Dataset DOI: 10.5061/dryad.tb2rbp0dv

Corresponding author information

Name: Marc Grünig
Affiliation: Swiss TPH, Switzerland
email: marc.gruenig@swisstph.ch

Alternative contact information

Name: Werner Rammer
Affiliation: TUM, Germany
email: werner.rammer@tum.de

Description of the data and file structure

Files and variables

This folder contains data for the publication Climate change will increase forest disturbances in Europe throughout the 21st century.

The data provided here allows to run code that is stored in the folder 00_code. We provide several folders with data. Each folder contains its own README file with more details about the files within the folder.

Additional details and description about the data can be found in the R and Python scripts.

The data is organized in the following folder structure that roughly follows the order of the steps taken:

Directory tree

├── 00_code
├── 01_simulation_data_pipeline
├── 02_dnn
├── 03_initial_forest_states
├── 04_disturbance_modules
├── 05_clim_data
├── 06_gis
├── 07_reference_grids
├── 08_svd
├── 10_results
└── 11_figures

Below, the detailed description of all folders is listed.

00_code.zip

00_code
├── .gitignore
├── .Rhistory
├── 00_functions
│   ├── check_attribute_cols_function.R
│   ├── check_height_cols_function.R
│   ├── check_species_cols_function.R
│   ├── check_species_condition.R
│   ├── create_aug_examples_function.R
│   ├── create_examples_function.R
│   ├── get_point_id.R
│   ├── height_max_mean_function.R
│   ├── height_transformer.R
│   ├── kahn_factors_function.R
│   ├── reorder_strings.R
│   ├── states_mapping_function.R
│   └── yield_table.R
├── 01_data_pipeline
│   ├── 00_examples.R
│   ├── 01_collect_soildata.R
│   ├── 02_climate_data_timeseries.R
│   ├── 03_collect_climatedata.R
│   └── 04_examples_complete.R
├── 02_fire_module
│   ├── 01_wind_directions.R
│   ├── 02_summer_vpd_calc.R
│   ├── 03_fire_freq_size_models.R
│   ├── 04_fire_occurrence_model.R
│   ├── 05_create_fire_events.R
│   └── 06_fire_states.R
├── 03_wind_module
│   ├── 01_wind_simulations.R
│   └── 02_wind_susceptibility.R
├── 04_bbtl_module
│   ├── 01_background_proba.R
│   ├── 02_bbtl_susceptibility.R
│   ├── 03_bbtl_kernels.R
│   └── phenips
│       ├── bbgenerations.cpp
│       ├── bbgenerations.h
│       ├── bbgenerations.o
│       ├── bb_kernels.csv
│       └── phenips.cpp
├── 05_mgmt_module
│   ├── 01_mgmt_proba.R
│   └── 02_mgmt_spatial_cap.R
├── 06_climate_data
│   ├── 01_cordex_biascorrection.R
│   ├── 02_svd_climate_data.R
│   └── 03_svd_climate_compr.py
├── 07_svd_initial_landscape
│   ├── 01_prepare_canopyheight_layer.R
│   ├── 02_vegetation_stacks.R
│   ├── 03_initial_soilconditions.R
│   ├── 04_initial_states_landscape.R
│   ├── 05_initial_vegetation_evaluation.R
│   └── 06_forest_state_lookups.R
├── 08_outputs_analysis
│   ├── 01_svd_grid_processing.R
│   ├── 02_extract_distrate_25km.R
│   ├── 03_annual_outputs_processing.R
│   ├── 04_extreme_events.R
│   ├── 05_old_young_forests.R
│   ├── 06_Fig1a_disturbance_rates.R
│   ├── 06_Fig1b_heatmap.R
│   ├── 07_Fig2a_biomes_dist_change.R
│   ├── 07_Fig2b_disturbancemaps.R
│   ├── 08_Fig3a_stacked_bars.R
│   ├── 08_Fig3bcd_extreme_events.R
│   ├── 09_Fig4_structure_trends.R
│   ├── 21_module_evaluation.R
│   ├── 22_uncertainties.R
│   └── 23_feedbacks_interactions.R
├── disturbance_scenarios_europe.Rproj
└── README.md

Contains the code and resources to create and analyze future disturbance scenarios for Europe. The code is also available at https://github.com/magrueni/disturbance_scenarios_europe

01_simulation_data_pipeline.zip

01_simulation_data_pipeline
├── climate_expl.csv
├── climate_slopes.csv
├── clim_bias_expl.csv
├── clim_slopes
│   ├── prec_ICHEC_EC_EARTH_historical.csv
│   ├── prec_ICHEC_EC_EARTH_rcp_2_6.csv
│   ├── prec_ICHEC_EC_EARTH_rcp_4_5.csv
│   ├── prec_ICHEC_EC_EARTH_rcp_8_5.csv
│   ├── prec_MPI_M_MPI_ESM_LR_historical.csv
│   ├── prec_MPI_M_MPI_ESM_LR_rcp_2_6.csv
│   ├── prec_MPI_M_MPI_ESM_LR_rcp_4_5.csv
│   ├── prec_MPI_M_MPI_ESM_LR_rcp_8_5.csv
│   ├── prec_NCC_NorESM1_M_historical.csv
│   ├── prec_NCC_NorESM1_M_rcp_2_6.csv
│   ├── prec_NCC_NorESM1_M_rcp_4_5.csv
│   ├── prec_NCC_NorESM1_M_rcp_8_5.csv
│   ├── temp_ICHEC_EC_EARTH_historical.csv
│   ├── temp_ICHEC_EC_EARTH_rcp_2_6.csv
│   ├── temp_ICHEC_EC_EARTH_rcp_4_5.csv
│   ├── temp_ICHEC_EC_EARTH_rcp_8_5.csv
│   ├── temp_MPI_M_MPI_ESM_LR_historical.csv
│   ├── temp_MPI_M_MPI_ESM_LR_rcp_2_6.csv
│   ├── temp_MPI_M_MPI_ESM_LR_rcp_4_5.csv
│   ├── temp_MPI_M_MPI_ESM_LR_rcp_8_5.csv
│   ├── temp_NCC_NorESM1_M_historical.csv
│   ├── temp_NCC_NorESM1_M_rcp_2_6.csv
│   ├── temp_NCC_NorESM1_M_rcp_4_5.csv
│   └── temp_NCC_NorESM1_M_rcp_8_5.csv
├── examples_expl.csv
├── metadata_expl.csv
├── metadata_processed_expl.csv
├── readme.txt
├── simdata_expl.csv
├── states_lookup_pruned.csv
├── states_lookup_pruned_init_ls.csv
└── time_series.csv

This folder contains data which can be used to have a look at the data pipeline that converts local forest simulations from the harmonized simulation database (https://zenodo.org/records/12750180) to deep neural network training data. Values in the example have been modified for demonstration purposes. While we cannot provide the full dataset here, the corresponding code allows to create the full training dataset from the underlying database. More information and code is currently under review (Grünig, M. et al. Loss of competitive strength in European conifer species under climate change. Under review)

The four main variables that are contained in the .csv files with climate information are: temperature [°C], precipitation [mm], VPD [kPa] and radiation [W/m²]. The file names represent one of the four climate variables (tas, prec, vpd, rad). All files containing climate data, start with clim_. The folder clim_slopes contains .csv tables with annual mean temperatures which are used to translate climate data from the forest simulations to the climate change scenarios used in SVD. File naming follows the convection of climatevariable_GCM_RCP.csv where climate variable can be prec or temp, GCM is one of ICHEC_EC_EARTH, MPI_M_MPI_ESM_LR, NCC_NorESM1_M. RCP is historical,  rcp_2_6, rcp_4_5 or rcp_8_5.

Further, the folder contains .csv files with examples of simulation data and related metadata. Those files contain predominantly the three dimensions of forest states: species proportions [%], LAI [m²/m²], Dominant height [m], as well as stand soil information on WHC [mm], texture (sand, silt, clay [%]), depth [mm], plant available nitrogen [kg/ha/year]. The file examples_expl.csv contains training data examples with forest state transitions and residence times.

02_dnn.zip

02_dnn
├── dnn_states_lookup.csv
├── models
│   ├── climcompressor_v4.h5
│   └── simple_frozen_graph_new_dnn_v17_7.pb
├── readme.txt
├── states.csv
└── states_europe_meta.csv

This folder contains information that is relevant for the DNN part of SVD. It includes .csv files with forest state information (e.g. PIAB_20_22_2) and corresponding state IDs which are needed as SVD inputs.
The trained models for climate compression and the model driving SVD are stored in the models subfolder.

03_initial_forest_states.zip

03_initial_forest_states
├── evaluation
├── initial_soil.csv
├── init_restime.tif
├── init_soil_rast.tif
├── init_veg_v2.tif
├── readme.txt
├── sampler
│   ├── vegetationSampler.html
│   └── vegetation_sampler.cpp
├── states_lookup_pruned.csv
└── vegetation

This folder contains underlying data and results for the initial forest landscape and soil condittions that are used as starting point in SVD.

• .tif & .csv files are the final layers and data for vegetation and soil that are used in SVD. The values represent forest state ids, residence time (1-10) or soil conditions (soil depth [mm], sand content [%], water holding capacity [mm], plant available nitrogen [kg/ha/year]) for each grid cell.
• evaluation contains the results from the corresponding script as figures (.png) and tables (.csv). These files are not relevant to recreate the pipeline.
• sampler contains C++ code for the tree species sampling performed for the initial vegetation
• vegetation is currently empty. We cannot share the data aggregate from different sources that do not underly CC0 licensing. The relevant links to the sources are provided in the methods, suppl. material and code.

04_disturbance_modules.zip

04_disturbance_modules
├── bbtl_module
│   ├── age_estimator.rds
│   ├── background_equalproba_100km.tif
│   ├── bb_kernels.csv
│   ├── bb_patches_vpd.csv
│   ├── patches
│   ├── phenips
│   ├── readme.txt
│   ├── spruce_share.tif
│   ├── states_bb.csv
│   └── transition_matrix_bb.csv
├── fire_module
│   ├── climate
│   ├── complexes
│   ├── figures
│   ├── firestates.csv
│   ├── fire_event_series
│   ├── fire_transitions.csv
│   ├── gis
│   ├── models
│   ├── readme.txt
│   └── wind_table.csv
├── mgmt_module
│   ├── management.csv
│   ├── mean_harvest_100km.tif
│   ├── mgmt_transitions.csv
│   ├── patches
│   └── readme.txt
└── wind_module
    ├── copernicus_storms
    ├── patches
    ├── readme.txt
    ├── results
    ├── wind_event_series
    └── wind_transitions.csv

This folders contain all relevant data to the SVD external part of the disturbance modules (i.e., creating disturbance events).

Readme files are included in each subfolder. We provide predominantly the calculated disturbance events along their position for wind and fire modules. The wind event series files contain information on the location of the event (cell_position, x, y), the number of 10km grid cells affected (number), proportion of 100m grid cells affected (proportion) and timepoint in the simulation (year). The wind event series tables follow the naming convention simulated_storms_drawX.csv, where X corresponds to the draw 1 to 90. The fire event series contains .csv tables with a unique fire id (fire_id), the timepoint in the simulation (year), coordinates of ignition (x, y). The allowed maximum size (max_size), the wind direction and speed (winddirection, windspeed). The table follow the naming convention future_fire_simulation_GCM_RCP_simulation_REP.csv. GCM is one of ncc, ichec, mpi, RCP is one of historical, rcp26, rcp45, rcp85 and REP is the replicate 1 to 10. For bark beetle and management modules no such event series exist, however all relevant input for SVD is provided in the folders.

The underlying data of the disturbance patches can be downloaded from https://zenodo.org/records/7080016, https://zenodo.org/records/8202241 and https://zenodo.org/records/11070255.
Additionally, we provide VPD data [kPa] for fire and bbtl, as well as .tif and .csv files with results.

05_clim_data.zip

05_clim_data
├── annual_climate
│   ├── svd_clim_ICHEC-EC-EARTH_historical_v3_aux_biascorr.csv
│   └── svd_clim_ICHEC-EC-EARTH_rcp_8_5_v3_aux_biascorr.csv
├── biascorr_ICHEC-EC-EARTH_v3.csv
├── daily_climate
└── readme.txt

This folder contains the examples of the climate data pipeline and the resulting annual climate that drives SVD.
the example annual climate is also used for the SVD example landscape and has been modified.

Due to limited data storage capacities, we can only provide one example for historical and future climate dataset here.
The full datasets will be shared upon request.

The .csv files of annual climate contain data on temperature [°C], precipitation [mm], VPD [kPa] and radiation [W/m²].

06_gis.zip

06_gis
├── countries
│   └── readme.txt
├── dem5.tif
├── europe_shp
│   ├── europe.dbf
│   ├── europe.prj
│   ├── europe.shp
│   ├── europe.shx
│   ├── europe_lowres.dbf
│   ├── europe_lowres.prj
│   ├── europe_lowres.shp
│   └── europe_lowres.shx
├── forest_mask_new_laea_masked.tif
├── forest_mask_new_laea_masked_nas.tif
├── forest_mask_wgs.tif
├── LAI_annual_mean_2020_2021_projwgs.tif
├── predicted_N_available.tif
├── readme.txt
└── texture_categories.tif

This folder contains various shapefiles (.shp) and raster layers (.tif) that are used for calculation steps, including masks of the study region.

07_reference_grids.zip

07_reference_grids
├── biomes_hex.gpkg
├── eu_mask.cpg
├── eu_mask.dbf
├── eu_mask.prj
├── eu_mask.shp
├── eu_mask.shx
├── eu_mask_hexagons_25km.cpg
├── eu_mask_hexagons_25km.dbf
├── eu_mask_hexagons_25km.prj
├── eu_mask_hexagons_25km.shp
├── eu_mask_hexagons_25km.shx
├── gridID_10k.tif
├── hex_forest_mask_25km.gpkg
├── readme.txt
├── reference_grid.tif
├── reference_grid_100km.tif
├── reference_grid_10km.tif
├── reference_grid_1km.tif
└── reference_grid_tab.csv

This folder contains raster layers (.tif) and sf objects (.shp) that are needed as reference grids in the code.
Additionally, we provide a .csv table with the reference grid showing grid Ids along latitude and longitude.

08_svd.zip

08_svd
└── svd_example
    ├── climate
    ├── dnn
    ├── example_ls_small
    │   └── output_example_1
    ├── fire
    ├── gis
    ├── log
    └── wind

This folder contains the data for a small landscape to explore SVD.
All relevant files to run a simulation are stored in the svd_example subfolder.
Visit https://github.com/edfm-tum/SVD/tree/main/SVDModel to download SVD and look at the documentation.
Note that we provide data for the fire and wind module only here for the sake of simplicity.

09_svd_simulations.tar.gz

09_svd_simulations
├── annual_outputs
├── biodiv
├── processed_data
├── raw
├── readme.txt
└── svd_simulations_ids.csv

This folder contains the output of 120 SVD simulations. The file svd_simulation_ids.csv shows the configuration of each simulation with columns for simulation ID, RCP, GCM, Fire scenario, wind scenario and Replication ID.

The folder raw contains the raw spatial outputs at 100 x 100m resolution at 10 year timesteps.
There are four subfolders for each module output containing .tif raster files.
name coding for the 10 year outputs is module_timestep_RCP_GCM_REP_10year.tif.
values in the files correspond to number of disturbances per grid cell within the focal 10 year timestep.
Additionally, we provide summed layers at 40 and 80 years of simulation (i.e. 2060 and 2100). Subfolders exist for barkbeelte, fire, management and wind. All files in the subfolers follow the naming convention AGENT_YEAR_RCP_GCM_REP_10year.tif (e.g. fire_10_rcp26_ncc_1_10year.tif), summed layers at 40 and 80 years simulation end with _summed.tif. AGENT is one of bbtl, fire, wind, mgmt; YEAR corresponds to simulation decade (10, 20, 30, 40, 50, 60, 70, 80); RCP is one of historical, rcp26, rcp45, rcp85; GCM is one of ichec, ncc, mpi; REP is the simulation replicate 1 to 90 (future) and 91 to 120 (historical).

The folder annual_outputs contains the annual timeseries of disturbed area across Europe
the folder is organised in subfolders for each simulation which contain .csv files for barkbeetle, fire and wind. Subfolders are named output_sim_eu_X where X corresponds to the simulation replicate. The file svd_simulation_ids.csv shows the configuration of each simulation with columns for simulation ID, RCP, GCM, Fire scenario, wind scenario and Replication ID. Within each output folder, three .csv tables are provided with information on barkbeetles, fire and wind:

• The barkbeetle output files show yearly outputs with columns for simulation year (year), number of gridcells with background infestations (n_background_infestation), number of gridcells with wind infestation (n_wind_infestation), number of grid cells impacted (n_impact), number of active infestations at the end of the year (n_active_yearend). 
• The fire output files show records for each fire with columns for simulation year (year), fire id (id), coordinates of ignitions (x, y), planned size from external event series (planned_size), realized size in SVD (realized_size), severity (share_high_severity). 
• The wind output files show records of each storm event with columns for simulation year (year), wind event id (id), coordinates of ignitions (x, y), proportion of planned effect (proportion), planned affected regions (=10km grids) (regions_planned), affected regions (regions_affected), forested cells in region (cells_forested), average successibility of forest in regions (mean_susceptibility), 100m cells planned (cells_planned), 100m cells affected (cells_affected). 

The folder biodiv contains the biodiversity data aggregated to 25km hexagons stored as .gpkg files. The files are named undist_vs_recently_dist_forest_X.gpkg where is corresponds to the simulation ID 1 to 120. 

The folder processed_data contains the processed spatial data at 25km and 50km hexagons. 50km hexagons were only used for demstration purposes. In the subfolder dist_rates_25km all process simulations used for the analysis are stored. The .csv tables follow the naming all_dist_rates_YEAR_RCP.csv, where YEAR corresponds to the simulation decade (10, 20, 30, 40, 50, 60, 70, 80); RCP is one of rcp26, rcp45, rcp85

10_results.zip

10_results
├── all_agents_distarea_final.csv
├── all_agents_distarea_fut.csv
├── all_agents_distarea_hist.csv
├── bbtl_results_final.csv
├── bbtl_results_fut_final.csv
├── bbtl_results_hist_final.csv
├── disturbance_change_final.csv
├── fire_results_final.csv
├── fire_results_fut_final.csv
├── fire_results_hist_final.csv
├── fut_dist_annual_final.csv
├── hist_dist_annual_final.csv
├── wind_results_final.csv
├── wind_results_fut_final.csv
└── wind_results_hist_final.csv

Contains the results of the statistical analysis of the simulation outputs. Files are provided for individual all agents combined (all_agents_distarea_), individual agents (bbtl_, fire_, wind_), and split by historical and future simulations (_fut.csv, _hist.csv). The .csv files show simulation number (sim), simulation year (year), climate change scenario (scen), disturbed area [ha] (dist_area), time period (period) and disturbance agent (agent) as columns.

11_figures.zip

11_figures
├── figure_data
│   ├── Fig1a_FigS1_fut.csv
│   ├── Fig1a_FigS1_hist.csv
│   ├── Fig1b.csv
│   ├── Fig2a_biomes.csv
│   ├── Fig2b.csv
│   ├── Fig3a.csv
│   ├── Fig3b_fire_extremes.csv
│   ├── Fig3c_bbtl_extremes.csv
│   ├── Fig3d_wind_extremes.csv
│   ├── Fig4a_FigS23a.csv
│   ├── Fig4b_FigS23b.csv
│   ├── FigS10_data_map.csv
│   ├── FigS11_data_map.csv
│   ├── FigS12_data_map.csv
│   ├── FigS13_sd_map_fire.csv
│   ├── FigS14_sd_map_bbtl.csv
│   ├── FigS15_sd_map_wind.csv
│   ├── FigS16_fut.csv
│   ├── FigS16_hist.csv
│   ├── FigS17_fut.csv
│   ├── FigS17_hist.csv
│   ├── FigS18_fut.csv
│   ├── FigS18_hist.csv
│   ├── FigS19.csv
│   ├── FigS2.csv
│   ├── FigS20.csv
│   ├── FigS21.csv
│   ├── FigS22.csv
│   ├── FigS24a.csv
│   ├── FigS24b.csv
│   ├── FigS29.csv
│   ├── FigS3.csv
│   ├── FigS31.csv
│   ├── FigS32.csv
│   ├── FigS33.csv
│   ├── FigS34.csv
│   ├── FigS35.csv
│   ├── FigS36.csv
│   ├── FigS4.csv
│   ├── FigS5_data_map.csv
│   ├── FigS6_data_map.csv
│   ├── FigS7_data_map.csv
│   ├── FigS8_data_map.csv
│   └── FigS9_data_map.csv
└── figure_descriptions.txt

Contains the underlying data of all figures in the manuscript and supplementary material (.csv tables). Figure descriptions are provided in the figure_descriptions.txt file. 

Code/software

• Python: Version 3.9 or higher
• R: Version 4.4.1 or higher