The Land Surface Model (LSM) should have reliable forcing, validation, and surface attribute data as the foundation for effective development and improvement of the model. The vortex covariance flux tower data is considered as the benchmark data for LSM. However, currently available flux tower datasets often require multi-faceted processing to ensure data quality before being applied to LSM. More importantly, these datasets lack on-site observational attribute data, which limits their use as benchmark data. Here, we conducted a comprehensive quality screening of the existing reprocessed flux tower dataset, including gap filling data ratio, external interference, and energy balance closure (EBC), and ultimately identified 90 high-quality sites. For these sites, we collected vegetation, soil, terrain information, and wind speed measurement heights from literature, regional networks, and biological, auxiliary, interference, and metadata (BADM) files. Then, we obtained the final flux tower attribute dataset through global data product supplementation and classification of plant functional types (PFTs).
The data used in this study can be divided into four groups. Firstly, PLUMBER2 serves as the dataset for data quality screening. The second group consists of attribute sources, including 113 literature related to the site, 7 flux area networks, and biological, auxiliary, interference, and metadata (BADM) files provided by FLUXNET and AmeriFlux.
The website is: a: https://ameriflux.lbl.gov/
b： http://www.biomet.co.at/
c ： http://www.chinaflux.org/
d： http://www.europe-fluxdata.eu/
e ： https://www.gml.noaa.gov/
f： https://ozflux.org.au/
g： https://www.swissfluxnet.ethz.ch/. ）
To establish the final dataset, we mainly took three processing steps: location and time period selection, attribute collection, and data processing. Firstly, the data selection process involves selecting years with low filling rates for flux (latent and sensible heat) and vapor pressure difference (VPD) gaps, while removing years that are subject to interference. The years of latent heat and sensible heat, as well as vapor pressure deficit (VPD), while excluding stations affected by external interference and unable to pass through EBC. Subsequently, we collected vegetation, soil, and terrain data observed on site. Vegetation attributes include FVC, maximum LAI, and average canopy height. Soil properties include soil texture, bulk density, and organic carbon concentration. The terrain attributes include altitude, slope, and aspect. In addition, the reference measurement height (used to simulate the lowest layer of the atmospheric model, with which LSM will be coupled) is revised based on wind speed measurement height whenever possible. Then, we supplemented vegetation attributes and soil texture with global data. Finally, FVC is further subdivided into different PFTs.
The model simulation shows that there is a significant difference in the output between the observed attribute data on site and the default values of the model, highlighting the crucial role of observed attribute data on site and increasing the importance of LSM boundary on flux tower attribute data. This dataset to some extent solves the problem of lack of on-site attribute data, reduces the uncertainty of LSM data sources, and helps diagnose parameter and process defects.
This work is licensed under
CC BY 4.0 (Creative Commons Attribution 4.0 International License).
CSTR:11738.11.NCDC.ZENODO.DB6474.2024
10.5281/zenodo.10939725
