Micrometeorological measurements of CO2, H2O and energy exchanges between agroforestry systems and the atmosphere
Doctoral thesis
Date of Examination:2025-09-26
Date of issue:2025-12-11
Advisor:Prof. Dr. Alexander Knohl
Referee:Prof. Dr. Alexander Knohl
Referee:Prof. Dr. Ivan Mammarella
Referee:Prof. Dr. Annette Menzel
Files in this item
Name:PhD_thesis_CallejasRodelas_JoseAngel.pdf
Size:12.4Mb
Format:PDF
Abstract
English
Alternative land-use systems, such as agroforestry (AF), provide nature-based solutions to mitigate the impact of agriculture on climate change. Despite their multiple environmental benefits, AF systems remain poorly studied. Particularly, ecosystem-scale measurements of carbon, water and energy exchanges between AF and the atmosphere are rare and only few studies exist on the matter. The eddy covariance (EC) technique was established during the past decades as the major tool to directly quantify flux density exchanges of energy, momentum and gas species, such as CO2, over terrestrial ecosystems. Nevertheless, the requirements of EC translate into large equipment and operational costs, which impedes its implementation in numerous ecologically and sociologically important ecosystems. Moreover, spatial heterogeneity over many ecosystems compromises the reliability of EC measurements, as the flux densities quantified from a single station can only be attributed to a certain footprint area and not to the whole ecosystem. Lower-cost EC (LC-EC) setups can offer a solution to both the lack of representativeness of conventional, single tower measurements over heterogeneous sites, and to the lack of measurements over important biomes and eco-regions. This thesis was framed within a long term research project, called SIGNAL. This project consisted in the comparison of AF to open cropland (OC) or open grassland (OG) across several locations in northern Germany, with the objective to evaluate multiple ecosystem functions related to soil nutrient cycles, biodiversity abundance or yield, among others. This thesis was performed as part of the atmospheric branch of SIGNAL, that was related to the quantification of carbon sequestration, evapotranspiration and water use efficiency of AF systems, across a total of five study regions, one pair AF/OG and four pairs AF/OC. Each AF, OC or OG was equipped with an EC station and ancillary meteorological instrumentation. Concretely, the objective of this thesis was to perform, first, a validation and characterization of LC-EC setups, in comparison to conventional EC (CON-EC); and then, to study ecosystem-atmosphere energy, CO2 and H2O exchanges over the study regions. This thesis was divided in four chapters, which are listed below. First, three LC-EC setups, similar to the ones that were installed at all EC stations at the AF and OC/OG sites, were compared to a CON-EC setup. The validation was done on CO2 flux densities (FC) and latent heat flux densities (LE). The LC-EC setups performed well in comparison to CON-EC (slopes ranging from 0.95 to 1.05 for FC and from 0.78 to 0.99 for LE, while r2 coefficients were between 0.7 and 0.92). The errors arising from the LC-EC were characterized in detail. The main source of uncertainty from LC-EC was attributed to the larger need for spectral corrections. However, the setups were successfully validated, which was considered as the base work for the other sections of the thesis. Second, the spatial heterogeneity in FC, LE, sensible heat flux densities (H) and turbulence parameters was quantified across one of the AF sites. This was done by installing a distributed network of three stations equipped with LC-EC setups, following their previous validation. The temporal variability of all parameters was demonstrated to be larger than the spatial variability. However, during some periods of the growing season and after harvest of some of the crops, the spatial variability dominated, particularly for FC. Coefficients of variation were the largest for FC (values between 0.3 and 0.4), followed by LE (values between 0.2 and 0.3), and were largest for the western wind sectors due to the largest degree of heterogeneity arising from the differential crop distribution in between tree stripes. The main message from this study was that the distributed network of three LC-EC stations improved the statistical representation of flux densities across the AF site, due to its large spatial heterogeneity. Third, the processing routines developed and implemented during the previous two sections of the thesis were extended and applied to all data corresponding to a total of five pairs of AF and OC or OG regions in northern Germany. Meteorological data from the first, second and third phases of the SIGNAL project were compiled and gap-filled using ERA5-Land data as a reference. Eddy covariance flux densities (FC, LE and H) were compiled for the second and third phases of the SIGNAL project, together with other turbulence parameters, and gap-filled using a combination of marginal distribution sampling and Extreme Gradient Boosting. Finally, FC was partitioned into gross primary production (GPP) and ecosystem respiration (RECO). Hence, we produced a harmonized, completely gap-filled dataset from a total of ten stations. Two of them corresponded to a paired grassland AF and OG, while the remaining eight corresponded to four pairs of cropland AF and OC. Both GPP and RECO were larger over AF than OC/OG. The difference between both is an indicator whether the site is a sink or a source of CO2. The net ecosystem carbon uptake (FC) was greater more than half of the site-years at the AF sites, compared to OC or OG. We did not find a consistent pattern for evapotranspiration, since the difference between AF and OC/OG was small. The uncertainty in the gap-filling methods was evaluated and considered appropriate for the different datasets. Finally, general guidelines about how to compare newer, lower-cost EC setups were provided as a comment paper. The aim was to establish a robust, standard way to validate LC-EC against CON-EC, based on the previous work developed during the thesis. This thesis section was motivated by a previously published paper which compared a newer LC-EC setup to a CON-EC setup. In conclusion, the work presented in this thesis provided (i) a technical validation of LC-EC setups; (ii) an implementation of the LC-EC setups to quantify spatial variability and statistical robustness of EC measurements over one AF site; (iii) a harmonized, gap-filled dataset on EC and meteorological parameters measured over several AF and OC or OG sites, together with the data processing routines; (iv) general guidelines and recommendations for EC instrumental validation. This thesis contributed to the development of LC-EC setups, establishing guidelines on how to perform instrumental validation; provided examples on the LC-EC implementation; established a robust processing routine from the raw, high-frequency data until final gap-filled and partitioned flux density time series; and provided a harmonized dataset expanding seventy eight site-years over several AF and OC/OG sites in northern Germany.
Keywords: Micrometeorology; lower-cost eddy covariance; evapotranspiration; agroforestry
