This page has the information of how to access and use the FLUXCOM energy fluxes data products. The data re publicly available for download from Data Portal of the Max Planck Institute for Biogeochemistry. All data download will be handled through the portal.
The native FLUXCOM energy flux products amount to more than 4 TB of data. Products with daily or 8-daily temporal resolution or customized ensemble estimates are available on request to Martin Jung (email@example.com).
The overview paper is under consideration in Scientific Data. A preprint of the manuscript is available at arXiv.
Monthly energy flux data of all ensemble members as well as the ensemble estimates from the FLUXCOM initiative are freely available (CC4.0 BY licence) from the Data Portal after registration. Choose ‘FluxCom’ in the dropdown menu of the database and select FileID 257.
The users will be provided with an access to a ftp server.
FTP Directory Structure
The ftp directory stores 214 GB of data and is structured in a consistent way with the file naming in Table below. The folder structure was designed to facilitate easy download of relevant subsets of the archive and is as per the following convention:
SETUP/TYPE/sReso or METEO/tRESO"
- SETUP is either “RS” or “RS_METEO”.
- TYPE is either “ensemble” or “member”.
- At the third level,
- RS uses sRESO (“720_360” or “4320_2160”)
- RS+METEO uses METEO (see Table below).
- tRESO is always “monthly” here.
File Naming Conventions
The files are provided in network Common Data Form, version 4 (netCDF-4) data format. The data files are named using the following convention:
Key to naming convention used for the folder structure and naming conventions.
|1||EF||Energy flux||no prefix||- LE, H, Rn|
|2||SETUP||Upscaling set up||no prefix||- RS: for RS products
- RS_METEO: for RS+METEO products
|2||EBC||Energy balance correction||EBC-||- ALL: for ensembles that include both energy balance corrected and uncorrected fluxes
- NONE: uncorrected energy fluxes
- BWR: energy fluxes corrected by Bowen’s ratio method
- RES: energy fluxes corrected by residual method
Note that EBC is always NONE for Rn, because it was never corrected.
|3||MLM||Machine learning method||MLM-||- ALL: for ensembles that include energy fluxes from all machine learning methods
- ANN, MARS, or RF for energy fluxes from RS+METEO
- Rfmiss, ANNnoPFT, MARSens, GMDH_CV, KRR, MTE, MTE_Viterbo, MTEM, or SVM for energy fluxes from RS
|4||METEO||Meteorological data||METEO-||- ALL: for RS+METEO ensembles that include energy fluxes from all meteorological data
- CERES_GPCP, CRUNCEP_v8, GSWP3, or WFDEI: for RS+METEO products using different meteorological data
- NONE: for all energy fluxes from RS (because RS does not use meteorological forcing data)
|5||sRESO||Spatial resolution||no prefix||- 720_360: for 0.5°x0.5° native RS+METEO and spatially aggregated RS data
- 4320_2160: for 0.0833°x0.0833° native RS data
|6||tRESO||Temporal resolution||no prefix||- monthly: for all data files|
|7||YYYY||Year||no prefix||- the year for which the data is|
Usage Notes for Energy Fluxes
For cross-consistency analysis and evaluation of LSM simulations we suggest focusing on spatial patterns of mean annual and mean seasonal fluxes.
For comparison with offline LSM simulations we recommend using products from the RS+METEO setup forced with corresponding meteorological forcing to minimize deviations due to different climate input data. However, as RS+METEO inputs prescribe seasonal and spatial land surface properties--through mean seasonal cycles and PFT-based tiling respectively--full consistency with forcing-specific LSM simulations cannot be achieved.
Patterns of interannual variations in these products are expected to be more uncertain than spatial patterns of mean annual or seasonal fluxes.
The energy flux densities of FLUXCOM product are defined per vegetated area in each grid cell. This needs to be considered when calculating global and continental budgets, in particular for sensible heat and net radiation where these fluxes have sizeable magnitudes over non-vegetated areas.
The land fraction provided with the FLUXCOM data should be multiplied with the flux densities for a correct accounting of fluxes over land.
FLUXCOM ensemble products provide the median absolute deviation of ensemble members per grid cell and time step which might be scaled to a robust estimate of the standard deviation of a normal distribution by multiplying by 1.4826.