OFFSHORE BOUNDARY-LAYER MODELLING H. Bergström1 and R. Barthelmie2 1) Uppsala Univ., Dept. of Earth Sc.-Meteorology, Villavägen 16, SE-752 36 Uppsala, Sweden Tel: +46 18 4717181. Fax: +46 18 551124. Email:
[email protected] 2) Dept. of Wind Energy and Atm. Physics, Risø Nat. Lab., P.O. 49, DK-4000 Roskilde, Denmark Tel: +45 46775020. Fax: +45 45775970. Email:
[email protected] ABSTRACT: Higher-order closure model results are used to climatologically quantify the impacts on the wind resource of features not normally accounted for in commonly used simplified models, as e.g. thermally driven flows and low-level jets. The paper describes results from the boundary-layer modelling work package (WP4) of the ENDOW (EfficieNt Development of Offshore Windfarms) project, where the objectives are to provide an improved boundary-layer input to wake models and to account for wake/boundary-layer interactions. Statistics from a number of studies are presented, where wind predictions made with the higher-order closure MIUU-model are compared with winds predicted by more simplified models. Tests with the MIUU-model are made in order to investigate the importance of large-scale thermally driven flows, which cannot currently be incorporated into a wind farm design tool. The offshore thermal stratification climate is also investigated. 1 INTRODUCTION Mapping the offshore wind climate, it is important to take into account both land and sea areas, together with the interaction between the two. Although the horizontal air pressure gradient, i.e. the geostrophic wind, is the primary driving force for the wind, it is well known that over land also topography, roughness and thermal stratification, are factors affecting the wind climate. In coastal areas also factors related to the land-sea transition are of importance, and may contribute to a complex wind field, and affect the wind also at large offshore distances. Spatial variations of wind and the wind climate over large offshore areas, like the Baltic Sea, are difficult to measure. Instead models may be used. Often these models just include a simplified description of the physics determining the boundary layer winds. But this means that the simplified models will e.g. not be able to account for the effects of large-scale thermally drive flows since the physics needed for this are excluded. Field experiments in the Baltic Sea area, [1], have shown that the atmospheric boundary layer far from seldom departs from often used simplified relations. There are several reasons for complex offshore wind conditions, e.g. the low-level jets (LLJ), which are often observed in the Baltic Sea area. There are several possible causes of low-level jets. One is an inertial oscillation initiated with air coming from land out over the sea, [2], which may give rise to a low-level jet with even super-geostrophic winds at heights of the order 100-300 m. Predicting the wind at e.g. 50 m with a model not capable of generating low-level jets, using measurements at 10 m, will thus result in wind speeds which are systematically too low, and one can expect the error to increase with height. Non-homogeneous offshore conditions may also be due to thermally driven flow modifications, such as the sea breeze. Offshore internal boundary layers may develop much more slowly than predicted by simplified models, and observations show that with stable conditions over the sea the wind speed may even, in contrary to what is normally assumed, decrease with distance offshore up to at least 100 km from the coast [1]. Due to complex interaction between land and sea, the offshore wind is often not only a function of distance from the coast but may also be affected by the curvature of the shoreline.
Thermally driven flows, low-level jets and other features, which do not follow simplified ‘normal’ behaviour may, however, be modelled quite well using higher-order closure boundary layer models like the MIUU-model from Uppsala University [1]. Results from a number of studies will be presented, where winds predicted by the MIUUmodel are compared with predictions by more simplified models. An attempt is made to quantify the impacts on the wind resource of features not normally accounted for in simplified models.
2 THE MIUU-MODEL AS A WIND CLIMATE TOOL The MIUU-model is a three-dimensional hydrostatic mesoscale model developed at Uppsala University, Sweden, [3]. The model has prognostic equations for wind, temperature, humidity and turbulent kinetic energy. The coordinate system is terrain-influenced, roughly following the terrain close to the surface and gradually transforming to horizontal at the model top. To reduce influences from the boundaries, the model area is chosen to be much larger than the area of interest. To limit the number of grid points in the horizontal, a telescopic grid is used, with the highest resolution only in the area of interest. In the vertical, the lower levels are log spaced while the higher levels are linearly spaced. The lowest grid point is at height z0, where z0 is the roughness length, and the model top is typically at 10000 m. At the lower boundary, roughness length, height above sea level, and temperature have to be specified at each grid point. The land surface temperatures are estimated with a surface energy balance routine using as input solar radiation and land use. Over sea the observed monthly average sea-surface temperatures have been used. In the ideal climate study, all synoptic and boundary conditions should be covered. But since the MIUU-model is computer time consuming to run, this would require an unrealistically large number of simulations. Thus the most important flow forcing parameters have to be identified, and varied in order to cover a reasonably wide range of atmospheric conditions. The most important parameters are: geostrophic wind (strength and direction), thermal stratification (through the daily temperature variation), surface roughness, topography, and land-sea temperature differences. The temperature shows a clear an-
nual variation, why the model has been run for 4 months (January, April, July and October), selected to represent the four seasons. For each season runs were made with three values of the geostrophic wind speed, and with 8 wind direction sectors, giving a total of 96 model runs used to determine the wind climate. Each simulation was run for a 36-hour time period, of which the last 24 hours were used. The initial potential temperature and humidity profiles were taken from climatologically averaged radiosonde data for the different seasons. All runs were finally weighted together using climate data of the geostrophic wind, giving the mean wind speed (annual or seasonal), or wind energy potential, at different heights. The model domain covers the Baltic Sea area with a horizontal resolution of 9 km. The resulting annual mean wind speed at 48 m, shown in Fig. 1, is predicted to be around 5-6 m/s over southern Sweden and 8.5-9 m/s over the Baltic Sea. A comparison between the modelled winds and observations show good agreement both offshore and over land [4]. This agreement between modelled wind climate and observations, together with several earlier model verifications, [1], support the accuracy of the MIUU-model. We may thus with confidence use the MIUU-model simulations to study influences on the wind climate of e.g. flow modifications due to differences in temperature and stratification between land and sea.
action, together with the temperature differences between land and sea and its diurnal cycle. These heterogeneous offshore winds are not expected following the analytical descriptions of the wind commonly used in simplified wind models, which would give a more or less homogeneous wind field at 12 m height after an offshore distance of 10 km or so, disregarding changes in stability. We will now further study the offshore stratification, investigate the influence from the thermally driven flows, and also compare with estimates of the influence on the offshore wind from topography and the higher roughness over land. The influence of low-level jets will also be investigated and the offshore thermal stratification will also be investigated. July daily average wind speed at height 12 m, SW10
8
(m) 400000 300000 200000 100000
200000
400000
600000
800000
y−distance (km)
5.5 5
400
1000000
(m)
Figure 1: Annual mean wind speed in the Baltic Sea area at 48 m height. Estimations made with the MIUU-model. 3 THERMALLY DRIVEN WIND MODIFICATIONS The Baltic Sea is surrounded by land, which means that land areas might influence the wind and turbulence structure over the sea regardless of wind direction. We may expect temperature differences between land and sea to be of importance to the offshore wind, giving sea breeze circulations and other thermally driven flow modification affecting large offshore areas. The growth of a stable internal boundary layer may affect the wind speed for large distances from the coast, even giving decreasing wind speeds at lower levels [1]. At the top of this internal boundary layer a low level jet may develop as a result of frictional decoupling at the coast, [2]. It is thus important to take thermal factors into consideration when the offshore wind field is studied. They may contribute to a spatial variability of the wind field also in the averages. An example of this is shown in Fig. 2., where the daily average wind speed at 12 m from a simulation with a geostrophic wind of 10 m/s from southwest is plotted. In spite of the constant pressure gradient, the offshore wind field is not constant. The main reasons for this are the complex geography leading to a complex land/sea inter-
4.5 4
300
3.5 3
100
9.4 9.2 9.0 8.8 8.6 8.4 8.2 8.0 7.5 7.0 6.5 6.0 5.5 5.0 4.5 4.0
500000
0 0
6
500
200
m/s
600000
6.5
600
0
700000
7
700
MIUU-model - annual mean wind speed at 48 m
800000
7.5
800
2.5 2 0
200
400 600 x−distance (km)
800
1000
Figure 2. Daily average wind speed in the Baltic Sea area at 12 m height as modelled by the MIUU-model using July temperature conditions and the geostrophic wind 10 m/s from southwest. 4 OFFSHORE THERMAL STRATIFICATION Earlier investigations using wind and temperature profile data from two meteorological towers, one at Alsvik on the west coast of the island Gotland, and the other on the island Östergarnsholm east of Gotland [5], show quite different stability statistics. The results from Alsvik, using data from the sea sector, show stable stratification (Richardson number Ri>0.05) during about 55% of the time, neutral conditions were found 20% of the time, and unstable conditions occurred in 25% of the data. The results from Östergarnsholm do not show this dominance of stable stratification over the sea. Instead unstable stratification (Ri
