PHYSICAL MODELING OF A FINITE FOREST AREA TRANSPORT OF BIOGENIC EMISSIONS

Transcription

1 International Conference Wind Effects on Trees September 6-8,, University of Karlsruhe, Germany PHYSICAL MODELING OF A FINITE FOREST AREA TRANSPORT OF BIOGENIC EMISSIONS Sandrine Aubrun, Bernd Leitl, Michael Schatzmann University of Hamburg, Meteorological Institute, Bundesstrasse, D-6 Hamburg, Germany Abstract An improved concept for the physical modeling of a forest area has been designed in order to characterize the flow structure in the model of the finite and inhomogeneous forest area of the Research center of Jülich. The turbulence properties of the approach flow and of the canopy flow are quantified and compared with guidelines or previous studies. The comparison with other dense forest canopies is satisfying since the mean and turbulent velocity profiles and the spectral distributions above and inside the canopy are in agreement with field data. Thus, the profiles of integral length scales through the forest canopy show similar evolutions to those found in literature. Since the quality of the physical modeling has been proved, this model of a finite forest area can be used for dispersion applications in order to provide information about the origin, the trajectory and the age of the emissions sampled at the measurement towers during field campaigns (presented orally). Introduction Forests are comple sources of biogenic Volatile Organic Compounds (VOC) in the planetary boundary layer. Biogenic VOC contribute significantly to the formation of photo-oidants in the troposphere. Due to the fast vertical transport processes, they may even have an impact on the chemistry of the upper troposphere. The objective of the interdisciplinary project ECHO (Emission and Chemical transformation of biogenic volatile Organic compounds), which takes part in the German research program AFO (funded by the German Federal Ministry of Education and Research) is to provide a better understanding of forest stands as a comple source of reactive trace gases into the troposphere. The contribution of the University of Hamburg to this project is the physical modeling of the forest site in the wind tunnel in order to study the flow structure inside and above the forest canopy and thus, to understand its influence on the transport of chemical compounds. Using an improved concept for the modeling of a forest area, the presented work focuses on the model validation through comparison with field data. If no in-situ data are available, the comparison is performed with previous field or wind tunnel studies (Raupach et al. 986, Brunet et al. 99, Gardiner 99, Raupach et al. 996, Novak et al. ). The approach flow is fitted to the requirements given by the German guideline VDI 78/ (). The field site The chosen field site is the finite forest area surrounding the Research Center of Jülich (Germany). This deciduous forest stand is representative of a typical European forest area. It covers ha and is located in a farmland-type region (figure ). The inhomogeneity in the distribution of tree species, of tree age and of tree height is significant. As a consequence, a

2 careful tree inventory was carried out. The meteorological conditions are permanently recorded at the meteorological (meteo) tower at 7 stations located from to meters above ground. The meteo tower is located in a clearing. Since a redevelopment of the boundary layer occurs within this clearing, the measured aerodynamic properties are not representative of the forest canopy. Nevertheless, these measurements are also used for comparison with the wind tunnel measurements. In the framework of the research project ECHO, three additional measurement towers were built-up inside the forest in order to measure the meteorological conditions as well as the biogenic VOC concentrations inside and immediately above the forest canopy. The main tower (see location on figure ) is located in an area planted with oak and beech trees, which are years old. The average tree height in this area is between and m. The LeafArea-Inde is.6. Figure : The Research Center of Jülich and the surrounding finite forest. Left: Aerial view, right: physical model at a scale of :, in the large boundary layer wind tunnel of the Meteorological institute of Hamburg University. Eperimental set-up A model of the Research Center of Jülich and the surrounding finite forest was built at a scale of :. One of the most frequent wind directions (7, westerly wind) was studied. The complete model was. m long and m wide ( m and m in full scale respectively). The scale of : ensures the embedding of the upwind edges of the finite forest area in the model. Buildings were made from Styrofoam. By processing aerial views from the field site, the cartography of the tree height distribution was assessed. Si tree height Ht classes were defined (Elbers ). The modeled areas where Ht < m were covered homogeneously by irregular turning chips. This tangled material had a thickness lower than mm ( m in full scale). For the higher ranges, a specific arrangement of rings made from metallic mesh was used to simulate the forest stand. The metallic mesh was made out of steel wires of diameter. mm and a mesh size of.8 mm. The ring height was directly related to the average height of the modeled Ht class. On the uppermost third of the ring height, the mesh was bent twice in order to decrease the local porosity and consequently, thus simulating the influence of the vegetation at the tree crown. The aspect ratio and the arrangement of rings were analogous for all ring categories. This set-up was not designed with respect to shape, drag coefficient or Leaf Area Inde (LAI) similarities between field and model. The strategy was to achieve the same aerodynamic properties of the inside and above canopy flow, as measured in field by iterative preliminary testing (Aubrun et al. ). The model was placed in the large boundary layer wind tunnel of the Meteorological Institute of Hamburg University (figure ). The wind tunnel is characterized by a test section of length

3 8 m, width m and height. m. The boundary layer was initialized using Counihan-type spires at the entrance of the wind tunnel and roughness elements covering the first 7. meters of the test section floor. The roughness elements were metallic bluff obstacles of height z r = mm, width mm and thickness. mm. This set-up controls the properties of the flow, which approached the edge of the finite forest area. The fluid acceleration due to the presence of the model in the test section and the growth of the boundary layer was compensated by adjusting the ceiling. Flow measurements were performed with a D fiber-optic Laser-Doppler-Anemometer (FVA- LDA, Dantec ) with 8 mm focal distance. The flow was seeded with micro-particles of µm diameter. Zero- and one-order moments of the velocities were processed with at least samples and a minimum averaging time of seconds. The sampling frequency was related to the particle seeding but was usually around Hz. The structure of the approach flow No field data of the approach flow of the Jülich forest are available. Nevertheless, the surroundings of the finite forest area are farmlands and grasslands, which belong to moderately rough surfaces. Based on a compilation of field data, the German guideline VDI 78/ () provides requirements for the proper modeling of the neutral atmospheric boundary layer up to a height of m. For moderately rough surfaces, the power law eponent should be between. < α <. 8, the roughness length between. < z <. m and the displacement height d. These ranges for α and z subsequently parameterize the range of the turbulence intensity profiles. The properties of the modeled boundary layer were checked at.m (m in full scale) upstream of the edge of the forest area. The best fit to the measured wind profile with an eponential function was obtained with a power law eponent α =.9. The wind profile fits to a logarithmic function from z =.m (i.e.. zr ) up to z = 9m full scale. The etrapolation of the logarithmic law towards the zero-value of velocity gives the associated roughness length z =.m. The power law eponent and the roughness length are slightly higher than the authorized ranges. Knowing z and taking into account ESDU documents (98), a depth of the atmospheric boundary layer can be estimated: δ 6 m. It is commonly assumed that the depth of the surface layer is % of the total atmospheric boundary layer: δ s 9m. The vertical distribution of the shear stresses is constant within the surface layer (constant shear layer). The friction velocity is estimated to u * =. m/s (i.e. u * U δ =. ) by averaging the shear stresses measured in the range zr < z < δ s. The turbulence intensity profiles are enclosed within the ranges advised by VDI 78/ () for a height up to 9m. The spectra of turbulent fluctuations of the three velocity components measured in the wind tunnel at and m full scale above ground are presented in figures a-c. They are compared to the empirical model based on field measurements over a flat featureless terrain and published in Kaimal and Finnigan 99. The vertical fluctuations in the wind tunnel are slightly larger than those epected from the Kaimal model. The longitudinal integral length scale Lu was calculated applying the autocorrelation method on time series of velocity fluctuations at different heights above ground. The autocorrelation function is integrated in time until its first zero crossing. In figure d, the vertical distribution of Lu is compared to the empirical law proposed by Counihan 97. The agreement is satisfying.

4 a) b) -/ -/ f S uu /u * - Z = m Z = m Kaimal et al. (97) slope -/ f S vv /u * f z/u mean f z/u mean c) -/ 8 6 d) wind tunnel Counihan (97) f S ww /u * - Z [m] - f z/u mean Lu [m] Figure : Spectral density distributions of turbulence for the three velocity components of the approach flow a) longitudinal, b) lateral, c) vertical. d) Vertical distribution of the longitudinal integral length scale of the approach flow. Measured in the wind tunnel. The structure of the canopy flow As the flow encounters the forest area, the atmospheric boundary layer is redeveloping under very rough conditions due to the presence of trees. The inhomogeneity of tree distribution and tree height leads to a very comple configuration. Field time series of velocity were acquired from July to September 998 with ultra-sonic anemometers-thermometers with a sampling frequency of Hz at the meteo tower at the heights z =.9 and.. Only data characterized by a wind direction between Ht meteo 68 < 7 < WD, a mean velocity at z =. higher than m s - and neutral stability Ht meteo conditions. < z L <. (L is the Monin-Obukhov length) were selected. From the remaining hours of data, mean velocity and turbulence profiles, turbulence spectra and integral length scales were computed, using the identical methods as for wind tunnel data. At the main tower, -min average horizontal velocity and its standard deviation were measured with ultra-sonic anemometers-thermometers during hours on the th of September at z =.,.,.7 and. with a mean wind direction of 8 and Ht main near-neutral conditions. These selected field data are compared with the equivalent wind tunnel eperiments (figure ). The flow properties are non-dimensioned with the flow properties measured at the height z Ht =., which is the highest measurement location main in the field. In figure a, the field wind profiles are compared with those measured in the wind tunnel. The wind profile measured at the main tower location is typical for a dense forest canopy (Gardiner 99, Raupach et al. 996), since the velocity is very low and nearly constant within the canopy and has a very strong gradient in the upper part of the canopy. The field

5 data show similar properties but the mean velocity measured inside the canopy is even smaller. Since the meteo tower is located in a clearing, the lower part of the boundary layer is subject to a local redevelopment, which is underestimated in the wind tunnel. The vertical distribution of the standard deviation of the mean horizontal velocity (figure b) shows that the turbulence properties of the flow are completely different in the two distinct regions, inside the canopy and above the canopy. The inside region is dominated by the local effect of the forest whereas the above one is representative of an atmospheric boundary layer developing on a very rough surface. In both regions, the standard deviation stays nearly constant and the transition between these two states takes place between.8 < z Ht <.. For the reasons eplained previously, this feature is not so visible in the case of the meteo tower. The transition region stretches between. < z Ht <.. Longitudinal standard deviations, as well as the Reynolds stresses, measured in the wind tunnel and in field are in agreement. a) b) Main tower WT Main tower field Meteo tower WT Meteo tower field c) z/ht z/ht z/ht U/U.Ht.. Urms/Urms.Ht.. u'w'/u'w'.ht Figure : Vertical profiles measured at the meteo tower and at the main tower in field and in the wind tunnel. a) Dimensionless mean horizontal velocity, b) dimensionless standard deviation of the horizontal velocity c) dimensionless Reynolds stresses. Figure presents the spectral densities of turbulence for the three velocity components measured in the wind tunnel at the main tower location. Each figure presents a different measurement height. The spectra measured above the canopy (figures a and b) show the classical distribution epected in an atmospheric boundary layer. The / slope in the inertial sub-range proves the local isotropy of the turbulence scales. The dissipation subrange is not visible on spectra, since the dissipation occurs for higher time scales than those actually captured with the used sampling frequency. At the top of the canopy (figure c), the frequency range, where the inertial sub-range is isotropic, is reduced. The dissipation subrange starts at f Ht U =. Meanwhile, the energy-containing sub-ranges for the three Ht components tend to one similar distribution with a maimum of energy at f Ht U... This feature is confirmed in the spectra measured at z Ht =. Ht (figure d). The spectral distribution is similar for the three components and is characterized by a very narrow shape. The local isotropy of the inertial sub-range is completely destroyed. A very small range of scales centered on f Ht U.. drives the turbulence structure within the canopy. Ht

6 Gardiner 99 noticed this evolution of the spectral distribution from the above-canopy flow to the in-canopy flow in a very dense spruce forest (LAI,). In his study, the turbulence spectra also showed a shift of the maimum energy peak to higher frequencies within the canopy. Brunet et al. 99 studied a model of slightly dense waving wheat crop (LAI.7) and found a progressive reduction of the / slope range as z Ht decreases. On the other hand, they did not obtain very sharp-shape spectra within the canopy. The lower density of the canopy is probably responsible for this feature, since the coupling between the aboveand in-canopy is never completely damped. Novak et al. studied in a wind tunnel a forest model made of Christmas tree branches (LAI between. and.). The spectral distributions presented different behavior than the previous studies since their shape is not sharper as z Ht decreases. Furthermore, the slope at the end of the inertial sub-range is smoother inside the canopy than above. In Novak s study, the vertical component spectra showed an absence of the / slope in all cases and at all heights, which is attributed to the limitations of the physical modeling of the turbulence properties of the atmospheric boundary layer in the wind tunnel used. Figure presents the comparison of the longitudinal and vertical turbulence spectra measured at the meteo tower in field and in the wind tunnel. The field spectra represent averages processed from the selected hours. The plotted spectra show a / slope in the inertial sub-range but, in the wind tunnel, the peak of energy is shifted to higher frequencies. Spectra measured at z Ht =. show the same characteristics (not plotted). Since the meteo tower is located in a clearing, the spectra measured within the canopy are not characterized by a sharp shape distribution. a) -/ b) -/ f S ii /σ u with i = u, v, w f S ii /σ u with i = u, v, w - u v w slope -/ - f Ht/U c) - -/ u v w slope -/ - f Ht/U Ht ii u ii u - u v w slope -/ - f Ht/U d) - -/ u v w slope -/ - f Ht/U Ht Figure : Spectral density of the turbulence for the three velocity components at the main tower location measured in the wind tunnel. a) z Ht =. 8, b) z Ht =. 6, c) z Ht =. 9, d) z Ht =.. An other approach to characterize the turbulence properties of a flow is to study lateral and vertical deviations of the instantaneous velocity vector from the mean. Figure 6 presents the histogram of the lateral and vertical deviations measured in field and in the wind tunnel. The lateral deviations are perfectly reproduced in the wind tunnel, whereas the vertical ones are overestimated. One reason could be that the geometry of the metallic rings used to model

7 the forest area offers no blockage in the vertical direction, while real trees. Histograms measured at z Ht =. show the same characteristics (not plotted). f S uu /σ u - -/ Field Wind tunnel slope -/ - - f Ht/U Ht f S ww /σ u - -/ Field Wind tunnel slope -/ - - f Ht/U Ht Figure : Spectral density of the turbulence for the longitudinal and vertical velocity components at the meteo tower location at z Ht =. 89. no. of events [%] [lateral] Field Wind tunnel no. of events [%] (vertical) Field Wind tunnel - lateral deviation α [deg] - vertical deviation β [deg] Figure 6: histogram of lateral and vertical deviations of the instantaneous velocity vector from the mean one at the meteo tower location at z Ht =. 89. main tower WT meteo tower WT Meteotower field z/ht Lu /Ht Lw /Ht Figure 7: Dimensionless integral length scales. a) Longitudinal, b) vertical. Figure 7 presents the vertical profiles of the longitudinal and vertical length scales ( Lu and Lw respectively) obtained inside the forest area. The Lu measured at the main tower is very low inside the canopy and dramatically increases at the top of the canopy. This feature is characteristic for a dense canopy (Raupach et al. 996). On the other hand, the Lu

8 measured at the tree height is.ht, whereas the literature usually indicates a minimum value of H t. The Lu measured at the meteo tower also show the influence of the clearing. The Lu increases gradually, even inside the canopy, similar to a slightly dense canopy. Ht =. From z, the two vertical Lu profiles collapse together, indicating the end of the influence of local inhomogeneities (i.e. the clearing). The vertical Lw profiles lead to the same conclusions. The Lw measured at the tree height is.h t, which is in agreement with literature. The integral length scales processed from the -hour time series measured in field are additionally plotted. Although these hours presented some very similar atmospheric conditions, a significant scatter in Lu and Lw are noticed. This indicates the difficulty to precisely quantify these values. The comparison between wind tunnel and field data shows that the integral length scales in wind tunnel are underestimated but nevertheless belong to the right value range all the same. Conclusion An improved concept for the physical modeling of a forest area had been developed in order to characterize the flow structure in the model of the finite and inhomogeneous forest area of the Research center of Jülich. The turbulence properties of the approach flow and of the canopy flow have been quantified and compared with guidelines or previous studies. The comparison with other dense forest canopies is satisfying, since the mean and turbulent velocity profiles and the spectral distributions above and inside the canopy are in agreement with field data. Thus, the profiles of integral length scales through the forest canopy show similar evolutions to those found in literature. The clearing surrounding the meteo tower generates a local disturbance in the flow structure. Even if this tower is in the middle of the forest area, local measurements are not representative for a dense forest canopy. They present all the characteristics of a slightly dense canopy, whereas they are the signature of the local redevelopment of the boundary layer. Since the quality of the physical modeling of the finite forest area had been proved, this model can be used for dispersion applications in order to provide information about the origin, the trajectory and the age of the emissions sampled at the main tower during field campaigns (presented orally). References Aubrun, S., Leitl, B. (): Development of an improved physical modelling of a forest area in a wind tunnel. Submitted to Atmospheric Environment as short communication. Brunet, Y., Finnigan, J.J., Raupach, M.R. (99): A wind tunnel study of air flow in waving wheat: single-point velocity statistics. Boundary-Layer Meteorology 7: 9-. Counihan J. (97): Adiabatic atmospheric boundary layers: a review and analysis of data from the period Atmospheric Environment 9, pp Elbers J., Göttel H., Schüttenberg T. (). Projekt Entwicklung eines numerischen Oberflächenmodells des Forschungszentrums Jülich. Engineering Sciences Data Unit, 98. Characteristics of atmospheric turbulence near the ground. Part II: Single point data for strong winds (neutral atmosphere). Item n 8. ESDU International, London. Gardiner, B.A. (99): Wind and wind forces in a plantation spruce forest. Boundary-Layer Meteorology 67: Kaimal, J.C., Finnigan, J., (99): Atmospheric Boundary Layer Flows. Oford University Press. Novak, M.D., Warland, J.S., Orchansky, A.L., Ketler, R., Green, S. (): Wind tunnel and field measurements of turbulent flow in forests. Part I: uniformly thinned stands. Boundary-Layer Meteorology 9: 7-9. Raupach, M.R., Coppin, P.A., Legg, B.J. (986): Eperiments on scalar dispersion within a model plant canopy. Part I: the turbulence structure. Boundary-Layer Meteorology : -. Raupach, M.R., Finnigan, J.J., Brunet, Y., (996): Coherent eddies and turbulence in vegetation canopies: the miing-layer analogy. Boundary-Layer Meteorology 78: -8. VDI-guideline 78/, (): Physical modelling of flow and dispersion processes in the atmospheric boundary layer Application of wind tunnels. Beuth Verlag, Berlin.