Göckede, M*; Thomas, C; Markkanen, T; Ruppert, J; Mauder, M; Foken, T: On the sensitivity of Lagrangian stochastic footprint modeling to within canopy flow statistics derived from wavelet analysis
Poster, 1st iLEAPS Science Conference, Boulder, CO, U.S.A.: 2006-01-21 - 2006-01-26

As the locations of flux towers which are part of flux monitoring networks such as FLUXNET (e.g. Baldocchi et al., 2001 [1]) have to be chosen because of their ecological importance, in many cases micrometeorological aspects play only a minor role when setting up a site. Hence, these sites are often located in complex terrain and are characterized by heterogeneous land cover composition with short fetches over the target land use type, conditions that compromise the collection of high quality meteorological data sets. The application of the eddy-covariance technique, which is commonly used for flux determination (e.g. Aubinet et al., 2000 [2]; Baldocchi et al., 2000 [3]), requires therefore a site dependent quality control to allow for a robust interpretation of the flux measurements (Foken et al., 2004 [4]). A key component in the data quality protocol are footprint models that determine the spatial context of a measurement by defining a transfer function between sources or sinks of the signal and the sensor position. The derived source area is crucial for the interpretation of micrometeorological data sets, e.g. by determining the fetch requirements under changing atmospheric stability regimes or by assessing the influence of distorting terrain elements on the measurements.
A powerful tool to model source areas over forest are Lagrangian Stochastic footprint models (e.g. Baldocchi, 1997 [5] Rannik et al., 2000 [6]; 2003 [7]), because this technique allows the consideration of horizontally heterogeneous flow conditions, effects of canopy flow on the measured fluxes, and a more realistic treatment of diffusion. As drivers, Lagrangian stochastic models use characteristics of prevailing turbulence to calculate trajectories of individual air parcels, such as the profiles of the mean wind speed u, the wind fluctuations (sigmau; sigmav; sigmaw), or the dissipation rate of turbulent kinetic energy. However, as only few generally valid theories are known for the flow in the canopy space (e.g. Lee, 1998 [8]; Finnigan, 2000 [9]), for within canopy flow these parameters often have to be approximated with crude generalizations and certain ad hoc assumptions (Schmid, 2002 [10]).
This study aims at testing the sensitivity of a Lagrangian Stochastic footprint model to the input parameters describing the turbulent flow field, with a focus on the within canopy flow processes. A long-term dataset of turbulence measurements collected during the WALDATEM-2003 (WAveLet Detection and Atmospheric TurbulencE Measurements, see Thomas et al., 2004 [11]) experiment within and above a tall spruce canopy is used to extract detailed turbulence statistics as input parameters for the footprint model. This dataset includes a profile of sonic anemometers, with the highest instrument at 14 m above the canopy, which was used to obtain high-frequency time series of turbulent variables and to monitor the turbulent exchange. In addition, the main tower was equipped with vertical profiles of cup-anemometers and mean temperature and humidity probes. The turbulence structure in the lower atmospheric boundary layer was observed with a SODAR-RASS system located in a clearing 200 m away from the main tower. From these measurements, representative profiles of the input parameters required for the footprint modeling are derived by application of different filters to the original data set. We employed spectral analysis using a wavelet analysis tool (Thomas and Foken, 2005 [12]) to determine the exchange regimes based on the detailed analysis of coherent structures along the vertical profile. These analyses allow the characterization of several typical states of coupling and decoupling between the canopy space and the atmosphere. The resulting description of the turbulent flow field varies in both the spatial and temporal context, as statistics were derived separately for typical different exchange regimes and wind direction sectors.
Besides using the WALDATEM dataset which allows a detailed characterization of the turbulent flow regimes at the Waldstein Weidenbrunnen site, two additional simpler methods to describe the canopy flow regime will be applied for means of comparison. Firstly, the parameterization of the flow statistics presented by Rannik et al. (2003 [7]) will be used, which have been tested thoroughly for Lagrangian Stochastic footprint modeling. This dataset was derived by measurements at the Hyytiälä site in Finland, for a forest architecture that has characteristics significantly different from the Waldstein Weidenbrunnen site. Secondly, a model by Massman and Weil (1999 [13]) to parameterize the profiles of the flow statistics based on profiles of the leaf area index will be employed. The impact of the application of these different descriptions of canopy flow will be tested by comparing size and position of the source areas computed by the footprint model, as well as by the determined composition of land cover types within the source area and their correlation to the measured eddy-covariance fluxes.

[1] D.D. Baldocchi et al., 2001, Bull. Amer. Meteorol. Soc., 82, 2415-2435.
[2] M. Aubinet et al., 2000, Adv. Ecol. Res., 30, 113-175.
[3] D.D. Baldocchi et al., 2000, Boundary-Layer Meteorol., 96, 257-291.
[4] Th. Foken et al., 2004, in: Handbook of Micrometeorology, X. Lee et al., Eds. (Kluwer Academic Publishers), pp. 181-208.
[5] D.D. Baldocchi, 1997, Boundary-Layer Meteorol., 85, 273-292.
[6] Ü. Rannik et al., 2000, Boundary-Layer Meteorol., 97, 137-166.
[7] Ü. Rannik et al., 2003, Boundary-Layer Meteorol., 109,163-189.
[8] X. Lee et al., 1998, Agric. For. Meteorol., 91, 39-49. [9] J.J. Finnigan, 2002, Ann. Rev. Fluid Mech., 32, 519-571.
[10] H.P. Schmid, 2002, Agric. For. Meteorol., 113, 159-183.
[11] C. Thomas et al., 2004, Univ. Bayreuth, Abt. Mikrometeorologie, Arbeitsergebnisse, Print ISSN 1614-8916, 24, 59 pp.
[12] C. Thomas and Th. Foken, 2005, Boundary-Layer Meteorol., submitted.
[13] W.J. Massman and J.C. Weil, 1999, Boundary-Layer Meteorol., 91, 81-107.

last modified 2006-01-23