The project Forest Ecosystem Responses to Climatic Drivers aims at understanding time lags between climatic drivers and the respective ecosystem responses in terms of net ecosystem productivity (NEP) at the two Swiss Fluxnet sites Davos and Lägeren. The project has a highly interdisciplinary character and brings together detailed knowledge from plant physiology, forest ecology and meteorology to disentangle the effects on NEP of (i) actual physical drivers, and (ii) biotic conditions determined by past and recent climatic conditions. Understanding the natural processes determining the carbon balance of forest ecosystems is of great global interest for estimating country-specific carbon budgets within the United Nations Framework Convention on Climate Change (UNFCCC). The topic is a current hotspot at which precise questions from politics meet the incomplete knowledge of various environmental science disciplines. We hypothesize that the current year NEP is significantly driven by climatic drivers of the recent past (month to years) and not only by present conditions as typically assumed. Furthermore we hypothesize that observed time lags between climatic drivers and NEP are due to storage dynamics of carbon and water in trees. Our specific aims are to (i) identify climatic drivers of NEP at two contrasting forest sites, (ii) to quantify the impact of the climatic drivers on observed time lags of ecosystem responses, and (iii) to assess the underlying physiological mechanisms explaining such time lags and therefore flux partitioning. In order to address these aims, we plan: o to measure microclimate profiles, eddy covariance net ecosystem CO2 and H2Ovapor exchange, continuous stem radius changes, and (CO2) in tree stems at two Swiss Fluxnet sites Davos and Lägeren. o to analyse statistical patterns of historic time series which quantify the time-dependent weight of climatic drivers on ecosystem responses. o to compare the results of two forest types from the Swiss Fluxnet sites Davos and Lägeren. o to test the applicability of physiological concepts to explain the observed time lags and thus flux partitioning of CO2 and H2Ovapor fluxes at the ecosystem level. Our long-term forest ecosystem research sites (Davos since 1997; Lägeren since 2005) are predestined locations to address topics that depend on temporally and spatially highly resolved field data at different integration levels with long-term records.
This repository contains all the data used for the article "Monitoring cropland daily carbon dioxide exchange at field scales with Sentinel-2 satellite imagery" by Pia Gottschalk, Aram Kalhori, Zhan Li, Christian Wille, Torsten Sachs. The data are used to exemplify how ground measured CO2 fluxes of an agricultural field can be linked with remotely sensed vegetation indices to provided an upscaling approach for spatial CO2-flux projection.
The provided data form the basis for running the data processing scripts sequentially for (re-)producing all statistical analyses, results and figures in the article. The data are given in the formats as used in the data-processing scripts written in R, MATLAB and JavaScript of Google Eearth Engine. All codes for processing the data and a workflow description can be found here.
The dataset covers three types of data: half-hourly eddy covariance (EC) data, satellite derived vegetation indices and GIS/image data.
Continuous EC CO2 fluxes (03/2020 - 08/2023) are measured at the agricultural site "Heydenhof" in Northeastern Germany. The data file is provided in .mat (MATLAB) format containing the standard EddyPro software output variables which are described in an accompanying meta data file. The land use information used for footprint modeling is included as .jpeg and .png-files for visulisation and as .mat-file to be used for running the footprint modeling script. Sentinel-2 vegetation indices are provided as .csv files. These files are provided for convenience and version control only as the JavaScript for generating Sentinel-2 derived vegetation indices in Google Earth Engine is provided in the associated code repository. Here, the field boundaries are provided as shape file.
Data file description:
"HEY_LandUse_image.mat": MATLAB file in raster format, containing the land use codes in a 4x4 km raster with a resolution of 1 m used for running the Korman-Meixner foot print model for flux source area attribution.
"meta_data_HEY_LandUse_image.txt": description of land use codes used in the "HEY_LandUse_image.mat"
"HEY_LandUse_image.png": Visualisation of HEY_LandUse_image.mat. Figure A2 in manuscript. Showing the land use distribution around the measurement tower encoded in the number of land use classes used for foot print modeling.
"HEYDENHOF.jpeg": Visualisation of land use classes from digitisation. Auxiliary information. Showing the land use distribution around the measurement tower.
"HEY_FluxData_20200304_20220824_all_data.mat": MATLAB data file containing the half-hourly EC measurements plus auxiliary meteorological variables from 04/03/2020 to 24/08/2022 in matrix format with rows being the half-hourly measurements and including the unique time identifier "Timestamp", and "NaN" as missing data value.
"meta_data_HEY_FluxData.txt": text file accompanying "HEY_FluxData_20200304_20220824_all_data.mat" containing the variable names, units, format, range and description for the variables of "HEY_FluxData_20200304_20220824_all_data.mat"
"TERENO_prec_data_2020_2022.csv": comma separated text file containing the half-hourly precipitation data for the measurement site (HEY) from 01/01/2020 to 13/10/2022.
"meta_data_TERENO_prec.txt": text file accompanying " TERENO_prec_data_2020_2022.csv " containing the variable description of the TERENO precipitation data.
"HEY_tower_field.zip": zipped shape file outlining the agricultural field used as source area for the satellite data retrieval.
"S2.csv": comma separated text file containing the vegetation indices from Sentinel-2 for the agricultural field from 02/03/2020 to 29/08/2022.
"meta_data_Sentinel2_S2.txt": text file accompanying "S2.csv" containing the variable description of Sentinel-2 derived vegetation indices, i.e. "S2.csv".
"S2_SD.csv": comma separated text file containing the standard deviation of the vegetation indices for the agricultural field from 02/03/2020 to 29/08/2022.
"meta_data_Sentinel2_S2_SD.txt": text file accompanying "S2_SD.csv" containing the variable description of the standard deviation for the Sentinel-2 derived vegetation indices.
This repository provides the code used for the article "Monitoring cropland daily carbon dioxide exchange at field scales with Sentinel-2 satellite imagery" by Pia Gottschalk, Aram Kalhori, Zhan Li, Christian Wille, Torsten Sachs. The data are used to exemplify how ground measured CO2 fluxes of an agricultural field can be linked with remotely sensed vegetation indices to provided an upscaling approach for spatial CO2-flux projection.
The repository contains the codes produced for the article "Monitoring cropland daily carbon dioxide exchange at field scales with Sentinel-2 satellite imagery" by Pia Gottschalk, Aram Kalhori, Zhan Li, Christian Wille, Torsten Sachs. In this article, the authors present how local carbon dioxide (CO2) ground measurements and satellite data can be linked to project CO2 emissions spatially for agriculutral fields.
The codes are provided for
- footprint analysis and raw flux data quality control (MATLAB codes);
- retrieving Sentinel-2 vegetation indices via Google Earth Engine (GEE code);
- subsequent quality control, gap-filling and flux partitioning following the MDS approach by Reichstein et al. 2005 implemented by the R-package "REddyProc" (R codes);
- statistical analyses of combined EC and Sentinel-2 data (R codes);
- code for all figures as displayed in the manuscript (R codes).
This software is written in MATLAB, R and JavaScript (GEE). Running the codes (R and .m files (Code)) and loading the data files (CSV files and .mat files (Data)) requires the pre-installation of [R and RStudio] (https://posit.co/downloads/) and (MATLAB). The GEE script runs in a browser and can also be opened/downloaded here: https://code.earthengine.google.com/858361ae4aac7c3fe5227076c9733040
The RStudio 2021.09.0 Build 351 version has been used for developping the R scripts. The land cover classification work was performed in QGIS, v.3.16.11-Hannover. Data were analyzed in both MATLAB and R; and plots created with R (R Core Development Team 2020) in RStudio®.The R codes in this repository contain a suite of external R-packages ("zoo"; "REddyProc"; "Hmisc"; "PerformanceAnalytics") which are required for data analysis in this manuscript. The data to run the codes are published with the DOI https://doi.org/10.5880/GFZ.1.4.2023.008 (Gottschalk et al., 2023).