Published 2026 | Version v1

A transient simulation of global vegetation dynamics over the past 20,000 years

  • 1. Peking University

Description

This repository accompanies “A transient simulation of global vegetation dynamics over the past 20,000 years”. It provides simulations of potential natural vegetation and associated land-surface variables generated with ORCHIDEE-MICT, together with factorial experiments, model source code, configuration information and forcing-processing scripts.

The main transient simulation spans 20,000 years before present (BP) to 1999 CE. BP is referenced to 1950 CE, so the annual age coordinate runs from 20,000 to −49 BP. The simulations use prescribed climate and atmospheric CO₂. Climate forcing is derived from TraCE-21K-II, with bias correction and within-month temporal disaggregation based on CRU-JRA v2.2 for 1901–1920, as described in the accompanying manuscript. The files use the TraCE-21K-II T31 grid (approximately 3.75° resolution), with 96 longitude coordinates and 48 Gaussian latitude coordinates. 

Transient simulation outputs

Data_transient_annual.zip contains eight NetCDF files, which are annual-mean outputs from the transient simulation:
•    Veg_annual.nc: fractional cover for each archived PFT category. 
•    LAI_annual.nc: leaf area index aggregated across PFTs.
•    GPP_annual.nc and NPP_annual.nc: gross and net primary productivity aggregated across PFTs.
•    Biomass_annual.nc: plant biomass carbon stock aggregated across PFTs.
•    Evap_annual.nc: evapotranspiration rate.
•    Snowdz_annual.nc: annual mean snow depth.
•    Megabiome_annual.nc: megabiome classifications derived from the simulated vegetation. Megabiome codes are 1: tropical forest; 2: warm-temperate forest; 3: savanna and dry woodland; 4: grassland and dry shrubland; 5: warm desert; 6: temperate forest; 7: boreal forest; and 8: tundra and polar desert.

These files use the dimensions year, lat and lon; Veg_annual.nc additionally has a pft dimension. The archived PFT categories comprise bare ground, eight tree PFTs and two natural grass PFTs. PFTname identifies the categories in their stored order. Common mask variables (landfrac and icefrac) are included in each file.

SoilT_monthly_<start>_to_<end>.nc and SoilM_monthly_<start>_to_<end>.nc contain monthly soil temperature and soil water content, respectively. They use the dimensions year, mon, depth, lat and lon, with six archived soil-temperature levels and four soil-moisture levels. These files are divided into time intervals to limit file size.

The file naming follows a left-closed, right-open interval ([startyr, endyr)), i.e., startyr ≤ year < endyr. The decreasing BP coordinate follows the simulation forward in chronological time.

Factorial experiment outputs

Data_factorial.zip contains CTRL.nc, fixCO2.nc, fixP.nc, fixT.nc, fixTcold.nc and fixTwarm.nc. Each file contains 36 independent time-slice simulations. CTRL is the all-forcing snapshot control. The other experiments hold atmospheric CO₂, precipitation, temperature, coldest-quarter temperature or warmest-quarter temperature, respectively, at the LGM conditions specified in the experimental design.

For each time slice, a randomly selected 20-year forcing sequence is repeated to drive 200 model years to reach equilibrium for vegetation. Archived variables summarize the final 20 model years. The time coordinate identifies the 36 TraCE-21K-II forcing segments; it does not represent 36 consecutive years. The segment chronology and experimental settings are described in the accompanying manuscript. Factorial comparisons document sensitivity to the prescribed forcing changes and should be interpreted in the context of this snapshot protocol.

Code and configuration

ORCHIDEE.zip contains the model source code. The ORCHIDEE version used in this study is ORCHIDEE-MICT (r5308). It is open source and distributed under the CeCILL (CEA CNRS INRIA Logiciel Libre) license.  run.def records model configuration settings. bias_correction.py is the Python script for generating the bias-corrected climate forcing.

Animations

Video_treecover_10y.avi displays simulated tree-cover maps at 10-year intervals.

Video_megabiome_comparison_500y.avi displays maps at 500-year intervals, with simulated megabiomes in the upper panel, pollen-based megabiome reconstructions in the middle panel, and the corresponding Best Neighbor Score in the lower panel, similar to Figure 4 in the manuscript.

Evaluation and reuse

The archive contains one realization and thus does not quantify uncertainties associated with climate forcing, model structure and parameter choices. Evaluation against pollen-based reconstructions provides greater support for broad vegetation patterns and some long-term regional changes than for absolute tree cover or the timing of individual transitions. Moreover, the fine output frequency does not establish accuracy for individual years or months. See Technical Validation and Usage Notes in the accompanying manuscript for the evaluation results and guidance on reuse.

Cite this dataset using https://doi.org/10.5281/zenodo.19754917 and cite the associated article when available.

Files

Data_factorial.zip

Files (12.5 GB)

Name Size
md5:229cd3dd3e9ee2897fbe20784218a8a3
9.7 kB Download
md5:358f66011624155124c44df0544bf51e
29.7 MB Preview Download
md5:b2ccb18d99432e9c39f255bca6b6e10f
1.5 GB Preview Download
md5:ad3dcb1adb65cc803a9fe744dd26e150
67.2 MB Preview Download
md5:1b5a7f76665eb01d140a8d4c2139e006
13.0 MB Download
md5:9ccfc1df25fe50181b2179d8804008d1
620.6 MB Download
md5:5ecdc5c55c80e7f86512d94a823b4131
593.5 MB Download
md5:67a0aec6150abf5e334af01a7abb2989
593.2 MB Download
md5:3ac41f0841012da2f672ef56f6265894
545.6 MB Download
md5:20769ab83780508505971e3180e16d4e
533.2 MB Download
md5:f5b397aa84b7cb3e4338503a06ff5bc3
623.6 MB Download
md5:720ecfc107573dba02b6a5b952216dd2
628.8 MB Download
md5:b990ee1476d48dc1fdbf60d1b0ce2688
635.5 MB Download
md5:3c2ea5cec95d62d30ca0d3c58ee31f29
16.0 MB Download
md5:ce0fcdbbda5f8af519133e0cd379be00
758.2 MB Download
md5:9699808c0e1962e784cc5b36aedbabb3
742.9 MB Download
md5:09c2783f49915afb0703b36190f16148
740.3 MB Download
md5:4e6ef414f93effc9bb8f767969bce105
721.4 MB Download
md5:80a070a2d7dfa76a78696c411f5ed168
707.6 MB Download
md5:d6b6fd1ac60e06b72189443951d16ecd
778.0 MB Download
md5:67c7e71d9e91306cb2e0c5d968e1023b
775.8 MB Download
md5:564db34e446b66a8e522ea10269e4788
778.7 MB Download
md5:2dffbf61d041dc716a74489bd68fa9f7
154.1 kB Download
md5:da1e214ec6989ae6e9823757ca7d408e
8.1 MB Download
md5:215a5d7cd05f5cfcdf62f78ff3dea2a7
105.9 MB Download