Turbulent fluxes were measured by an eddy covariance system at three levels over an intricate land surface on the southern part of the Loess Plateau, consisting of heterogeneous flat terrain and a large valley 500 maway from the observation site to the southeast. The surface roughness length, the seasonal variation of bulk transfer coefficient for sensible heat (CH), and the seasonal variation of surface moisture availability (β) were also analyzed based on the observation. The flux footprint was carefully considered in this study. A relatively dry period of the experimental area existed from June to the first week of July 2004 when the land surface offered turbulent energy to the atmospheric surface layer mainly by sensible heat flux with a maximum value of around 230 Wm-2. A wet duration lasted from the second week of July to the end of September 2004 with very frequent rainfall events in conditions when the winds were mainly from the southeast;latent heat flux was dominant during the wet season and reached a peak value of around 280 Wm-2. The surface parameters of CH and β were calculated when the mean winds coming from the flat terrain, i.e., from the northwest direction. The values of CH ranged between 0.004 and 0.006 during the observational year of June 2004 to June 2005. The surface moisture availability β changed with seasons as anticipated with high values during June and July 2004 and lowest values around0.03 inFebruary 2005. Its peak value of 0.91 occurred in July; the mean value of β during the wet season was 0.29. Furthermore, the relationship between the surface soil water content and β indicated that changes in soil water content contributed much to variations of surface moisture availability β.
Turbulent fluxes in the atmospheric surface layer (ASL) are not only governed by properties of the airflow, they are also closely connected to the structure and function of the underlying surface. Generally, land surfaces are not homogeneous; the vertical structure of the atmospheric boundary layer (ABL) is modified over heterogeneous surfaces compared with that over homogeneous surfaces. As Wyngaard [
The Loess Plateau in China, as a semi-arid plateau, located within the middle reaches of the Yellow River basin and being affected by the Asian monsoon activities in summer, possesses a unique heterogeneity of topography and consequently induces typical processes of energy budget and fluxes transportations in the ABL. Measuring accurate surface energy fluxes and surface parameters over the Loess Plateau have significant meanings for the ABL evolution study over such a heterogeneous landscape, the Asian monsoon and climate change studies, and the research of the water cycle systems of the Yellow River Basin and even of the Eastern Asia. The field campaign in this study which was equipped with a flux tower and a wind profile Radar was the first ABL field observation conducted over the Loess Plateau [
In literatures, Kimura et al. [3,4] ever studied seasonal variations of heat balance including sensible and latent heat fluxes and soil moisture on the Loess Plateau using a three-layer soil model and verified the model by observations over bare soil fields. Liu et al. [
Based the ABL experiment in this study, many works have been done to investigate the effects of topography and surface heterogeneity on the ABL properties over the experimental region on the Loess Plateau. Li et al. [
In this paper, we will estimate the surface energy fluxes over the experimental site. The main objective is to study the variability of surface fluxes with the surface wetness. The seasonal variations of the bulk transfer coefficient for sensible heat and the surface moisture availability will be presented to investigate the properties of energy transfer and evapotranspiration over the studied region on the Loess Plateau.
The ABL observation in this study was carried out at a field site (35˚12'N, and 107˚40'E) located in the middle southern area of the Loess Plateau in China. The Loess Plateau has a semi-arid land climate, and lies within 34˚ to 40˚N and 100˚ to 115˚E. The Loess Plateau consists generally of strongly dissected flat terrain and gullies extending long to several tens kilometers with typical depths in the order of 100 m and surface widths in the order of 1 km [
The land cover of the studied site is heterogeneous, consisting of wheat, apple trees, and residences; the area is mainly flat, except for a large valley with the nearest edge about 500 m away from the observation tower in the southeast. The valley has the depth of around 100 m, a surface width of about 1 km, and the length around 20 km extending to the southeast (see
The monthly precipitation and mean air temperature averaged during 45 years from 1957 to 2001 at the experimental site are shown in
The daily precipitation observed during 13 months from June 2004 to June 2005 is shown in
In
not much. During the remaining days of July, through August, and until the end of September, rainfall events occurred very frequently, many of which had quite prominent precipitation. The maximum precipitation occurred on August 11 and August 19, with the daily values of 42 mm and 41 mm respectively. The accumulated precipitation during the period of July 9 to September 30 was 339 mm, whereas that during June 1 to July 8 was only 36 mm. Therefore, we will use the duration of June and the first week of July to represent the dry season over the studied region, and from the second week of July until the end of September to represent the wet season. We will focus on analyzing fluxes properties during these dry and wet seasons.
In this experiment, a flux and radiation observation system (FROS) was installed to measure turbulent fluxes and radiation components in the atmospheric surface layer. The FROS included three ultrasonic anemometer/thermometers (1210R3; GILL Instruments, Ltd., UK) mounted at 2 m, 12 m, and 32 m heights measuring turbulent quantities of wind and air temperature, as well as three open-path infrared CO2/H2O gas analyzers (LI-7500; Li-Cor, Inc., USA) recording the fluctuations of CO2 and water vapor densities at the three heights. Two shortwave radiometers (CM21; Kipp & Zonen B.V., Netherlands) and two longwave radiometers (CG4; Kipp & Zonen B.V., Netherlands) were installed at 2 m level to measure the upward/downward shortwave radiations and upward/downward longwave radiations. Soil thermometers (CS107; Campbell Scientific Inc., USA) and soil moisture sensors (TDR-615; Campbell Scientific Inc., USA) were also used to measure the soil temperature and soil moisture at the depth of 2 cm, 10 cm, 20 cm, 40 cm, and 80 cm respectively. A rain gauge (34-T; Ohtakeiki Ltd., Japan) measured the precipitation. The data logger was the type of CR5000 by Campbell Scientific, Inc., USA.
In this study, 10 Hz raw turbulent data measured by the ultrasonic anemometer/thermometers at three heights (2 m, 12 m, and 32 m) during the observational year (June 2004 to June 2005) were used to calculate sensible and latent heat fluxes by eddy correlation method. In section 5.1, we will focus on analyzing the properties of surface heat fluxes in the wet and dry seasons of a summer during four months (June to September in 2004) within the observational year.
Because the surface parameters CH, CE, and β are based on the assumption of surface homogeneity, to avoid the topography complexity caused by the southeast valley, these parameters were calculated only in conditions when winds came from the northwest flat terrain. In Sections 5.2 and 5.3, we will investigate the seasonal variations of bulk transfer coefficient CH for sensible heat and surface moisture availability β based on the data collected at 32 m height during the whole observational year (June 2004 to June 2005). In all these cases, we only selected clear days according to the precipitation data.
In previous studies, Matsushima and Kondo [
The turbulent sensible and latent heat fluxes at 2 m, 12 mand 32 m heights were calculated by the eddy covariance method, which can be formulated as following:
Here H denotes the sensible heat flux, LE the latent heat flux; cP and ρ are the specific heat and the air density, l the latent heat for vaporization of water, the vertical turbulent wind speed, the turbulent potential temperature, the turbulent specific humidity.
Firstly dynamic calibrations of the water vapor and CO2 densities measured by the infrared CO2/H2O gas analyzer were conducted by a standard hygrometer and standard CO2 gas [
For flows over complex terrain, errors may arise when vector quantities are measured in a reference framework that is not consistent with that of the equations used to analyze them. To solve this problem, three mathematical rotations were applied here to transfer the sampled velocity data from the instrument’s reference frame to the streamline reference frame according to the scheme by Kaimal and Finnigan [
Sensible heat flux H and latent heat flux LE from a surface can be generally determined from the bulk transfer method [
Here Ts is the surface temperature, the saturation specific humidity at a temperature of Ts. U, T, and q represent the mean wind speed, mean air temperature, and mean specific humidity at a reference level, respectively. Here CH and CE are the bulk transfer coefficients for sensible heat and latent heat.
For a well-wetted surface, is expected. Since the surface moisture availability β is defined as, for a well-wetted surface, and for a completely dry surface β = 0 can be expected; the values of β for land surfaces with different wetness vary between these limits.
One of the methods to determine the bulk transfer coefficients is based on the Monin-Obukhov similarity theory (MOST):
Here and are stability correction functions for momentum and heat; is von Kármán’s constant; z0 and zt represent the roughness length for momentum and sensible heat, respectively; and d is the displacement height [
There are many formulations for the MOST stability functions in literatures [21-26]. In this study, we applied the interpolation stability functions for the ASL proposed by Brutsaert [
The values assigned to the constants are a = 0.33, b = 0.41, m = 1.0, c = 0.33, d = 0.057, and n = 0.78. Equations (5) and (6) can be readily integrated into
and then yield the stability correction functions:
in which, , and denotes a constant of integration, given by [
After CH is determined from (4), as an alternative, the surface temperature Ts in Equation (2) can be replaced by the effective surface temperature for sensible heat flux Th as:
Finally from (3) the surface moisture availability β can be derived:
After sensitivity tests, we found that the displacement height d had little effect on CH and β. Thus we applied an estimated value of d = 0.4 m and a value of z0 = 0.5 m calculated from the neutral logarithmic wind relationship based on data at 32 m height during June 2004 in determination of CH and β. To determine zt values, Sugita et al. [
Before calculating surface fluxes, the flux footprint was estimated to investigate the influence of the upwind spatial distribution of the surface emission to the vertical fluxes measured at some height. The footprint was calculated by the methods given by Horst and Weil [
In this section we will focus on analyzing the properties of the sensible and latent heat fluxes during the dry season (June and first week of July) and the wet season (remaining days of July, August and September in 2004) of the experimental area.
Figures 4 shows the time series of daily mean sensible and latent heat fluxes as well as net radiation from June to September in 2004. The mean sensible heat flux (denoted as H) and the mean latent heat flux (denoted as LE) showed similar fluctuation trend with the net radiation and with each other during the four months. The sensible heat flux H was close to the latent heat flux LE in June.
During July, H decreased on the whole; LE started to increase in clear days when the net radiation was also large and LE was getting larger than H gradually. The latent heat flux LE kept increasing in August and was much larger than the sensible heat flux H. The maximum of LE during August was around 200 Wm−2; in August at the study site rainfalls were concentrated and the precipitation was prominent, and most of the land surface was bare soil after the wheat was harvested in June. Thus these latent heat flux values are comparable to the results of Kimura et al. [
These changes of surface heat fluxes were mainly caused by the variations of the solar radiation and the precipitation during the wet season and the dry season in the region. On one side, the decreasing of the solar radiation from June to August caused the fall of the surface temperature. In consequence, less sensible heat was available to be transferred into the atmosphere by means of turbulent processes between the land surface and the atmosphere. On the other side, the predominant increasing of the precipitation during July and August (see
According to the different properties of sensible and latent heat fluxes during the dry and wet seasons, it should be meaningful to investigate the detailed characteristics of H and LE respectively in each season. For this purpose, we selected some typical days from both seasons with proper micrometeorological conditions such as no precipitation, clear sky in daytime, and mild wind speeds.
Considering the radiation components and the precipitation (see
We show the radiation components of each representative period in
The sensible and latent heat fluxes during both representative periods of dry and wet seasons are shown in
After calculating ensemble averages of H and LE during each representative period, the daily variations of H and LE are shown in
In wet period, the maximum of H in daytime occurred at 11:30 with a value of 123 Wm−2. Around 15:00, H showed a second peak of 118 Wm−2. LE in daytime also showed double peak at 12:30 and 15:00, with values of 280 Wm−2 and 265 Wm−2, respectively. In nighttime, both H and LE were near to zero or slightly minus. These four representative days were shortly after the rainy days with large precipitation. Thus the surface humidity was prominently increased. Therefore, in daytime after around 10:00, strong evaporation was produced from the land surface
due to heating effect by solar radiation. Such strong evaporation transferred large latent heat flux to the atmosphere, so that LE started to increase rapidly and be much higher than H. After sunset around 18:00, when the solar radiation tended to zero, LE decreased rapidly to zero.