PhD Thesis
Site-specific modelling of turbulent fluxes on the Tibetan Plateau
Wolfgang Babel (04/2009-05/2013)
Support: Thomas Foken
The Tibetan Plateau attracts attention in recent decades due to its influence on the East-Asian Monsoon and regional hydrology. Therefore estimates of the regional energy and water balance have come into the focus, utilising remote sensing and regional model approaches, but such attempts require surface-specific flux data of high quality for validation. Eddy-covariance measurements are qualified for this task, but these are scarce on the Tibetan Plateau, incomplete due to quality filtering and potentially biased due to the well-known closure gap of the observed energy balance as well as small-scale heterogeneity. This thesis is related to the infrastructural EU project CEOP-AEGIS, aiming at a standardised processing of eddy-covariance data – including correction of the energy balance closure and gap-filling – on the Tibetan Plateau. In a pre-analysis step, particular issues about data quality of turbulent fluxes (sensible heat flux and latent heat flux/evapotranspiration) at Tibetan Plateau sites have been addressed. One of them is the degradation of data quality due to the frequent occurrence of near-ground free convective conditions. Another issue arises from coordinate rotation for non-omnidirectional sonic anemometer, which requires a careful handling. In consequence, a sector-wise planar-fit is recommended, disregarding the sector influenced by the anemometer's mounting structure. This can reduce occurrences of invalid momentum flux data, whilst no effect on scalar fluxes can be seen. As a main topic, this thesis investigates the application of process-based modelling to estimate turbulent flux exchange between the surface and the atmosphere for typical surface types on the Tibetan Plateau. Therefore a case study has been carried out at Nam Co, Tibetan Plateau. Turbulent flux measurements over dry and wet grassland as well as over a shallow lake have been conducted during the summer monsoon season of 2009, and modelled with the land surface scheme SEWAB and a hydrodynamic multilayer model for the lake. Adaptations were implemented to the land surface scheme with regard to the special conditions on the Tibetan Plateau, such as extreme diurnal variation of surface temperature and variation in soil moisture, further called TP version. The analysis includes a consequent model comparison with eddy-covariance data, using model parameters derived independently rather than applying optimisation strategies. Specific attention has been devoted to the impact of observed energy balance closure and its correction, establishing a new correction method according to the Buoyancy flux. The land surface model reasonably represented the dry and the wet grassland site by only setting the site-specific model parameters, and the TP version performed overall better than the original version, while laboratory measurements of soil parameters failed to improve model performance in comparison to standard parameter values. Soil temperature and moisture measurements as well as field based knowledge of the soil type have been identified as minimum requirements for model parameter acquisition. Lake surface fluxes have been modelled reliably, the lake depth has been taken into account. These results can be transferred to any lake on the Tibetan Plateau given the required forcing data including a representative lake surface temperature. The choice of the surface model and the selection of the energy balance closure correction method are inter-related problems. The correction partitions the balance residual to the sensible and latent heat flux. This can be typically done according to the Bowen ratio, or according to the presented new method which attributes a larger fraction to the sensible heat flux. Testing both methods leads to partly ambiguous model performance, especially with respect to the used parameter sets. It clearly leads to shifts in model bias, while the R² metric suggests higher model compatibility to the Bowen ratio correction. The latter agrees with previous findings with respect to SEWAB modelling, but is in contradiction with recent experimental findings, attributing the closure gap to secondary circulations, driven by buoyancy. Future research on model structure should account for such processes. As expected, the flux measurements showed distinct differences between the investigated land use types in magnitude and dynamics. The used models were able to resolve these differences in general with contrasts between surface types exceeding model errors. This must be considered when validating regional flux estimates with eddy-covariance data from the dry Nam Co station. The findings from this thesis provide the basis to process eddy-covariance data on the required level as described above.