The Inter-Sectoral Impact Model Integration and
Intercomparison Project (ISI-MIP2)
Comparison of Forest Landscape
Models (CoFoLaMo)
(Simulation protocol for forest landscape models only, for the general, cross-sectoral part of the
protocol refer to the file: ISI-MIP2_protocol_general.docx )
Contributors: Matthias Speich, Dirk Schmatz, Giorgio Vacchiano, Heike Lischke, Laura Schuler,
Björn Reineking, Paola Mairota, Vincenzo Lerronni, Josef Bruna, Dominik Thom
Harald Bugmann, Koen Kramer, Rupert Seidl, Annabel Porte, Robert Scheller, Hong He
Rationale for multi-landscape-model simulations in ISI-MIP2
and PROFOUND
This is a protocol to support multi-model simulations of forest landscapes, i.e. spatially interacting
forest areas embedded in non-forest-areas, for both model evaluation with observed data but also
for model projections under climate change. In addition to the general ISI-MIP input data to be used
for these simulations (mostly climate data), a number of regions has been selected for which a
range of forest landscape models can be rather easily initialized and observational data is available
for model evaluation.
General setup of the study
The study comprises several regions, models and scenarios and will be done in three phases:
Warm-up (until Beginning 2016)


Dischma
Standard-Scenario, Validation-simulation
Comparison (until mid 2016)



Dischma, Aosta, Bavarian/Bohemian forest, Puglia
Standard-Scenario + Variations
Validation + Future („Catch-up“)
Spin offs (later…)


Other sites: Atlantic coast of France, Veluwe-forest in the Netherlands
Specific questions: e.g. invasion of mediterranean species in Atlantic France, large herbivores in
Veluwe
1

Please formulate
Models involved





LANDIS II
Landclim
TreeMig,
ForHyCS
Iland
Regions




Dischma in North-eastern Switzerland
Aosta in the North-west Italian Alps
Bavarian Forest in Czech Republic and Germany
Puglia in Southern Italy
Resolution



Input data: climate 100m
Models as they come but as a fraction of 100: Landclim 25m and LANDIS 20m Treemig 100m
Forhycs100m Iland???
Evaluation: 100m, or smallest common denominator
Research questions



Overarching research questions for forest landscape models

How are forest landscapes (on different scales) affected by

Climate change (different scenarios, in space)

Forest management and land use change
ISI-MIP questions

Model evaluation

Extreme events (both in future and how well do models get extremes in past
(model evaluation).
Region specific questions

…
Scenarios
Climate/CO2

Validation:

Hindcasted (1500-1900) + Historical observations (1900-2012). Spatial grid
100m. Current CO2
2


For Hindcasting we use a predefined random sequence of the first 20 years of
the observations from 1500-1900. The climate for these years is used for the
entire simulation area.
Future:


Hindcasted (1500-1900) + Historical observations (1900-2012)+ HadGEM2ES (1951-2005) + HadGEM2-ES- Scenarios
RCP 2.6, 4.5, 6, 8 (2006-2100). Corresponding CO2, where possible,
otherwise current CO2
Spin-up
TreeMig, ForHyCs, Landclim: 1500-1950 (2012 in Validation run)
Landis II: takes output of TreeMig or Landclim from 1900 as initialization
3
Management


Standard per site, also in future
In future also “no management”
Land use



Land use here means a) “forests are allowed by humans to grow ” versus b) “forests are not
allowed to grow by humans”. For current forests a) means all current forest area + other not
managed natural areas (meadows) . For future (potential) this means current forest area plus
areas where forest might be allowed in future minus areas where forests might be converted
by humans to other land use categories. additionally currently extensively used pastures and
meadows. This has to be defined by expert knowledge of the region responsible. One example is
that extensively used meadows and pastures will be abandoned. In the simulations this future
potential forest becomes available in 2006!
Validation simulations: current land use
Future simulations:
1. Current
2. potential land use
Disturbances



As included in the models.
Random in TreeMig/ForHycs Random means by frequency distributions of intensity or areas
Both variants:
1. Random or spatially linked in LandClim and LANDISII (Paola and Giorgio)
2. spatially linked in LANDISII (Josef)
Dispersal
1.
2.
Seeds enter a cell only by dispersal
Seeds are present everywhere („extraterrestrial seeds“)
Input data
As grids, spatial resolution 100m
Climate




Daily T, P, solar radiation
Observational data (1900-current) ideally as timeseries of grids, otherwise station data
Downscaling (Dirk)

of observational data and HadGem scenarios

Output: As timeseries, as netcdf
Upscaling to monthly data by Heike or Laura
Soil:


Field capacity, AWC (water holding capacity),or bucket size with information to which depth it
refers
Soil type
4

Soil depth, if available
Usual forest management
As simple rules, ideally as probability and intensity. + additional information
Random disturbances
As frequency distribution of intensities
Land use mask
On sufficiently fine resolution, e.g. 25 m in Puglia, 100m in Davos.
Current and potential (see above)
Comparison data
Whatever we can get!
Modis, Sentinel, lidar, NFI, historical maps, MAB
Note year!
Preferedly including 2003/2007
Output data
All per cell and spec ies
From 1900 – 2100 : entire region, transient
Biomass
LAI
Variables per height –class or age
Experiments and possible analyses
Standard simulations
To avoid to have to run all combinations of scenarios we will focus on one standard simulation and
explore differences with respect to this standard. This standard consists of


Climate: Hindcasted + Historical climate + RCP4.5;
With dispersal, disturbances, management, current land use for past and future
Variations=Scenario combinations
5
Dimensions of ISI-MIP
comparison scenarios
5
2: Past/Future
Playground
2 Dispersal/ubiquitous seeds
System
dispersal, disturbances,
management, current land
use for past and future
Comparison against
standard:
5*(5+2+2+2+2+2+2)=55
1
Climate model
1
Climate
Standard: RCP4.5, with
2: yes/no
Disturbances
•Stand forest
models
•Landscape forest
models
2: yes/no
5 Current,
RCP 2.6,
4.5, 6, 8
2 Current/none
2
2 Current /potential
Human influence
Table 1 Variables to be reported by forest landscape models. Abbreviations are provided in Table 2.
Variables should be reported as documented in section XXX (section 6 main protocol).
long name
units
output variable name
Essential (mandatory)
outputs
All outputs per species and
grid-cell and averaged over
grid-cells and species
Biomass
Kg/ha
per species and cell
biom_<species/total>
Average Tree Height
m
per species and cell total
height_<species/total>
Basal Area
m²ha-1
per species and cell total
ba_<species/total>
Volume of Dead Trees
m³ha-1
per species and cell total
mort_<species/total>
Harvest
m³ha-1
per species and cell total
harv_<species/total>
Remaining stem number
after disturbance and
Trees/ha
per species and cell total
stemno_<species/total>_<dbhclass>
6
management by dbh class
Cell Volume
m³ ha-1
per species and cell total
vol_<species/total>
Tree age by dbh class
yr
per species and cell total
age_<species/total>_<dbhclass>
Gross Primary Production
kg m-2 s-1
per species and cell total
gpp_<species/total>
Net Primary Production
kg m-2 s-1
per species and cell total
npp_<species/total>
Autotrophic (Plant)
Respiration
kg m-2 s-1
per species and cell total
ra_<species/total>
Heterotrophic Respiration
kg m-2 s-1
per species and cell total
rh_<species/total>
Net Ecosystem Exchange
kg m-2 s-1
per cell
nee_<total>
Mean Annual Increment
m³ ha-1
per species and cell total
mai_<species/total>
Fraction of absorbed
photosynthetically active
radiation
%
per species and cell total
fapar_<species/total>
Leaf Area Index
m2 m-2
per species and cell total
lai_<species/total>
Species composition
% (biomass)
per ha
species_<species>
Total actual
Evapotranspiration
kg m-2 s-1
per species and cell total
evap_<species/total>
Evaporation from Canopy
(interception)
kg m-2 s-1
per species and cell total
intercep_<species/total>
Water Evaporation from Soil
kg m-2 s-1
per cell
esoil
Transpiration
kg m-2 s-1
per species and cell total
trans_<species/total>
Soil Moisture
kg m-2
per cell
soilmoist
7
Optional outputs
Distribution limit distance
from reference point; total
(=treeline), per species
treeline_<species/total>
Removed stem numbers by
size class by management
Trees/ha
per species and cell total
harvstemno_<species/total>_<dbhclas
>
Volume of disturbance
damage
m³ha-1
per species and cell total
dist_<dist_name>
Nitrogen of annual Litter
g N m-2a-1
per species and cell total
nlit_<species/total>
Nitrogen in Soil
g N m-2a-1
cell total
nsoil_<total>
Net Primary Production
allocated to leaf biomass
kg m-2 s-1
per species and cell total
npp_landleaf__<species>
Net Primary Production
allocated to fine root
biomass
kg m-2 s-1
per species and cell total
npp_landroot_<species>
Net Primary Production
allocated to above ground
wood biomass
kg m-2 s-1
per species and cell total
npp_abovegroundwood_<species>
Net Primary Production
allocated to below ground
wood biomass
kg m-2 s-1
per species and cell total
npp_belowgroundwood_<species>
Root autotrophic respiration
kg m-2 s-1
per species and cell total
rr_<species/total>
Carbon Mass in Leaves
kg m-2
per species and cell total
cleaf_<species>
Carbon Mass in Wood
kg m-2
per species and cell total
cwood_<species>
Carbon Mass in Roots
kg m-2
per species and cell total
croot_<species>
Temperature of Soil
K
per cell
tsl
8
Note: If you cannot provide the data at the temporal or spatial resolution specified, please provide it
the highest possible resolution of your model. Please contact the coordination team (isi-mip@pikpotsdam.de) to for any further clarification, or to discuss the equivalent variable in your model.
9
Table 2: Codes for species, disturbance names and dbh classes as used in protocol (species, dist_name, dbhclass)
OTHER SPECIES to be added
long name
Short name
Fagus sylvatica
fasy
Quercus robur
quro
Quercus petraea
qupe
Pinus sylvestris
pisy
Picea abies
piab
Pinus pinaster
pipi
Eucalyptus globulus
eugl
fire
fi
wind
wi
insects
ins
drought
dr
grazing
graz
diseases
dis
DBH_class_0-20*
dbh_c0
comment
10
DBH_class_20-40*
dbh_c20
DBH_class_40-60*
dbh_c40
DBH_class_60-80*
dbh_c60
DBH_class_80-100*
dbh_c80
DBH_class_100-120*
dbh_c100
DBH_class_120-140*
dbh_c120
DBH_class_>140*
dbh_c140
*the boundaries of the dbh classes should interpreted as follows: dbh_class_0-20 = 0 to<20 cm; dbh_class_20-40 =20-<40 cm, etc…. the dbh class dbh_c140
includes all trees of 140cm dbh and larger.
11
12
Experiments from the stand modelers of TG3(?). We should discuss latest in Davos what exactly we do.
Historic runs and validation exercise – Experiments 1a
These are the core simulations for ISI-MIP2.1A. For the sites mentioned in Error! Reference source not found., a detailed comparison of model-data(mis)match is envisaged, especially with a focus on past extreme events (e.g. 2003) and variability. The data in <BIOME data table, link once in final document>
may also be interesting for some additional validation tasks that can be carried out during postprocessing. The simulations of Experiment 1a listed in Error!
Reference source not found. are needed for this experiment.
ISI-MIP Fast-track catch-up runs – Experiments 2a
These are simulations for the sites mentioned in Error! Reference source not found. using ISI-MIP Fast track climate scenarios to project forest development
under climate change in the future. These are interesting for cross-scale comparisons with DGVMs, cross-sectoral analysis of climate impacts and multi-model
climate change impact projections. The simulations of Experiment 2a listed in Error! Reference source not found. are needed for this experiment.
Influence of disturbances – Experiments 1b and 2b
These are historic and future simulations as described in sections 0 and 0 but with dynamic disturbances switched on for those models that actually simulate
such dynamics. These simulations can be used to isolate the effects of disturbances vs. climate or to consider the joint impact of climate change and
disturbances on forest products and services. The simulations of Experiment 1b and 2b listed in Error! Reference source not found. are needed for this
experiment.
Isolation of climate effects (optional, future experiment)
Simulate time slices (i.e. same cell as growing in past simulations is repeatedly simulated for different time slices of maybe 20-30 years) to isolate the effects of
climate change from the effects of forest dynamics. Some cells are already very old and would reach 200 years or more of age in 2100.
Climate input uncertainty (optional, future experiments)
To do list
Warm up
13
Comparison
14
15