Master Thesis
Quality assessment tool for flux measurements under changing land use conditions
Wolfgang Babel (10/2008-03/2009)
Support: Thomas Foken
Spatial heterogeneity has a strong influence on the reliability of meteorologically measured fluxes which transfer energy and matter between a given type of land surface and the atmosphere. Some QA/QC techniques and respective tools already exist to evaluate the quality of such fluxes from heterogeneous landscapes. In contrast, the aim of this thesis is to lay a foundation for a framework, how the footprint concept can contribute to estimate spatial representativeness of flux data used for upscaling. This specific goal is motivated by the recently launched projects CEOP-AEGIS and MESO-TiP, both aiming at water balances and ecologic processes on a regional scale over the Tibetan Plateau. In this context a decision supporting scheme has to be developed whether measurements sufficiently represent fluxes from a certain target area for different grid sizes or additional modelling is needed to operate the desired flux conversion. This decision is not trivial as often (typically during nighttime) high deviations between the source area and the target area come along with low fluxes diminishing its relevance for some purposes. Therefore deviances of a measured flux from the target area flux are discussed in principle. Furthermore, model experiments are carried out with observed flux data (sensible heat flux, latent heat flux and momentum flux) and a simple artificial landscape, containing only two types of land-use. The source weight functions required to estimate the contribution of a certain land-use type are calculated from a forward Lagrangian stochastic model. The investigated datasets stem from the LITFASS experiment 2003 in Lindenberg, Germany, covering a period of one month during the growing season from maize fields and rye fields, respectively. Additionally the performance of SVAT model runs (SEWAB) are evaluated and compared with the observed flux deviance in general. Observed fluxes are then assigned to the target and surrounding land-use of the artificial landscape and comparisons are conducted dependent on the land-use contribution of the fluxes. Due to the pronounced relationships between flux magnitude and the size of its source area, integration of the footprint concept offers a promising approach to quantify flux errors over heterogeneous terrain. Quantifying this relationship is helpful to estimate an effective influence of fluxes from adjacent land-use on the target flux. As one would expect, parameterisation remains the crucial step for SVAT modelling, therefore the findings for model perfomance cannot be transferred to other case studies, as following specific requirements complicate their usage: Model performance on low fluxes turned out to be decisive, while calibration usually focuses on large fluxes and the planned extrapolation of the model to unknown surrounding fluxes require a sound reproduction of the ongoing processes. Facing those problems a more statistical approach to quantify and correct footprint related flux errors whenever possible is seen to be more promising than flux correction by modelling the surrounding flux.