As also reflected in this study, the Danish government in 2020 decided to set aside
75,000 ha of forest for conservation purposes. This drastically changed the forest management on
more than 10 pct. of the forest area. Future political goals therefore have the potential to introduce
even more drastic changes to future forest management and hence to the development of the forest
carbon pools and their climate change mitigation potential.
41
Carbon pool development on set aside forest areas
Albeit we succeeded in differentiating management of the forest resource according to geographical
data on set aside forest in eastern and western Denmark, very little knowledge is available on the
actual implications of the management on forest carbon pools.
We assumed that a proportion of the area will be deforested but have no actual knowledge if the
area is correct or whether the areas will grow into forest again. In particular, a portion of the set
aside area will be rewetted and although we have assumed that this will not impact emissions of
other green house gasses such as nitrous oxide and methane, we know that rewetting might affect
the emissions of these very potent climate gasses.
We have assumed limited harvesting of the areas but have no knowledge on the possible reduction
in forest biomass owing to the introduction of large grazers (horses and cattle) in particular in the
nature national parks. It is likely that the animals will both damage trees and hence affect the
present carbon pools while also hampering regeneration of the forest and in time cause substantial
deforestation in sensu climate reporting.
5.2 Connection with greenhouse gas reporting
To enable comparison of reported and projected values possible, we entered the projection results
into the reporting tool routinely used for making emissions estimates based on carbon pool
estimates from the national forest inventory. The reporting tool calculates emissions as the average
annual differences between 5-year moving averages of the forest carbon pool estimates to alleviate
significant inter-annual fluctuations in emissions resulting from overlapping cycles of the national
forest inventory [28]. Consequently, the reporting tool is well suited to flatten the cyclic emission
pattern resulting from the EFISCEN-space model repetition of harvesting cycles, which is an
artefact of the model setup rather than reflecting overall model trends.
When merging reported and projected emissions it should be noted that there are prominent
methodological differences in the calculations of carbon pools between the reporting on one hand
and the projection on the other. Firstly, the height of individual trees that is used in the tree biomass
estimation is largely estimated from local diameter/height regression specific to the individual plots
[11]. When making the calculations from the EFISCEN-Space output it is not possible to produce
localised diameter/height regressions and we used a set of general, albeit species specific, equations
estimated from the national forest inventory data. Secondly, to allow for the scaling also of small
42
trees in plots covering more than one land use, we included only sample plots with the plot centre
covered by forest in the projections. To accommodate for the difference in statistical design, we
used a consistent estimator assuming full forest cover of included plots and zero cover for plots not
included. However, although the estimator is consistent, one should not expect numerically
identical results. For the initialization of the EFISCEN-space model, we estimated the above ground
biomass carbon pool at 37,728.71 kt C whereas the estimate provided from the usual calculation
used for the reporting estimated 36,206.40 kt C – a difference of 4 pct.
The difference between the most recent reported value of forest carbon stocks and the first year of
the projection makes coupling of the reporting and projection difficult as the difference will result
in technical emissions or uptake, not resulting from forest growth or management but from a basic
difference in statistical design. We therefore made a technical correction to the reporting tool by
subtracting the difference between observed and projected carbon pools in year 2022 that
constitutes the last year in the reporting and first year of the projection (37,728.71 kt C - 36,206.40
kt C ~1,500 kt C for above ground biomass) respectively, hereby alleviating the effect of the
difference in sampling scheme between the NFI data and the data used in the projections. This
correction has no implications for the projected emissions except for the first year of the
projections, where observed and projected carbon pools are linked. The observed pattern now
shows fluctuations that are not different in magnitude from historical observations (Figure 5.3).
Figure 5.3. Observed (solid lines) and projected (dashed lines) emissions connected by the normal reporting tool.
43
5.2.1 Comparison to previous projections
The methods applied in this projection differ largely from previous projections in many aspects.
Previous projections relied on an age-class based approach in which the underlying assumption in
both model estimation and application was the harvesting of entire forest stands rather than
individual trees. Oppositely, EFISCEN-Space relies on an individual tree-based approach, in which
individual tree development is projected into the future. Here, the model estimation and application
relied on individual tree observations on national forest inventory plots. Based on an analysis of the
National Forest Inventory data, we found that the approach used in this study is much closer to the
actual management of the Danish forests and also that the model makes a more direct use of the
available data.
In this study, we compared the Forest Carbon Pool Projections 2022 (KF22) with the current
Forest Carbon Pool Projections 2024 (KF24) (Figure 5.4). The differences between methods, and
not least in the underlying models, are expected to result in differences between previous and
present projections. However, there are also differences in the underlying frozen policy scenario,
that should be observed when comparing the two projections. Such differences include differences
in the size and composition of afforestation, the pace of implementation of set-aside forest for
biodiversity protection, as well as the expected composition and growth of forest regeneration.
The KF22 projected a decline in forest stocks at the onset of the projections (dashed red line in
Figure 5.4) owing to an initial increase in harvesting. This was presumably caused by a skewed age
class distribution, in particular for beech where low prices for decades had resulted in a buildup of
large resources of mature trees. Notably, the current projections (KF24) predict a similar initial
increase in harvests similar to but not of the same magnitude as projected in KF22. Oppositely, the
recovery of net carbon uptake is faster in the current projections, seemingly owing to the use of a
tree-based approach rather than the previous approach assuming harvest of entire stands. The
observed difference may likely be attributed to the modelling of individual trees, that allow for a
gradual turnover of the stand as is normal practice in particular in beech in Denmark, but likely also
in the stability of the underlying modelling framework. In previous modelling efforts, the Markov
chain models were built upon statistical modelling of the chance of forest transition from one age-
class to the next and the associated chance that the forest is converted to the youngest age-class (age
0). Owing to the scarce number of incidents where such conversion took place, and the common
conversion of forest to entirely different age-classes made such modelling difficult and uncertain.
44
This in turn likely results in less certain projections of emissions from forests and is likely
responsible for much of the difference observed between current and previous projections.
Collectively, when comparing the 5-year moving averages produced by the climate reporting tool
this leads to differences in emissions projections from live biomass between KF22 and KF24
totalling 1.7 mi. tCO2-eq/yr in 2030 and 3.1 mi. tCO2-eq/yr in 2040 (Figure 5.4, Table 5.1 labelled
“MA”). Since the moving averages in the climate reporting tool utilize data from two consecutive 5-
year periods, the resulting emissions are affected by previous reported/projected emissions. If
instead considering the periodic averages (i.e. average emissions of 2023-2027, 2028-2032, 2033-
2037, 2038-2042, and 2043-2047) differences in emissions totalled 2.5 mi. tCO2-eq/yr in 2030 and
3.2 mi. tCO2-eq/yr in 2040 (Table 5.1 labelled “PA”). In particular the peak in emissions in the first
5-year period of the projection (2023-2027) affects the moving average the following years and
hence the 2030 estimate. Subsequently, the stable level of emissions from 2028 and onwards leads
to lesser differences between the moving and periodic averages.
As the project evolving around Forest Carbon Pool Projections 2024 was at the initiation meant to
entail only a simple projection, we opted to assume no change in the dead wood and litter layer
carbon pools. This assumption was based on the relatively minor size of these two pools and their
commonly slow change, collectively resulting in only minor contribution to the annual emissions.
As a consequence of this assumption, it is not meaningful to compare emissions from dead wood
and litter between the Forest Carbon Pool Projections 2022 and Forest Carbon Pool Projections
2024.
45
Figure 5.4. Comparison of projections of forest emissions. Here we compare the KF22 (total emissions, dashed red
line) projection and the present KF24 projection (from 2022, solid red line). Prior to 2022 (marked with a vertical
dashed line), lines show the reported emissions.
Table 5.1. Comparison of emissions from live biomass (above and below ground) for the forest carbon pool projections
Klimafremskrivning 2022 and Klimafremskrivning 2024. Climate projections 2024 (MA) represents the moving
averages depicted in Figure 5.4; Climate projections 2024 (PA) represents simple 5-year averages of the emissions
depicted in Figure 4.1.
Year Climate projections 2022 Climate projections 2024 (MA) Climate projections 2024 (PA)
Above
ground
Below
ground
Total Above
ground
Below
ground
Total Above
ground
Below
ground
Total
1,000 tCO2-eq
2025 119 88 207 -1,815 -352 -2,167 -913 -170 -1,084
2030 -181 -35 -216 -1,645 -281 -1,925 -2,220 -481 -2,701
2035 -305 -51 -357 -2,552 -533 -3,085 -2,636 -576 -3,212
2040 -124 -14 -138 -2,682 -566 -3,247 -2,720 -590 -3,310
2045 -247 -40 -287 -2,700 -570 -3,269 -2,597 -555 -3,152
Although not relying on species and age-class specific treatment of the forest but rather on
individual tree projections from a mere scaling of observed diameter and species distributions
observed on inventory plots, the data used in the previous Forest carbon pool projection 2022 and
the current Forest carbon pool projection 2024 are the same: data from the national forest
inventory. Nonetheless, the projections produced here differ largely from recent projections.
Importantly, however, is that both methods entail uncertainties. As explained in section 5.1.2 on
uncertainties, the uncertainty (standard error) of the biomass carbon pool estimate is around 0.9 %
46
of the total, resulting in an uncertainty of 1.5 mi. tCO2 eq. As the emissions are calculated as
differences between pools, the associated uncertainty may be estimated as the sum of the
uncertainties of the two carbon pool estimates minus their covariance. Although we do not know the
covariance, the uncertainty is likely to be even larger than the uncertainty related the estimate of
carbon pools; i.e. larger than 1.5 mi. tCO2 eq. Consequently, although the differences in carbon
emission projections may seem large, they are likely not statistically different.
5.3 Forest carbon projection methods
5.3.1 Simple projections
Initially, the assignment for this task was to produce a simple projection of forest carbon pools and
the associated emissions. It should be noted that even apparently simple projections require
underlying assumptions, which must be valid in order to justify the projections. From the beginning
of the project, it was appreciated that a simple method that captures the current diversity in forest
structure and developments in forest management does not exist. Nonetheless, to demonstrate the
possible application of a very simple projection method, we made an analysis of carbon pools and
emissions from the mere reported values from the forest inventory and the emissions reporting for
the UNFCCC. Furthermore, we intend to demonstrate that even a very simple method for projecting
carbon emissions produce estimates with significant uncertainty and hence that results should be
interpreted with care, when developing policies upon the estimates.
Projections were made using the Autoreg procedure in SAS with a 2nd order autocorrelation
(AR(2)) model estimated through maximum likelihood. The model fits a linear regression model to
the time series data, where each data point is predicted based on a linear combination of its two
most recent past values. The coefficients of this linear combination are estimated using maximum
likelihood, and the model is then used to forecast future values in the time series based on this
learned pattern.
The forecasting of the carbon pools shows an increasing trend and a relatively narrow confidence
interval (Figure 5.5). This is well in line with previous studies that the uncertainty of the live
biomass estimates is around 0.9 pct. and hence a narrow confidence interval is expected. However,
this estimate does not include uncertainties related to e.g. future afforestation, age distribution of the
forest and resulting changes to the harvest levels, or uncertainties related to changes in forest
management on set-aside areas.
47
Figure 5.5. Examples of forecasting carbon pools in living biomass using an autoregressive model on reported carbon
stocks from the Danish NFI.
When applying the autoregressive model to reported emissions, confidence intervals became very
large (Figure 5.6), owing to the large variations resulting from estimating differences between large
pools with a relatively small change (~1-2 pct.) even if the pools are estimated with a small standard
error. The figures illustrate the effect of even small uncertainties in the carbon pool estimates when
considering annual differences that make up the emissions estimates. This effect should also be kept
in mind when evaluating the more complex projections made with the EFISCEN-Space model.
Figure 5.6. Examples of simple projections using an autoregressive model (lag=2) on reported emissions data from
above-ground biomass and total biomass.
5.4 Future development
This project has made it possible to implement a new modelling framework that has the potential to
continuously refine the projection of forest carbon and other resources in the Danish forests.
However, the short timeframe of this project did not allow for implementation of all the capabilities
of the EFISCEN-Space model.
48
A roadmap of future development and testing is envisioned, both to further take advantage of the
capabilities of EFISCEN-Space and to complement the existing input data from the NFI with
auxiliary data e.g. from long-term experiments and remote sensing. An example is to further
develop new Danish growth increment functions based on data from the NFI that could be used
within the EFISCEN-Space model. These were in fact developed during this project but have not
been adequately tested to include in this report.
Some specific additions to the modelling framework should be considered in future editions. Firstly,
our analyses do not take into account possible effects of changing forest management on dead wood
carbon pools. Especially when analysing various scenarios of setting aside forest for biodiversity
protection where increasing dead wood carbon pools is expected, the current projection system
would fall short of the target. However, such an analysis would to some extent be possible, as the
model provides projections on natural mortality and hence the contribution to dead wood pools.
This would, however, require knowledge on the oxidation of the dead wood pool, commonly
expressed in half-lives of the biomass [29, 30]. Albeit this could be an important extension of our
analyses, we expect that the effect on overall carbon pools would be limited owing to the size of the
dead wood relative to the live carbon pool and the 25-year time perspective in our analyses in which
it is unlikely that any significant build-up of dead wood would occur.
In addition to the pool estimates, the current modelling framework does not take into account
emissions of other greenhouse gasses such as methane or nitrous oxide from the soil. A prominent
feature of current trends in closer-to-nature forest management and the setting aside forest land for
biodiversity protection is the reversion to natural hydrological conditions by ceased maintenance
and even destruction of ditches and drainage pipes. Inhibited soil drainage eventually leads to
wetter conditions in forest soils and to the formation of intermediate or permanently wet soils that
may affect emissions of CO2, methane, and nitrous oxide. As the latter two greenhouse gasses have
a high global warming potential, this may have significant impact on the climate effect of changed
forest management practises, as even small proportions of wet soils contribute substantially to the
emissions of methane and nitrous oxide [31].
As we have little knowledge on the actual rewetted area resulting from e.g. the setting aside of
forest for biodiversity protection in our scenarios, we referred from analyses on the consequences
on emissions of other climate gasses in our study. However, the EFISCEN-Space model has been
coupled with the YASSO soil carbon model in a way that enables outputs and plot data from
49
EFISCEN-Space to serve as inputs to YASSO (i.e., biomass inputs from mortality, harvesting, and
litterfall). Such development could enable soil carbon projection that reacts in a dynamic way to
changes in modelling future climate scenarios or forest management.
The EFISCEN-Space model is itself under continuous testing and development with a dedicated
team of software developers and forest scientists. It is planned to add functionality to e.g., explicitly
simulate plot level development under future climate scenarios as well as to consider inputs to
Harvested Wood Product (HWP) pool development from forest harvesting.
5.5 Assessing actual climate effects
In the context of assessing the climate impact of forestry activities, it is imperative to adopt a
comprehensive approach that goes beyond merely accounting of direct emissions associated with
changes in forest carbon pools as is the basis of this report. Traditional metrics including the climate
reporting often focus on the immediate carbon fluxes resulting from forest management practices,
such as carbon sequestration and emissions related to deforestation or afforestation. However, this
narrow perspective overlooks the broader climate benefits derived from the utilization of forest
products and the substitution of these products for more carbon-intensive materials and energy
sources [29, 30].
Wood harvested from forests serves as a critical input for a variety of products and energy solutions
that play a significant role in the transition towards a green economy. When wood products replace
materials that are more carbon-intensive to produce, such as concrete, steel, or plastic, there is a net
reduction in greenhouse gas emissions. This substitution effect extends to the energy sector, where
biomass sourced from sustainably managed forests can displace fossil fuels, further contributing to
the reduction of greenhouse gas emissions. The processing of wood into products and energy is
generally less energy-intensive compared to the manufacturing processes for other materials. This
results in lower emissions from the production phase, enhancing the overall climate benefit of using
wood.
It's essential to recognize that the energy requirements and emissions associated with the processing
of wood are significantly offset by the carbon storage in wood products and the substitution
benefits. However, our analyses do not account for the possible effects of forest products in total
societal emissions to a large degree occurring outside the forests. As an example, designating forest
areas for nature protection obviously results in a decline in the wood production after the
50
conversion and hence in time the inflow of wood to the HWP pool. This will result in increased net-
emissions from the HWP pool, since only nationally produced wood is accounted for in the pool
while part of the HWP pool is continuously being oxidized. This effect is included in our model, as
HWP are being projected in the simulations.
While the direct emissions from changes in forest carbon pools provide valuable information on the
immediate impacts of forest management, they fail to capture the full climate effect of forestry
activities. The reduction in wood production resulting from designating forests to biodiversity
protection further results in increased emissions from related sectors such as the energy (relying
more on fossil resources rather than bioenergy), building (relying more on fossil-expensive
materials such as concrete and steel), and transport (transporting wood from larger distances)
sectors. These emissions are however not accounted for in the LULUCF sector and hence also not
in our model. Ignoring the substitution effects of wood products and biomass energy overlooks a
crucial component of the forest's role in climate mitigation. Furthermore, emerging practices such
as Bioenergy with Carbon Capture and Storage (BECCS) present additional opportunities to
enhance the climate benefits of using wood for energy by potentially reducing the carbon debt
associated with biomass energy use.
The developments to the EFISCEN-Space model presented in this study, allows for expansion of
the scope including scenario analyses, not only of direct emissions but including the entire systemic
emissions related to changes in forest management, wood production, and the utilization of wood.
This comprehensive perspective is essential for accurately assessing the potential contribution of
forestry to climate change mitigation and for informing policies and practices that maximize the
climate benefits of forest resources.
5.6 Concerns Related to the Discontinuation of the National Forest Inventory
The methodologies employed in this report are fundamentally dependent on the comprehensive data
provided by the National Forest Inventory (NFI). The NFI has been instrumental in supplying the
foundational data necessary for initiating our projections and crafting the underlying models that
inform our analyses. Recently, the Environmental Protection Agency decided to discontinue the
NFI in its current form. This decision poses profound challenges and raises significant concerns
regarding our capacity to accurately monitor, report, and project the climate effects of the nation's
forests within the Land Use, Land-Use Change, and Forestry (LULUCF) sector. This section delves
51
into the critical implications of this discontinuation, focusing on its impact on climate reporting, the
feasibility of forest climate effect projections, and the integrity of future projections due to the
potential loss of continuous data series.
Impact on Climate Reporting for the LULUCF Sector
Adherence to the Intergovernmental Panel on Climate Change (IPCC) guidelines for greenhouse
gas inventory reporting, especially emissions and removals in the LULUCF sector [32], has been
underpinned by the accurate and comprehensive field measurements conducted by the NFI. The
cessation of the NFI's operations in 2024 threatens Denmark's compliance with these international
reporting standards, as remote sensing, in isolation, lacks the capability to capture the nuanced
biophysical parameters essential for thorough LULUCF accounting. This transition jeopardizes the
integrity of Denmark's climate commitments by potentially compromising the credibility of its
reported data.
The potential impact of the decision to discontinue the NFI in its current form is particularly
surprising given the role that forests play in Danish and EU strategies for climate change mitigation.
The Danish government has decided to afforest 250,000 ha with the explicit aim to gain climate
neutrality and in time even net negative emissions. Afforestation is furthermore a pivotal part of
European Green Deal and the EU ambition to be the first climate-neutral continent. Given the
apparent role of forests in climate change mitigation, it is remarkable that Denmark so chooses to
discard of the only available tool for analysing the impact of political initiatives on forest carbon
emissions.
Implications for Forest Climate Effect Projections
The ability to project the climate effects of forests is indispensable for informing climate policies
and strategies at both national and international levels. Such projections are heavily reliant on
robust historical and present-day data regarding forest composition, growth rates, and carbon
sequestration capacities. The continuity of data series provided by the NFI has been invaluable for
understanding the dynamics of forest ecosystems, particularly in the context of changing
management and evolving climate conditions. This longitudinal data has enabled a nuanced
understanding of trends, the assessment of forest management practices, and the formulation of
informed policy decisions. With the termination of the NFI, this continuity is at risk, creating a
significant knowledge gap in our understanding of how forest ecosystems respond to environmental
changes. The resultant data discontinuity will severely hamper future ecological and climate
52
projections, detracting from the effectiveness of research initiatives and policy formulations.
Specifically, the discontinuation of the NFI disrupts the flow of this critical data, thereby impeding
the generation of reliable and accurate forest climate effect projections. This impediment
significantly undermines Denmark's strategic planning for climate change mitigation and
adaptation, diminishing the nation's contribution to global climate objectives.
The discontinuation of the National Forest Inventory presents significant obstacles to Denmark's
climate reporting capabilities, forest management strategies, and scientific research endeavours. It
undermines the nation's ability to fulfil its international reporting obligations, generate precise forest
climate effect projections, and maintain vital long-term ecological data series. In light of these
challenges, it is critical to reassess the decision to discontinue the NFI or to develop an alternative
solution that ensures the continuation of comprehensive, field-based forest monitoring practices in
line with IPCC guidelines.
53
References
1. Nord-Larsen, T. and B. Talbot, Assessment of forest-fuel resources in Denmark: technical
and economic availability. Biomass and Bioenergy, 2004. 27(2): p. 97-109.
2. Lind, C., Træbrændselsressourcer fra danske skove over ½ ha, in Forestry series. 1994.
3. Nord-Larsen, T. and N. Heding, Træbrændselsressourcer fra danske skove over ½ ha:
opgørelse og prognose 2002. Dansk skovbrugs tidsskrift, 2003. 87: p. 1-72.
4. Johannsen, V.K., T. Nord-Larsen, and N.S. Bentsen, Opdatering af skovfremskrivning:
Forventet drivhusgasregnskab for de danske skove 2020-2050, in GN Rapport. 2022 Institut
for Geovidenskab og Naturforvaltning, Københavns Universitet: FRederiksberg. p. 99.
5. Nord-Larsen, T. and V.K. Johannsen, Fremskrivning af danske biomasseressourcer –
skovressourcen, in IGN Rapport. 2022 Institut for Geovidenskab og Naturforvaltning,
Københavns Universitet: Frederiksberg. p. 38.
6. Nord-Larsen, T. and M.K. Suadicani, Træbrændselsressourcer fra danske skove over ½ ha:
opgørelse og prognose 2010, in Arbejdsrapport , Skov & Landskab 2010, Skov & Landskab,
Københavns Universitet: Hørsholm. p. 53
7. Gylling, M., et al., +10 mio. tons planen: muligheder for en øget dansk produktion af
bæredygtig biomasse til bioraffinaderier. 2012, Fødevareøkonomisk Institut: Frederiksberg.
8. Schelhaas, M.-J., et al., Model documentation for the European Forest Information Scenario
model (EFISCEN 3.1.3), in Alterra report 1559 and EFI technical report 26. 2007, Alterra
and European Forest Institute: Wageningen and Joensuu. p. 118.
9. Schelhaas, M.-J., et al., EFISCEN-Space 1.0 model documentation and manual. 2022:
Wageningen. p. 114
10. Schelhaas, E.A.a.M.-J., National Forestry Accounting Plan Submission of the Forest
Reference Level 2021-2025 for the Netherlands. 2019, Wageningen Environmental
Research Wageningen.
11. Nord-Larsen, T. and V.K. Johannsen, Danish National Forest Inventory: Design and
calculations, in IGN Raport. 2016, Institut for Geovidenskab og Naturforvaltning,
Københavns Universitet.
12. Nord-Larsen, T., et al., Skovstatistik 2022. 2023, Institut for Geovidenskab og
Naturforvaltning, Københavns Universitet: Frederiksberg. p. 64.
13. Schelhaas, M.-J., et al., Species-specific, pan-European diameter increment models based
on data of 2.3 million trees. Forest Ecosystems, 2018. 5(1): p. 21.
14. Schelhaas, M.-J., et al., Actual European forest management by region, tree species and
owner based on 714,000 re-measured trees in national forest inventories. PLOS ONE,
2018. 13(11): p. e0207151.
15. Nielsen, O.-K., et al., Denmark's National Inventory Report 2023. Emission Inventories
1990-2021 - Submitted under the United Nations Framework Convention on Climate
Change, in Scientific Report. 2023, DCE – Danish Centre for Environment and Energy. p.
933
54
16. Nord-Larsen, T., H. Meilby, and J.P. Skovsgaard, Simultaneous estimation of biomass
models for 13 tree species: effects of compatible additivity requirements. Canadian Journal
of Forest Research, 2017. 47(6): p. 765-776.
17. Nord-Larsen, T., et al., Skovstatistik 2021. 2023, Institut for Geovidenskab og
Naturforvaltning, Københavns Universitet: Frederiksberg. p. 60.
18. Nord-Larsen, T., et al., Skovstatistik 2020. 2021, Institut for Geovidenskab og
Naturforvaltning, Københavns Universitet.: Frederiksberg.
19. Denmark, S. Statistikbanken.dk. 2024 [cited 2024 April 9]; Available from:
https://statistikbanken.dk/statbank5a/default.asp?w=1920.
20. Pretzsch, H., et al., Forest growth in Europe shows diverging large regional trends. Sci Rep,
2023. 13(1): p. 15373.
21. Chagnon, C., et al., Strong latitudinal gradient in temperature-growth coupling near the
treeline of the Canadian subarctic forest. 2023. 6.
22. Dyderski, M.K., et al., How much does climate change threaten European forest tree
species distributions? 2018. 24(3): p. 1150-1163.
23. Smith, I.A., V.K. Dearborn, and L.R.J.P.o. Hutyra, Live fast, die young: Accelerated growth,
mortality, and turnover in street trees. 2019. 14(5): p. e0215846.
24. Piovesan, G., et al., Drought‐driven growth reduction in old beech (Fagus sylvatica L.)
forests of the central Apennines, Italy. 2008. 14(6): p. 1265-1281.
25. Martinez del Castillo, E., et al., Climate-change-driven growth decline of European beech
forests. Communications Biology, 2022. 5(1): p. 163.
26. Williams, A.P., et al., Temperature as a potent driver of regional forest drought stress and
tree mortality. Nat Clim Chang. 2013, Nature Publishing Group.
27. Tresch, S., et al., The cumulative impacts of droughts and N deposition on Norway spruce
(Picea abies) in Switzerland based on 37 years of forest monitoring. 2023. 892: p. 164223.
28. COWI, Analysis of Danish LULUCF monitoring, reporting and forecasting - Forest Land.
2019: Lyngby. p. 59.
29. Bentsen, N.S., Carbon debt and payback time – Lost in the forest? Renewable and
Sustainable Energy Reviews, 2017. 73: p. 1211-1217.
30. Nielsen, A.T., T. Nord-Larsen, and N.S. Bentsen, CO2 emission mitigation through fuel
transition on Danish CHP and district heating plants. GCB Bioenergy, 2021. 13: p. 1162-
1178.
31. Christiansen, J.R., L. Vesterdal, and P.J.B. Gundersen, Nitrous oxide and methane exchange
in two small temperate forest catchments—effects of hydrological gradients and
implications for global warming potentials of forest soils. 2012. 107: p. 437-454.
32. IPCC, 2006 IPCC Guidelines for National Greenhouse Gas Inventories, E. H.S., et al.,
Editors. 2006, Intergovenmental Panel on Climate Change: Japan.
55
6 Appendix
Table 6.1. Projected carbon pools in above and below ground biomass distributed to total pools and pools in the
afforestation made during the simulations (i.e. not including afforestation prior to the initiation of simulations in 2022).
Total Forest less than 30 year
Year Forest
Area
Above ground
biomass
Below ground
biomass
Above ground
biomass
Below ground
biomass
ha 1,000 t CO2-eq
2022 642.976 138.339 30.441
2023 642.976 139.673 30.705
2024 642.976 140.505 30.848
2025 642.976 141.339 31.004
2026 642.976 142.225 31.176
2027 642.976 142.905 31.293
2028 656.222 144.694 31.672 230 13
2029 656.222 147.186 32.214 286 17
2030 656.222 149.548 32.718 339 20
2031 656.222 151.755 33.210 402 23
2032 656.222 154.005 33.699 466 27
2033 667.556 157.011 34.368 686 40
2034 667.556 159.769 34.975 799 46
2035 667.556 162.340 35.540 912 53
2036 667.556 164.699 36.044 1.034 59
2037 667.556 167.184 36.580 1.169 67
2038 667.825 170.217 37.252 1.311 75
2039 667.825 173.112 37.888 1.443 83
2040 667.825 175.779 38.463 1.592 91
2041 667.825 178.256 38.994 1.753 100
2042 667.825 180.786 39.530 1.884 108
2043 666.975 183.767 40.193 2.036 117
2044 666.975 186.660 40.808 2.204 126
2045 666.975 189.303 41.375 2.335 133
2046 666.975 191.473 41.820 2.504 143
2047 666.975 193.770 42.304 2.676 153
56
Table 6.2. Projected carbon pool contribution to harvested wood products.
Year Total Sawn timber Panels Paper
1,000 t CO2-eq
2022 - - - -
2023 515 386 129 -
2024 619 464 155 -
2025 636 477 159 -
2026 654 491 164 -
2027 689 517 172 -
2028 454 340 113 -
2029 466 349 116 -
2030 512 384 128 -
2031 545 409 136 -
2032 569 427 142 -
2033 444 333 111 -
2034 498 373 124 -
2035 539 404 135 -
2036 587 441 147 -
2037 584 438 146 -
2038 499 375 125 -
2039 515 386 129 -
2040 581 435 145 -
2041 598 448 149 -
2042 615 461 154 -
2043 503 377 126 -
2044 556 417 139 -
2045 596 447 149 -
2046 679 509 170 -
2047 652 489 163 -
57
Table 6.3. Emissions reported from previous inventories and in the previous (KF22) and present (KF24) projection
based on the 5-year moving averages of forest carbon pools used in the reporting tool applied for annual reporting to
the UNFCCC. The year 2022 separates reported and projected values from KF24.
Climate projection 2022 Reorted values/Climate projection 2024
Year Above
ground
biomass
Below
ground
biomass
Dead
wood
Soil Total Above
ground
biomass
Below
ground
biomass
Dead
wood
Soil Total
1,000 t CO2-eq
1959 - - - - -
1960 41 11 -1 3 54
1961 -72 5 -1 -16 -84
1962 -72 5 -1 -16 -84
1963 -72 5 -1 -16 -84
1964 -72 5 -1 -16 -84
1965 -72 5 -1 -16 -84
1966 -72 5 -1 -16 -84
1967 -72 5 -1 -16 -84
1968 -72 5 -1 -16 -84
1969 -72 5 -1 -16 -84
1970 -72 5 -1 -16 -84
1971 -72 5 -1 -16 -84
1972 -72 5 -1 -16 -84
1973 -72 5 -1 -16 -84
1974 -72 5 -1 -16 -84
1975 -72 5 -1 -16 -84
1976 -1.646 -406 -29 -418 -2.498
1977 -1.646 -406 -29 -418 -2.498
1978 -1.646 -406 -29 -418 -2.498
1979 -1.646 -406 -29 -418 -2.498
1980 -1.646 -406 -29 -418 -2.498
1981 -1.646 -406 -29 -418 -2.498
1982 -1.646 -406 -29 -418 -2.498
1983 -1.646 -406 -29 -418 -2.498
1984 -1.646 -406 -29 -418 -2.498
1985 -1.646 -406 -29 -418 -2.498
1986 -1.646 -406 -29 -418 -2.498
1987 -1.646 -406 -29 -418 -2.498
1988 -1.646 -406 -29 -418 -2.498
1989 -1.646 -406 -29 -418 -2.498
1990 -1.006 -222 -18 -251 -1.498
1991 -1.006 -222 -18 -251 -1.498
1992 -1.006 -222 -18 -251 -1.498
1993 -1.006 -222 -18 -251 -1.498
1994 -1.006 -222 -18 -251 -1.498
1995 -1.006 -222 -18 -251 -1.498
58
Climate projection 2022 Reorted values/Climate projection 2024
Year Above
ground
biomass
Below
ground
biomass
Dead
wood
Soil Total Above
ground
biomass
Below
ground
biomass
Dead
wood
Soil Total
1,000 t CO2-eq
1996 -1.006 -222 -18 -251 -1.498
1997 -1.006 -222 -18 -251 -1.498
1998 -1.006 -222 -18 -251 -1.498
1999 -1.006 -222 -18 -251 -1.498
2000 -1.026 -227 -19 -255 -1.526
2001 -991 -215 -8 -240 -1.454
2002 -955 -203 2 -224 -1.380
2003 -919 -191 12 -208 -1.305
2004 -882 -179 23 -191 -1.229
2005 -797 -157 34 -165 -1.084
2006 -796 -156 35 -164 -1.082
2007 -967 -194 16 -164 -1.309
2008 -1.590 -328 3 -163 -2.078
2009 -1.624 -337 -19 -163 -2.142
2010 -1.786 -377 -56 -166 -2.384
2011 -2.430 -525 -73 -356 -3.383
2012 -2.598 -560 -74 -524 -3.757
2013 -2.311 -505 -86 -723 -3.625
2014 -2.478 -547 -111 -896 -4.031
2015 -2.381 -515 -82 -1.048 -4.027
2016 -1.680 -356 -102 -1.007 -3.146
2017 -1.347 -272 -122 -1.044 -2.785
2018 -907 -178 -134 -990 -2.209
2019 -1.225 -240 -137 -999 -2.601
2020 -1.114 -203 -133 -756 -2.206 -1.113 -200 -137 -775 -2.224
2021 -1.135 -192 -166 -580 -2.072 -1.933 -357 -110 -687 -3.087
2022 -1.027 -171 -195 -392 -1.784 -2.188 -452 -84 -730 -3.454
2023 -862 -120 -225 -152 -1.359 -2.413 -439 -57 -560 -3.469
2024 -151 45 -241 122 -226 -1.892 -346 -14 -357 -2.608
2025 119 88 -298 150 59 -1.815 -352 -11 -356 -2.534
2026 59 63 -252 138 8 -1.153 -225 -6 -246 -1.630
2027 -1 39 -206 126 -43 -917 -156 - - -1.073
2028 -61 14 -160 113 -94 -1.008 -181 - - -1.189
2029 -121 -10 -114 101 -144 -1.340 -205 - - -1.546
2030 -181 -35 -68 89 -195 -1.645 -281 - - -1.925
2031 -206 -38 -65 89 -220 -1.907 -345 - - -2.252
2032 -231 -42 -63 90 -245 -2.219 -403 - - -2.622
2033 -256 -45 -60 90 -270 -2.460 -471 - - -2.931
2034 -281 -48 -58 91 -295 -2.511 -522 - - -3.033
2035 -305 -51 -55 92 -320 -2.552 -533 - - -3.085
2036 -269 -44 -56 85 -283 -2.582 -545 - - -3.126
59
Climate projection 2022 Reorted values/Climate projection 2024
Year Above
ground
biomass
Below
ground
biomass
Dead
wood
Soil Total Above
ground
biomass
Below
ground
biomass
Dead
wood
Soil Total
1,000 t CO2-eq
2037 -233 -37 -56 79 -246 -2.628 -546 - - -3.175
2038 -197 -29 -56 73 -209 -2.635 -557 - - -3.192
2039 -160 -22 -56 67 -172 -2.662 -559 - - -3.221
2040 -124 -14 -57 61 -134 -2.682 -566 - - -3.247
2041 -149 -19 -55 56 -167 -2.706 -568 - - -3.274
2042 -173 -24 -53 51 -199 -2.715 -574 - - -3.289
2043 -198 -30 -51 47 -232 -2.705 -575 - - -3.279
2044 -223 -35 -49 42 -264 -2.704 -573 - - -3.278
2045 -247 -40 -47 37 -297 -2.700 -570 - - -3.269
2046 -246 -39 -45 39 -291 -2.638 -568 - - -3.206
2047 -244 -38 -44 41 -286 -2.592 -551 - - -3.143
university of copenhagen
department of geosciences and
natural resource management
rolighedsvej 23
dk - 1958 frederiksberg
tel. +45 35 33 15 00
ign@ign.ku.dk
www.ign.ku.dk