Ament, Felix: Energy and moisture exchange processes over heterogeneous land-surfaces in a weather prediction model. - Bonn, 2006. - Dissertation, Rheinische Friedrich-Wilhelms-Universität Bonn.
Online-Ausgabe in bonndoc: https://nbn-resolving.org/urn:nbn:de:hbz:5N-08348
@phdthesis{handle:20.500.11811/2656,
urn: https://nbn-resolving.org/urn:nbn:de:hbz:5N-08348,
author = {{Felix Ament}},
title = {Energy and moisture exchange processes over heterogeneous land-surfaces in a weather prediction model},
school = {Rheinische Friedrich-Wilhelms-Universität Bonn},
year = 2006,
note = {Land-surfaces exhibit significant variablity at very small scales - in contrast to the atmosphere, where horizontal diffusion reduces small scale fluctuations effectively. It is a challenging task for numerical weather prediction (NWP) to account for these different characteristics while calculating exchange fluxes between these two systems: Surface processes need to be considered with higher spatial resolution than atmospheric effects and high resolution initial conditions and parameters of the surface are required. This study evaluates methods to solve these surface heterogeneity problems on the basis of integrations of the non-hydrostatic weather prediction model Lokal-Modell (LM) both in a NWP configuration (grid spacing of 7 km) and in a regional climate model set up (grid spacing of 21 km). The runs are performed for the 30-day periode of the LITFASS-2003 experiment.
Two heterogeneity parameterisation schemes, the mosaic and tile approach, have been implemented into LM. Both methods decompose the surface within one atmospheric grid box into several patches to resolve subgrid scale variability. The mosaic approach utilises an explicit, geographical sub-grid, whereas the tile approach subdivides the surface according to a certain criteria, e.g. land-use. In general, the tile method requires less computational time since fewer patches are used. However, the mosaic technique is more flexible since it takes multivariate heterogeneity into account.
Two major model enhancements are needed to simulate the observed exchange fluxes during LITFASS-2003 successfully: land-use dependent stomatal resistance parameters and vegetation albedo, and the use of accurate soil moisture data for initialisation. The latter is obtained by multi-year assimilation runs of the soil module of LM driven exclusively by observations. This technique ensures a balanced model state and allows to capture heterogeneity effects due to soil moisture variations induced by inhomogeneous rainfall.
The flux predictions of all integrations using these enhancements agree well with the observations within the range of measurement uncertainty independently from the representation of heterogeneity. The impact of improved surface fluxes on forecasts of atmospheric state variables is beneficial. Using high resolution integrations (e.g. grid spacing of 1 km) as reference, a clear ranking of parameterisation schemes can be established: The mosaic approach leads to very accurate flux predictions, followed by the tile approach, and the operational homogeneous approach. The deviations in forecasted surface fluxes of all methods decay significantly, if averages over larger scales are considered. The ranking of the methods can be explained by analysing the small scale variance of high resolution runs: The variance of surface quantities is by far larger than those of corresponding atmospheric quantities. This supports the assumption inherent to the mosaic and tile approach to refine the surface only. During LITFASS-2003, a considerable fraction of flux variability is explained by soil moisture variations which are not correlated with land-use. These subgrid scale heterogeneities can only be resolved by the mosaic approach and not by a tile scheme.},

url = {https://hdl.handle.net/20.500.11811/2656}
}

The following license files are associated with this item:

InCopyright