SCIENCE CHINA Earth Sciences • RESEARCH PAPER •

doi: 10.1007/s11430-014-4881-9

Spatial distribution of surface energy fluxes over the Loess Plateau in China and its relationship with climate and the environment ZHANG Qiang1,2*, ZHANG Liang1,3, HUANG Jing4,1, ZHANG LiYang3, WANG WenYu1,3 & SHA Sha1 1

Institute of Arid Meteorology, China Meteorological Administration; Key Laboratory of Arid Climatic Change and Reducing Disaster of Gansu Province; Key Open Laboratory of Arid Climatic Change and Disaster Reduction of China Meteorological Administration, Lanzhou 730020, China; 2 Meteorological Bureau of Gansu, Lanzhou 730020, China; 3 College of Atmospheric Sciences, Lanzhou University, Lanzhou 730000, China; 4 Meteorological Bureau of Yongjia, Yongjia 325100, China Received June 4, 2013; accepted December 16, 2013

China’s Loess Plateau is located at the edge of the Asian summer monsoon in a transition zone of climate and ecology. In the Loess Plateau, climate and environments change along with space, which has an obvious impact on the spatial distribution of surface energy fluxes. Because of scarce land-surface observation sites and short observation time in this area, previous studies have failed to fully understand the land-surface energy balance characteristics over the entire the Loess Plateau and their effect mechanisms. In this paper, we first test the simulation ability of the Community Land Model (CLM) model by comparing its simulated data with observed data. Based on the simulation data for the Loess Plateau over the past thirty years, we then analyze the spatial distribution of surface energy fluxes and compare the pattern differences between the area averages for the driest year and wettest year. Furthermore, we analyze the relationship between the spatial distribution of the components of the surface energy balance with longitude, latitude, altitude, precipitation and temperature. The main results are as follows: the spatial distribution of surface energy fluxes are significantly different, with the surface net radiation and sensible heat flux increasing from south to north and latent heat flux and soil heat flux decreasing from southeast to northwest. The sensible heat flux at the driest point is nearly twice as high as that at the wettest point, whereas the latent heat flux and soil heat flux at the driest point are half as much as that at the wettest point. The impact of variations of annual precipitation on the components of the surface energy balance is also obvious, and the maximum magnitude of the changes to the sensible heat flux and latent heat flux is nearly 30%. To a certain extent, geographical factors (including longitude, latitude, and altitude) and climate factors (including temperature and precipitation) affect the surface energy fluxes. However, the surface net radiation is more closely related to latitude and altitude, sensible heat flux is more closely related to the monsoon rainfall and latitude, and latent heat flux and soil heat flux are more closely related to the monsoon rainfall. Loess Plateau, components of surface energy balance, spatial distribution, climatic and geographical factors, effect mechanism Citation:

Zhang Q, Zhang L, Huang J, et al. 2014. Spatial distribution of surface energy fluxes over the Loess Plateau in China and its relationship with climate and the environment. Science China: Earth Sciences, doi: 10.1007/s11430-014-4881-9

Atmosphere is the most typical coupling system of heat and power in nature, and external energy forces and internal *Corresponding author (email: [email protected])

© Science China Press and Springer-Verlag Berlin Heidelberg 2014

energy conversion are the root causes of its ever-changing conditions. The formation of weather processes and evolution of climatic states are in large part under the control of atmospheric energy transmission and conversion mechanisms. earth.scichina.com

link.springer.com

2

Zhang Q, et al.

Sci China Earth Sci

Severe convective weather and climate variations are generally the direct results of atmospheric energy accumulation and adjustment (Wu et al., 1995), and even small changes in the energy of the atmospheric system likely have a profound impact on weather and climate. Land-surface energy exchange processes are a primary source of energy in the earth atmosphere system (Huang et al., 2011; Zhang et al., 2011a). The energy force effect of solar radiation on the atmospheric system is also indirectly implemented through land-surface energy exchange processes. Therefore, important intrinsic relations occur between the spatial distribution pattern of land-surface energy and regional weather and climate (Dickinson 1995). The spatial distribution of land-surface energy is affected by the spatial change of climate and environmental factors. The Loess Plateau in China is located at a latitude of 34°– 40°N and longitude of 103°–114°E (Zhang et al., 2008), and it is the largest area of loess deposition in the world and the second step of topography of China. The area features an overlap of the edge of the Asian summer monsoon and northeast slope region of the Qinghai-Tibet Plateau, which has a unique geographical position and climatic environment (Wang Y Y et al., 2011). The Loess Plateau is not only a transition zone from the inland arid climatic zones to monsoon climate zones but also a transition zone from the Qinghai-Tibet Plateau to the coastal plain. Precipitation spatial changes (Huang et al., 2012; Li et al., 2008) and vegetation distribution are significantly different in this area (Li et al., 2003). The altitude drop and spatial heterogeneity of the underlying surface properties are also obvious. All of these features lead to the unique spatial distribution characteristics of the land-surface energy balance in this region. Simultaneously, the areas with a dry climate and fragile ecology feature a spatial distribution pattern of land-surface energy balance characteristics in the Loess Plateau region that is more sensitive to climate change (Jin et al., 2002) and significantly affected by climate change (Zhang et al., 2013a). The spatial differences of the land-surface energy balance over the Loess Plateau not only affect the motion of the boundary layer atmosphere, but they also cause local circulation or convection (Huang et al., 2011; Zhang et al., 2011a), which directly or indirectly impact the formation and evolution of weather in this area. The variation of the land-surface energy balance spatial patterns can affect adjustments of the climate system energy cycle and therefore impact regional climate change. Previous studies (Lu et al., 2010) have shown that the Loess Plateau and its surrounding areas have high occurrences of drough over the past 50 years and the current trend of enhancement and extension is most likely related to the unique spatial distribution pattern of land-surface energy balance and its changing trends in this region. The land-surface energy balance of the Loess Plateau region, therefore, is of great significance to regional weather and climate.

January (2014) Vol.57 No.?

Land-surface process observations and studies have recently been launched in the Loess Plateau region (Zhang et al., 2009; Huang et al., 2008), and observational data have been analyzed to determine the diurnal variation characteristics of the land-surface energy exchange under typical weather conditions in each station (Li et al., 2010; Sun et al., 2011). The spatial differences of the land-surface energy balance in the summers of typical years (Zeng et al., 2010; Zeng et al., 2007) and annual change of the surface energy balance at a typical observation station (Zhang et al., 2011b; Li et al., 2012) and its response to precipitation fluctuation (Zhang et al., 2013a) have been analyzed in preliminary studies. In addition, simulation studies of the land-surface energy exchange processes of ecosystems in the agriculture and animal husbandry ecotone over the Loess Plateau (Cheng et al., 2011) and satellite remote sensing inversions of the spatial distribution characteristics of the land-surface energy flux at typical time have been performed (Zhang J et al., 2010). However, previous studies are greatly limited as a result of the short time series of experimental data and unreliability of most satellite remote sensing inversions and numerical model simulation data. Overall, a limited number of current studies have focused on the spatial distribution characteristics of the land-surface energy balance and its response to climate change over the Loess Plateau region; this has led to an insufficient understanding of the physical mechanisms. All of these conditions hamper an objective and comprehensive understanding of the Loess Plateau land-surface energy balance characteristics and limit evaluations of the inner link between the spatial distribution pattern of the energy balance and weather and climate, which limits the establishment of weather and climate prediction technology. In recent years, however, significant improvements have been made to global land-surface process models (Dai et al., 2003), and simulation data have been successfully applied in practice (Song et al., 2009). Therefore, in this paper, the long-term time series of simulated data from the new generation of land-surface process models were used in the analysis of the spatial distribution of the landsurface energy balance and its physical relationship with climate factors in the arid Loess Plateau to explore the intrinsic links between land-surface energy balance spatial patterns and weather and climate.

1 Climate background and information from observational data The Loess Plateau covers the regions west of the Riyue Mountains in Qinghai, east of the Taihang Mountains in Shanxi, adjacent to the Qinling Mountains in Shaanxi, north of the Yinshan Mountains of Inner Mongolia. The plateau is approximately 1000 km from east to west and 700 km from north to south, and it has a total area of approximately 6.4×105 km2. The Loess Plateau region is affected by dry

Zhang Q, et al.

Sci China Earth Sci

and cold polar air in winter for half a year and warm wet monsoon airflow from the western Pacific and Indian Ocean in summer, and it generally has a continental climate background (Yu et al., 2002). To reveal the climate characteristics in different areas over the Loess Plateau, this paper applied the dry and wet index Im that was originally proposed by Thornthwaite (1948) and subsequently improved by Ma et al. (2005a, 2005b): I m = 1  Pr

n

P

ek

,

(1)

k=1

where Pr and Pek are the total precipitation and total potential evaporation, respectively. Pek is estimated by the following formulas: k  1°C;  a  k   Pek  d    101, 1C  k  C;  h   a  a T  a T 2   °C. k 3 k  1 2 k

(2)

where d is the result of the number of days of each month divided by thirty, and Tk is the average surface air temperature of the kth month. Coefficient a is defined as a = 0.49239 + 0.01792h  7.71 10 h 2 + 6.75 10 h3 ,

(3)

where h is the total heat index that is defined as the sum of the heat index i for all months, and i is the result of the total average surface temperature Tk of all months divided by the number 5: n

n T

h =  i , and i =  5k . k=1

(4)

k=1

Finally, the constants a1, a2, a3, respectively, are equal to

January (2014) Vol.57 No.?

a1=415.8547, a2=32.2441, a3=0.4325.

3

(5)

A greater value of Im shows a dry climate and less available water in the climate system. According to the division standard raised by Ma et al. (2005a, 2005b), Im<0.5 indicates a humid climate; 0.50.5 indicates an arid climate. As shown in Figure 1, the climate in this region has a significant spatial difference in climate change, with the dry-wet index Im changing between 0.5–1.0 and including arid, semiarid and semi-humid climate types. The southern area is semi-humid with 0.50.5 in, and most of the remaining areas are semiarid with 0
Figure 1 The distribution of the dry-wet index, altitude and observational sites over the Loess Plateau. The black line is the border of the Loess Plateau. The black dots represent the observational stations. The isoline suggests the elevation (m). Coloring shows the dry-wet index Im.

4

Zhang Q, et al.

Sci China Earth Sci

field observational data and routine data from weather stations. The data for the land-surface energy balance components were provided by the Semiarid Climate Observatory and Laboratory (SACOL) of Lanzhou University, which is located on the Loess Plateau. SACOL is a comprehensive scientific experimental station in a typical semiarid climate zone (Huang et al., 2008) and part of the series “Experimental Study of Land Surface Process in Chinese Loess Plateau (LOPEX)” (Zhang et al., 2009), which is represented by a small black triangle in Figure 1. This station began a continuous series of observations from April 2006 for six years. Not only does it have a long period of observations, but the data quality is also strictly controlled. EdiRe software is used to remove the outliers from the components of the land-surface energy flux data acquired by the eddy correlation method and a series of processes have been conducted that include coordinate rotation, turbulence stability calculation, and H2O and CO2 lag correction. Relevant references (Huang et al., 2008; Zuo et al., 2009, 2011; Yan et al., 2011; Wang et al., 2010; Guan et al., 2009) have introduced the main technical indicators of the observation instruments. Recent studies (Zhang and Li, 2010; Zhang, 2012a; Liu et al., 2011) suggest that the observation data are of high reliability. This paper used the SACOL data from January 2007 to December 2010 and air temperature and precipitation observation data from 49 routine meteorological stations on the Loess Plateau over nearly fifty years. The latter have passed the quality control requirements of the National Meteorological Information Center, China Meteorological Administration.

2 Land-surface process simulation and observation validation In this paper, we use nearly thirty years of land-surface process simulation data from the CLM (community-developed land surface model) from 1979 to 2010. The CLM model (Dai et al., 2003) is a new generation land-surface process model that combines the advantages of the NCAR land-surface model (LSM) (Bonan, 1998), bio-atmospheric transmission scheme (BATS) (Dickinson et al., 1995), IAP94 (Dai et al., 1997), and the land-surface model developed by the Institute of Atmospheric Physics, Chinese Academy of Sciences. The CLM is an important component of the CCSM (Community Climate System Model), and it has a hierarchical structure with one layer of vegetation, five layers of snow, and ten layers of soil. The soil temperature is calculated using the soil heat conduction equation, and soil moisture is calculated by Darcy’s theorem. The numerical integral adopts a finite difference space division method and fully implicit time integration scheme. The physical parameters of the model and numerical solution have all been verified. The land-surface process simulation data are calculated by the CLM model based on the atmospheric forcing

January (2014) Vol.57 No.?

data, which are derived from the Global Land Data Assimilation System (GLDAS), which has a spatial resolution of 1°×1° and time resolution of 3 h. The simulated results can be directly obtained from the NASA website (http://disc.sci. gsfc.nasa.gov/hydrology/data-holdings) (Rodell et al., 2004) and have been used successfully to analyze climate changes of land-surface processes in previous studies (Ngo-Duc et al., 2005). To validate the simulation capability and reliability of the model over the Loess Plateau region, Zhang et al. (2013b) compared the simulated results of the CLM, Mosaic (Koster et al., 1996), and Noah (Ek et al., 2003) models which are included in the GLDAS and Australian CABLE (Wang Y P et al., 2011) model with the results from field observations and found that the correlation coefficient of the CLM is higher compared with the observations and the root-square deviation is lower compared with the other three models for this region. The SACOL observational data are used to verify the reliability of the CLM model simulations. The simulated grid data have the same latitude and longitude as the SACOL station. Because the time resolution of the SACOL station data and simulated results of the CLM are different and this article focuses on the relationship between the surface energy and climate environment, we only compare the monthly average. Considering that the unbalanced phenomenon of surface energy flux observed by the eddy correlation method is common, we first conduct a vertical thermal advection and soil heat flux correction before contrastively analyzing the simulation data and observation data (Zhang et al., 2012a). After correction, the surface energy closure can be as high as 0.94. Figure 2 shows the correlations between the simulated surface net radiation and corrected observed value, sensible heat flux and observed value, soil heat flux and observed value, and latent heat flux and observed value. This graph shows that the CLM model produces accurate simulations of the surface energy balance components. The simulated and observed values are relatively close, and the correlation coefficient of the surface net radiation, sensible heat flux, latent heat flux, soil heat flux and observed value are 0.97, 0.97, 0.82, and 0.90, respectively, and the root mean square error values are 19.5, 12.0, 15.2, and 4.4 W/m2, respectively. The surface temperature is required to calculate the surface energy balance components, and it is usually obtained through field observation methods, including the use of surface temperature sensors, MODIS data inversion, and numerical model simulations. Of these three methods, observation values are generally regarded as the “true value”; however, there are a limited number of observation sites because of environmental factors. MODIS data have a high spatial resolution, but it must be corrected to obtain synchronous ground station observations. Although simulations by the land-surface model are not as accurate as MODIS data in terms of spatial resolution, this method can achieve continuous simulation results by using forcing data. For the

Zhang Q, et al.

Sci China Earth Sci

January (2014) Vol.57 No.?

5

Figure 2 Comparison of the correlations of simulated surface net radiation (a), sensible heat flux (b), soil heat flux (c), latent heat flux (d) and observed values after correction.

lack of observation sites on the Loess Plateau, to verify the model’s reliability in simulating surface temperature and energy components, we have compared the reliable monthly averaged CMG grid-surface daytime temperature data of MODIS from the Terra satellite with a spatial resolution of 0.05° with the CLM simulated monthly averaged surface temperature data over the Loess Plateau (Figure 1). As seen from Figure 3, the monthly average surface temperature of MODIS, CLM and SACOL station is consistent over a seasonal cycle. Temperature is the highest in summer and the lowest in winter, and the correlation is relatively high, with the correlation coefficient between MODIS and SACOL and CLM and SACOL both reaching 0.99 and passing the significance test of =0.01. Compared with the CLM, the difference in temperature between MODIS and SACOL station is greater, with the average value reaching 7.7°C. The difference in temperature between the CLM and SACOL station is 3.1°C. Studies have shown that the influence of the atmosphere may cause a star brightness temperature error in MODIS that is one of the main causes of regional heat spatial differences and leads to significant errors in the surface temperature (Mao, 2004). Therefore, before the application of MODIS data, an atmospheric correction is usually required. For the CLM, sensitivity experiments carried out by Han (2006) show that in regions where vegetation cover is less, the change of surface emissivity can have a

significant influence on the surface temperature. On the Loess Plateau, the detailed relationship between surface emissivity and surface temperature requires further study. By comparing the spatial distribution characteristics of the 30-year averaged surface energy balance components simulated by the CLM model on the Loess Plateau region (Figure 4) with that of the coordinated observation test data for North China in July to September 2008 from previous studies (Zeng et al., 2010), we found that although these two types of data are roughly consistent in their spatial distribution patterns, the energy distribution of the observation data over the Loess Plateau region is only a rough outline because for the lack of coordinated observation test sites, and we could not clearly determine the energy spatial distribution features. The inversion results of the land-surface energy balance components of the MODIS satellite remote sensing data in the summer of 2007 from previous studies (Zhang J et al., 2010) are generally higher than the simulated results over the Loess Plateau (Figure 4). This result was partly caused by the MODIS satellite passing only twice a day and only under clear-sky conditions, which results in the fact that data are limited in time. However, the uncertainty of the land-surface parameters in the inversion process greatly affects the accuracy of the inversion results. Therefore, using the simulated results of the land-surface model to analyze the components of the land-surface energy

6

Zhang Q, et al.

Sci China Earth Sci

January (2014) Vol.57 No.?

Figure 3

The monthly average temperature according to MODIS, CLM and SACOL station data for the Loess Plateau (January 2007–December 2010).

Figure 4

Spatial distribution that is characteristic of the 30-year averaged surface energy balance components on the Loess Plateau.

Zhang Q, et al.

Sci China Earth Sci

balance is better than those by other methods over the Loess Plateau.

3 Spatial change of surface energy balance components 3.1 Spatial distribution characteristics of the surface energy balance components The large span in latitude and longitude and obvious spatial differences in climate and altitude over the Loess Plateau region significantly influence the spatial distribution characteristics of the land-surface energy balance components (Zhang et al., 2012b). The spatial distribution of the 30-year averaged components of surface energy balance over the Loess Plateau region (Figure 4) indicates that the surface net radiation flux increases gradually from south to north and has a close relationship with latitude, with the net radiation flux increasing with increased latitude. However, the net radiation is more obviously affected by altitude in the southwest and generally decreases with lower altitude. Therefore, the high-value center appears in the north, and the low-value center appears in the southwestern corner, with the highest and lowest values being approximately 89 and 66 W/m2, respectively. The surface sensible heat flux also increases gradually from south to north but changes more regularly than net radiation; it is almost parallel to latitude in most areas, indicating that it has a close relationship with latitude. The high-value center also appears in the north, but the low-value center is in the southern region. The highest and lowest values are approximately 71 and 39 W/m2, respectively. Although the surface latent heat flux distribution is regular, it is not completely parallel to latitude, presents a gradual decreasing trend from southeast to northwest and has a closer relationship with the monsoon

January (2014) Vol.57 No.?

7

rainfall spatial distribution. The high-value center appears in the southeast, and the low-value center appears in the northwest, with the highest and lowest values beinges approximately 40 and 11 W/m2, respectively. The soil heat flux and latent heat flux have a similar spatial distribution and reduce gradually from southeast to northwest. However, the soil heat flux is more noticeably affected by altitude in the southwest and decreases with lower altitude, which is similar to the situation of the surface net radiation. Its high value appears in the southwest and low value center is in the north, and the highest and lowest values are approximately 0.62 and 0.08 W/m2, respectively. To further highlight the diversity of space, the lattice of the lowest average rainfall in the 30-year average for this region is selected as the driest point, and the lattice of maximum average rainfall is selected as the wettest point. We compare the surface energy balance characteristics of these two points and the spatial average of this region. Figure 5 shows the surface energy balance components and their distribution rate in the 30-year spatial average, including the driest and wettest lattice. The spatial average surface net radiation, sensible heat flux, latent heat flux, and soil heat flux over the Loess Plateau region are 78.4, 53.1, 25.2, and 0.21 W/m2, respectively. Compared with the energy fluxes at the driest and wettest points, the sensible heat flux of the driest point is nearly twice as high as that of the wettest point, whereas the latent heat flux and soil heat flux of the driest point are half as muth as that of the wettest point. At the driest and wettest points, the sensible heat fluxes are 64.8 and 36.2 W/m2 and approximately 11 W/m2 larger and 17 W/m2 smaller, respectively, than the spatial average, which is closer to the driest lattice. At the driest and wettest points, the latent heat flux is 12.5 and 36.4 W/m2 and approximately 13 W/m2 smaller and 11 W/m2 larger, respectively, than the spatial average, which is slightly closer

Figure 5 The surface energy balance components (a), their allocation rate (b) in the regional 30-year average, the driest point (40.5°N, 107.5°E) and the wettest point (34.5°N, 110.5°E). Rn, net radiation; SH, sensible heat; LH, latent heat; G, soil heat flux.

8

Zhang Q, et al.

Sci China Earth Sci

to the wettest lattice. At the driest and wettest points, the soil heat flux is 0.10 and 0.27 W/m2 and approximately 0.11 W/m2 smaller and 0.06 W/m2 larger, respectively, than spatial average, which is also closer to the wettest lattice. The surface net radiation is an exception, and its spatial average is closer to that value of the driest lattice and much larger than that of the wettest location. This result suggests that the average radiation force on the Loess Plateau is overall more similar to the climate of the drier areas in this region. From the allocation rate over the Loess Plateau region shown in Figure 5(b), the spatial average of the surface sensible heat flux, latent heat flux, and soil heat flux accounts for 67.7%, 32.1%, and 0.27%, respectively, of the net radiation. The sensible heat flux is dominant, which is a characteristic of semiarid zones. The surface sensible heat flux, latent heat flux, and soil heat flux of the driest point accounts for 83.8%, 16.1%, and 0.13%, respectively, of the net radiation. The dominance of the sensible heat flux is more obvious and occurs closer to the northwest arid area (Zhang et al., 2007). However, at the wettest spot, the surface sensible heat flux, latent heat flux, and soil heat flux account for 49.7%, 49.9%, and 0.37%, respectively, of the net radiation. The contribution of the sensible heat flux and latent heat flux is almost equal, which is similar to the characteristics of semi-humid regions and reflects that the Loess Plateau stretches across arid, semiarid, and sub-humid zones. The Loess Plateau is located at the edge of the summer monsoon. Variations of the summer monsoon cause significant fluctuations in the annual precipitation and change the spatial distribution pattern of the land-surface energy balance. To understand the influence of extreme climates on

January (2014) Vol.57 No.?

the spatial distribution pattern of the land-surface energy balance, we made a comparison of the spatial distribution characteristics of the surface energy balance components over the Loess Plateau region in the driest year (1996), when the annual average soil moisture was at the minimum, and the wettest year (1990), when the annual average soil moisture was at the maximum (not shown). The surface net radiation change in the driest and wettest years is not obvious, and it generally increases gradually from south to north and is similar to the 30-year average. However, the slant of the surface sensible heat flux is greater overall in the driest year, and the changing gradient from south to north is stronger. During the wettest year, the surface sensible heat flux is slightly smaller overall and the spatial change gradient from south to north is weaker and closer to the 30-year average. The surface latent heat flux in the driest year is obviously smaller, and the spatial change gradient from southeast to northwest is weaker. In the wettest year, the surface latent heat flux is slightly greater, and the spatial variation gradient from southeast to northwest is slightly stronger and closer to the 30-year average. The highest value center of the soil heat flux in the driest year is in the northeast, and in the wettest year it moves to the southwest. The dry-wet fluctuations can transform the spatial distribution pattern. In general, the influence of the interannual dry-wet fluctuations on changes of the surface heat flux spatial distribution is more significant than on the surface radiation budget spatial distribution. Figure 6 shows a comparison of the spatial average of the surface energy balance components and allocation rate of the 30-year average in the driest year (1996) and wettest year (1990). Based on a quantitative analysis, the surface

Figure 6 Comparison of the spatial average of the surface energy balance components (a) and allocation rate (b) of the 30-year average in the driest year (1996) and wettest year (1990). SH/Rn, the proportion of sensible heat in the surface energy; LH/Rn, the proportion of latent heat in the surface energy; G/Rn, theproportion of soil heat flux in the surface energy. The others are the same as in Figure 5.

Zhang Q, et al.

Sci China Earth Sci

net radiation fluxes in the driest and wettest years are both smaller than the 30-year average at approximately 8.0 and 2.5 W/m2, respectively. The surface sensible heat flux is approximately 11.1 W/m2 greater than the 30-year average in the driest year and approximately 4.5 W/m2 smaller than the 30-year average during the wettest year. Compared with the sensible heat flux, the latent heat flux is approximately 20.5 W/m2 smaller than the 30-year average in the driest year and approximately 1.8 W/m2 larger than the 30-year average during the wettest year. The soil heat flux in the driest and wettest years is approximately one and half times larger than the 30-year average and corresponds to the net radiation flux in the driest and wettest years, which is smaller than the 30-year average. In terms of the allocation rate of surface energy balance components, the proportion of the surface sensible heat flux, latent heat flux, and soil heat flux in the net radiation in the driest year is 92.4%, 6.8%, and 0.81%, respectively, and that in the wettest year is 63.9%, 35.4%, and 0.66%, respectively. Compared with the sensible heat flux, the value of the driest year is 24.8% greater than the 30-year average and that of the wettest year is 3.8% smaller than the 30-year average. Compared with the latent heat flux, the value of the driest year is 25.3% smaller than the 30-year average and that of the wettest year is 3.4% greater than the 30-year average. The characteristic of the driest year is similar to that of arid areas, whereas the characteristics of the wettest year and 30-year average are similar to that of semiarid areas. Based on the spatial average, these results show that no matter how the climate fluctuates, the Loess Plateau region has generally arid and semiarid climate conditions. However, annual changes in dry and wet conditions do have an effect

Figure 7

January (2014) Vol.57 No.?

9

on variations of surface energy in spatial distribution. 3.2 Change of the surface energy balance components along with latitude and longitude Because the solar radiation intensity changes regularly with latitude and the summer monsoon gradually moves from southeast to northwest, the land-surface energy balance components over the Loess Plateau region also change with latitude and longitude. Figure 7 shows the change of the surface energy balance components with latitude (Figure 7(a)) and longitude (Figure 7(b)). The spots in the figure represent the average of the surface energy balance components with latitude (Figure 7(a)) and longitude (Figure 7(b)). The surface energy balance components clearly vary with latitude, and the change of surface sensible heat flux can reach approximately twice as high as the initial value, but the latent heat flux and soil heat flux can only reach approximately half of the initial value. Although the overall change of surface net radiation is not obvious, it has a peak value at 37.5°N and a low value at 34.5° and 40.5°N, which reflects the combined influence of solar radiation, altitude, and monsoon rainfall. The surface latent heat flux and soil heat flux decrease with an increase of latitude, which mainly reflects the influence of monsoon rainfall. The surface sensible heat flux overall increases with an increase of latitude, and its weak peak and low values are synchronous with that of net radiation, which indicates that although the sensible heat flux is affected primarily by the summer monsoon rainfall, the net radiation forcing can produce an obvious disturbance. Figure 7 also shows that the change of the surface energy

The 30-year average surface energy balance components vary with latitude (a) and longitude (b). The others are the same as in Figure 5.

10

Zhang Q, et al.

Sci China Earth Sci

balance components with longitude is not as obvious as the change with latitude. The regularity is even more disparate, which is related to the physical mechanism implied by longitude, although it is not as clear as the mechanism implied by latitude. However, the surface net radiation increases more obviously with longitude than with latitude, which is closely related to the topography and elevation changes. From west to east, the area near western Wushaoling is generally 3000–3500 m a.s.l. (101.5°E). The elevation gradually reduces to the east, and then rises again to the Liupan Mountains, but it is still less than 3000 m. The Hetao Plain and Guanzhong Plain are located east of the Liupan Mountains (106.5°E), and the elevation declines further. The average elevation of the Hetao Plain is 1100–1200 m, and the Guanzhong Plain is much lower. Continuing to the east is the Taihang Mountains (110°E) where the altitude increases again. From the change of net radiation with longitude, the net radiation begins to increase gradually at 106.5°E and has a weak peak at 108.5°E near the Hetao and Guanzhong plains, and it then increases with decreases in height after 110°E. This change reflects the dual effects of altitude and the summer monsoon rainfall. The surface sensible heat flux has no obviously increasing trend, but it also has a peak at 108.5°E, which reflects the control of net radiation over the sensible heat flux. The surface latent heat flux increases with increased longitude, reflecting that the summer monsoon precipitation has a more significant influence on the

Figure 8 flux (d).

January (2014) Vol.57 No.?

surface latent heat flux than the other components of surface energy, which is consistent with the conclusions from the previous analysis. The overall soil heat flux decreases with increases of longitude, but it has a low point at 105.5°– 106.5°E.

4 Influence of spatial change of climate factors on the surface energy flux We previously discussed the spatial distribution of the surface energy balance components related to solar radiation, summer monsoon precipitation, altitude and numerous other factors. In many cases, the roles of various factors always overlap and it is difficult to distinguish between them. Therefore, a spatial correlation analysis is further performed between the surface energy balance components and climate elements such as temperature and precipitation, and environmental elements such as altitude. 4.1 Influence of climate factors Precipitation is the most important climate factor, so in Figure 8, the respective spatial correlations are provided between precipitation and the surface net radiation, sensible heat flux, latent heat flux and soil heat flux. Obviously, the surface net radiation does not change with precipitation

The respective spatial correlations between precipitation and the surface net radiation (a), sensible heat flux (b), latent heat flux (c), and soil heat

Zhang Q, et al.

Sci China Earth Sci

distribution. However, other surface energy balance components show a strong correlation with precipitation. The correlation coefficient of precipitation and the surface sensible heat flux, latent heat flux, and soil heat flux is 0.60, 0.91, and 0.71, respectively. The latent heat flux has the best correlation with precipitation, which is reasonable and consistent with previous analyses. The annual precipitation spans from 100 to 600 mm, with the surface sensible heat flux reducing from approximately 69 to 42 W/m2, surface latent heat flux increasing from approximately 13 to 35 W/m2, and soil heat flux increasing from 0.09 to 0.26 W/m2. For every 100 mm increased precipitation, the surface sensible heat flux decreases by approximately 5.0 W/m2 and latent heat flux and soil heat flux increase by 4.5 and 0.04 W/m2, respectively. Temperature is another important climatic element, and it can reflect the effects of solar radiation and precipitation. Over the Loess Plateau, the correlation between the surface energy balance components and temperature is not as good as the correlation between the surface energy balance components and precipitation, and the correlation coefficients are all relatively lower (not shown). However, the surface soil heat flux and latent heat flux have somewhat better correlations with temperature because they change with temperature more obviously. 4.2

Influence of altitude

Altitude is a key physical factor that affects the land-surface energy balance components, such as the intensity of solar radiation and surface temperature. The effect of the large altitude drop of the Loess Plateau region on the land- surface energy balance components cannot be ignored. Figure 9 shows the respective spatial correlation between altitude and the surface net radiation, sensible heat flux, latent heat flux, and soil heat flux, which indicates that the correlation between altitude and surface net radiation is the highest, with the correlation coefficient reaching 0.51. Sensible heat flux is the next highest with a correlation coefficient of 0.39. The correlations between surface soil heat flux and altitude and surface latent heat flux and altitude are not very strong, close to 0.2, which is consistent with the influencing mechanism of altitude on the surface energy balance. From the overall trend, the surface net radiation, sensible heat flux, and latent heat flux decrease with increases of altitude. For every 100 m increase of altitude, the surface net radiation, sensible heat flux, and latent heat flux reduce by approximately 0.63, 0.42, and 0.21 W/m2, respectively. The soil heat flux increases by approximately 0.04 W/m2 with every 100 m increase of altitude.

5 Discussion and conclusions CLM can accurately simulate the surface energy balance

January (2014) Vol.57 No.?

11

components of the Loess Plateau. The simulated results are highly correlated with observed values at the same grid point, and the root mean square error is small. Compared with the available observation data and satellite remote sensing inversion data, the simulated data have some advantages in studying the spatial distribution of the land- surface energy balance. The spatial variation of geographical factors such as latitude, longitude, altitude and climate elements such as precipitation and temperature affect the spatial distribution characteristics of the land-surface energy balance components. The latitude and longitude span, altitude drop, and rainfall spatial differences over the Loess Plateau area are all significant; therefore, the land-surface energy balance components are significant in the spatial variation and distribution patterns. Specifically, the surface net radiation and sensible heat fluxes gradually increase from south to north and have a closer relationship with latitude. The latent heat flux and soil heat flux gradually reduce from southeast to northwest and have a closer relationship with precipitation. The surface energy balance components over the Loess Plateau are significantly affected by the spatial variations in climate. In addition to the surface net radiation, differences in the surface energy balance components, such as the sensible heat flux at the driest point is nearly twice as high as that at the wettest point, whereas the latent heat flux and soil heat flux at the driest point are half as much as that at the wettest point. The surface energy distribution at the driest point is dominated by the sensible heat flux, but in wettest lattice, there is an obvious change to an equal contribution of sensible heat flux and latent heat flux. Although the interannual dry-wet fluctuations have an obvious influence on the surface energy balance components, they are less significant than the influence of the spatial dry-wet difference on the surface energy balance components. Although the surface energy balance components change with latitude and longitude, the change with latitude is more obvious. The surface sensible heat flux generally increases with increases of latitude, but the surface latent heat flux and soil heat flux reduce with increases of latitude. These differences reflect that the surface sensible heat flux is controlled mainly by solar radiation, whereas the surface latent heat flux and soil heat flux are primarily controlled by the summer monsoon precipitation. On the Loess Plateau, the land-surface energy balance components have strong spatial correlations with precipitation and altitude. For every 100 mm increase of annual precipitation, the surface sensible heat flux decreases by approximately 5.0 W/m2 and latent heat flux and soil heat flux increase by 4.5 and 0.04 W/m2, respectively. For every 100 m increase of altitude, the surface net radiation, sensible heat flux, and latent heat flux reduce by approximately 0.63, 0.42, and 0.21 W/m2, respectively. The soil heat flux increases by approximately 0.04 W/m2 with every 100 m increase of

12

Figure 9

Zhang Q, et al.

Sci China Earth Sci

January (2014) Vol.57 No.?

The spatial correlation between altitude and the surface net radiation (a), sensible heat flux (b), latent heat flux (c), and soil heat flux (d).

altitude. Temperature does not appear to be related to the spatial distribution of the land- surface energy balance components. Overall, the surface net radiation and altitude are more closely related. The sensible heat flux has a good relationship with the monsoon precipitation and altitude. The latent heat and soil heat flux are more closely related to the monsoon rainfall. In this paper, we provided a comprehensive distribution pattern of the land-surface energy balance and its relationship with climate and geographical factors using the simulated results from a land-surface model. However, because the land-surface process parameterizations of the CLM model have not been improved over the Loess Plateau region, the simulated surface energy balance components may be somewhat uncertain, which may affect the accuracy of its analyses. In the future, the parameterization results from LOPEX can be used to further improve the ability of the CLM model on the Loess Plateau. And by performing comparisons and validations with additional observational information to improve the quality of the simulations, we will be able to further analyze the distribution characteristics of the landsurface energy balance and influencing mechanisms over the Loess Plateau region. This work was supported by the State Key Program of National Natural Science of China (Grant No. 40830957) and the National Key Basic Re-

search Program (Grant Nos. 2013CB430200, 2013CB430206). In this paper, Global Land Surface Data Assimilation System developed by Goddard Space Flight Center of NASA and National Centers for Environmental Prediction of NOAA provides the land surface process simulation data. The SACOL station of College of Atmospheric Sciences of Lanzhou University provides the land surface process field observation data. We thank them all. Bonan G B. 1998. The land surface climatology of the NCAR land surface model coupled to the NCAR. Community Climate Model. J Clim, 11: 1307–1326 Cheng X R, Yu M K, Huang M B. 2011. Simulation of energy balance of SVTA system in the farming -pastoral zone of the Loess Plateau. Acta Prata Sin, 20: 160–180 Dickinson R E. 1995. Land-atmosphere interaction. US National Report to International Union of Geodesy and Geophysics 1991–1994. Rev Geophys, (Suppl): 917–922 Dai Y J, Zeng Q C. 1997. A land surface model (IAP94) for climate studies I: Formulation and validation in off-line experiments. Adv Atmos Sci, 14: 433–460 Dai Y J, Zeng X B, R E Dickinson, et al. 2003. The common land model. Bull Amer Meteorol Soc, 84: 1013–1023 Ek M B, K E Mitchell, Y Lin, et al. 2003. Implementation of Noah land surface model advances in the National Centers for Environmental Prediction operational mesoscale Eta model. J Geophys Res, 108: 8851 Guan X D, Huang J P, Guo N, et al. 2009. Variability of soil moisture and its relationship with surface albedo and soil thermal parameters over the Loess Plateau, Adv Atmos Sci, 26: 692–700 Han L J. 2006. The study on Assimilating MODIS LST product into the Common Land Model. Doctoral Dissertation. Beijing: Beijing Normal University Huang J P, Zhang W, Zuo J Q, et al. 2008. An overview of the semiarid climate and environment research observatory over the Loess Plateau.

Zhang Q, et al.

Sci China Earth Sci

Adv Atmos Sci, 25: 906–921 Huang R H, Chen W, Zhang Q. 2011. Land-atmosphere Interaction in Northwest Arid Areas of China and Its Impact on Climate Change in East Asia, Beijing: China Meteorological Press. 356 Huang J P, Guan X D, Ji F. 2012. Enhanced cold-season warming in semiarid regions Atmos. Chem Phys Discuss, 12: 4627–4653 Koster R D, M Suarez. 1996. Energy and water balance calculations in the Mosaic LSM. NASA Tech Memo 104606, 9: 59 Li B, Zhang J T. 2003. Analysis of relationships between vegetation and climate variables in Loess Plateau. Acta Ecol Sin, 23: 82–89 Li Z C, Wei Z G, Wen J, et al. 2008. Analyses on climate variation characteristics over the Loess Plateau in recent 50 years. J Arid Res Environ, 22: 57–62 Li H Y, Zhang Q, Wang S. 2010. Research on characteristics of landsurface radiation and heat budget over the Loess Plateau of central Gansu in summer. Adv Earth Sci, 25: 1070–1081 Lu G H, Yan G X, Wu Z Y. 2010. Analysis on aridification with the last 50 years in China. Adv Sci Tech Water Resour, 41: 78–82 Li H Y, Zhang Q, Zhao J H. 2012. The influences of air heat storage, plant photosynthesis and soil water movement on surface energy balance over the Loess Plateau. Acta Phys Sin, 61: 159–201 Liu Y, J Huang, G Shi, et al. 2011. Aerosol optical properties and radiative effect determined from sky-radiometer over Loess Plateau of Northwest China. Atmos Chem Phys, 11: 11455–11463 Ma Z G, Fu C B. 2005a. Decadal variations of arid and semiarid boundary in China. Chin J Geophys, 48: 519–525 Ma Z G, Huang G. 2005b. Multi-scale temporal characteristics of the dryness/wetnessover northern China during the last century. Chin J Atmos Sci, 29: 671–681 Mao K B. 2004. Study of methodology to invert land surface temperature basing on MODIS. Masters Dissertation. Nanjing: Nanjing University Ngo-Duc T, Polcher J, Laval K. 2005. A 53-year forcing data set for land surface models. J Geophys Res, 110: D06116 Rodell M, Houser P R, Jambor U, et al. 2004. The global land data assimilation system. Bull Am Meteorol Soc, 85: 381–394 Song Y M, Guo W D, Zhang Y C. 2009. Performances of CoLM and NCAR_CLM310 in simulating land-atmosphere interactions over typical forest ecosystems in China. Part II: Impact of different parameterization schemes on simulations. Clim Environ Res, 14: 229–242 Sun C Z, Huang B L, Chen H B, et al. 1998. Interaction between soil water conditions and different kinds of artificial plant cover in the Loess Plateau. J Beijing Forest Univ, 20: 7–14 Sun Z X, Zhang Q. 2011. Analysis of climate characteristics of land surface temperature and energy in the semiarid region in the Loess Plateau. J Desert Res, 31: 1302–1308 Thornthwaite C W. 1948. An approach toward a rational classification of climate. Geogr Rev, 38: 55–94 Wu G X, Liu H, et al. 1995. Climate Physics. Beijing: China Meteorological Press. 425 Wang G, Huang J, Guo W, et al. 2010. Observation analysis of land-atmosphere interactions over the Loess Plateau of northwest China. J Geophys Res, 115: D00K17 Wang Y Y, Zhang Q, Jiang S B. 2011. A review of recent study advances in climatic environment on China Loess Plateau. Adv Meteorol Sci Tech, 2: 38–41

January (2014) Vol.57 No.?

13

Wang Y P, E Kowalczyk, R Leuning, et al. 2011. Diagnosing errors in a land surface model (CABLE) in the time and frequency domains. J Geophys Res, 116: G01034 Yu W D, Min Q W, Li X G. 2002. The features of precipitation in the Loess Plateau and its possible impacts on vegetation distribution. Res Sci, 24: 55–60 Yan H R, Huang J, P Minnis, et al. 2011. Comparison of CERES surface radiation fluxes with surface observations over Loess Plateau. Remote Sens Environ, 115: 1489–1500 Zeng J, Zhang Q, Wang S. 2007. Regional differences in the characteristics of clear-sky land surface processes in distinct climatic zones over northern China. Chin J Atmos Sci, 35: 483–492 Zeng J, Shen J, Zhang Q. 2010. An overview of the spatial patterns of land surface processes over arid and semiarid regions. Sci Cold Arid Region, 2: 288–297 Zhang Q, Wang S. 2007. Study on annual variation of hydrothermal feature of soil and land surface energy balances in arid hungriness area. Prog Nat Sci, 17: 211–216 Zhang Q, Wang S. 2008. On land surface processes and its experimental study in Chinese Loess Plateau. Adv Earth Sci, 23: 167–173 Zhang Q, Hu X J, Wang S, et al. 2009. Some technological and scientific issues about the experimental study of land surface processes in Chinese loess. Adv Earth Sci, 24: 363–371 Zhang Q, Li H Y. 2010. The relationship between surface energy balance unclosure and vertical sensible heat advection over the Loess Plateau. Acta Phys Sin, 8: 5888–5895 Zhang J, Zhang Q, Huang J P. 2010. Research on land surface energy flux process in Loess Plateau in May–October 2007. Plateau Meteorol, 29: 855–863 Zhang Q, Huang R H, Wang S. 2011a. Discussion about special function of land surface process and atmospheric boundary on regional climate in arid area of northwest China. Arid Meteorol, 29: 133–136 Zhang Q, Sun Z X, Wang S. 2011b. Analysis of variation regularity of land-surface physical quantities over Dingxi Region of the Loess Plateau. Chin J Geophy, 54: 1727–1737 Zhang Q, Li H Y, Zhao J H, et al. 2012a. Modification of the land surface energy balance relationship by introducing vertical sensible heat advection and soil heat storage over the Loess Plateau. Sci China Earth Sci, 55: 580–589 Zhang Q, Zeng J, Zhang Li Y. 2012b. Characteristics of land surface thermalhydrologic processes for different regions over North China during prevailing summer monsoon period, Sci China Earth Sci, 55: 1872–1880 Zhang Q, Li H Y, Yue P. 2013a. Responses of the land-surface process and its parameters over the natural vegetation underlying surface of the middle of Gansu in loess plateau to precipitation fluctuation. Acta Phys Sin, 62: 019201 Zhang Q, Huang J, Zhang L, et al. 2013b. Warming and drying climate over Loess Plateau area in China and its effect on land surface energy exchange. Acta Phys Sin, 13: 139202 Zuo J, Huang J, Wang J, et al. 2009. Surface turbulent flux measurements over the Loess Plateau for a semiarid climate change study, Adv Atmos Sci, 26: 679–691 Zuo J, Wang J, Huang J, et al. 2011. Estimation of ground heat flux and its impact on the surface energy budget for a semiarid grassland. Sci Cold Arid Region, 3: 41–50