126 S.G. Li et al. Agricultural and Forest Meteorology 102 2000 125–137
dynamics of deserts in the Sahara be controlled by a biogeophysical feedback mechanism. Biological
feedbacks play an important role in desertification worldwide Schlesinger et al., 1990. Deforestation
and resulting hydrological changes of land surface affect regional and even global climates e.g. Shukla
and Mintz, 1982; Shukla et al., 1990; Nobre et al., 1991; Wright et al., 1992. Some GCMs suggest
that future global warming will likely exacerbate the degradation of semi-arid grasslands on a large scale
in North America and Asia Manabe and Wetherald, 1986. It is also true in the arid and semi-arid zones
of Northern China that deforestation and afforestation affect the regional pattern of precipitation and tem-
perature regime Chen et al., 1992; Lin et al., 1993. For example, in the Horqin Steppe, the desertified
grassland area is still expanding due to inappropriate anthropogenic activities such as prolonged overgraz-
ing, poor farmland management and over-lumbering Zhu and Wang, 1992. Vegetation plays very impor-
tant roles in mass and heat exchange between land surfaces and the atmosphere Grace et al., 1981.
Removal or deterioration of vegetation cover affects water and heat budget of the surfaces. Therefore, in
the desertification-prone areas such as the Horqin Steppe, it is requisite to conduct research on the re-
lationship between desertification mainly caused by degradation of vegetation and micrometeorological
changes. The primary objectives of this paper are: 1 to describe impacts of different levels of grazing
intensity on the micrometeorological characteristics of grassland; and 2 to assess initiation of grassland
desertification under increasing overgrazing.
2. Materials and methods
2.1. Site descriptions and instrumentation The studied area is located at Daliushu village
lat. 42
◦
58
′
N, long. 120
◦
43
′
E, 345 m above mean sea level, Naiman, Inner Mongolia, China, in a zone of
continental semi-arid monsoon climate Fig. 1. The mean meteorological conditions are summarized as
follows by Liu et al. 1997: The mean annual solar radiation is 5200 MJ m–
2
. The total annual sunshine duration is about 2946 h. The mean annual air tem-
perature is about 6.4
◦
C, and the coldest and warmest
Fig. 1. Location of measurement field. Plots A, B, C, and D represent the ungrazed plot, the lightly grazed plot, the moderately
grazed plot, and the overgrazed plot at Daliushu village, Naiman, Inner Mongolia, respectively. The dots in the plots indicate the
micrometeorological measurement sites. The arrow beside the plots indicates the prevailing wind direction during the measurements.
monthly mean temperatures are −13.1
◦
C of January and 23.7
◦
C of July, respectively. The mean annual accumulated air temperature above 10
◦
C ranges from 3000 to 3400
◦
C days. The frost-free period is ca. 137–150 days per year. The normal annual pre-
cipitation is about 362 mm, nearly 70 of which falls from June through August. The annual mean
pan-evaporation is around 1935 mm. Prevailing wind directions are northwest in winter and spring, and
southwest to south in summer and autumn. The mean annual wind-speed ranges from 3.4 to 4.1 m s
− 1
. The threshold for sand movement, 5 m s
− 1
, is exceeded during more than 200 days per year mainly in spring
and winter. Gales wind speed ≧ 17.2 m s
− 1
occur from 16.9 to 24.1 days per year. The geomorpho-
logic landscape in this region is characterized by sand dunes alternating with gently undulating interdunal
lowlands where farmlands and residential points are distributed. The zonal soils are sandy with light yel-
low color, coarse texture and loose structure. The sandy land is classified into three grassland types:
shifting, semi-fixed and fixed. The dominant plant in the shifting sand dunes is Agriophyllum squarrosum.
The dominant plant in the semi-fixed sandy land is Artemisia halodendron, with lesser numbers of Agrio-
phyllum squarrosum, Artemisia scoparia, Lespedeza davurica, Caragana microphylla, Hedysarum fruti-
cosum var. ligosum, and Salix gordejevii. The major plants in the fixed sandy land are Caragana micro-
S.G. Li et al. Agricultural and Forest Meteorology 102 2000 125–137 127
phylla, Lespedeza davurica, Artemisia frigida and some grasses.
Sandy grassland in Naiman region is ecologically fragile and is subject to desertification. There are many
quantitative andor qualitative criteria for assessing the degree of desertification e.g., Mabbutt, 1984, 1986;
Dregne, 1985, 1991; Zhu et al., 1989. Zhu’s crite- ria were used in the field to judge whether or not
grassland had been desertified. Desertification initiates when wind erosion and sand accumulation occur in
the grassland. If vegetation coverage was reduced to 30–50 and area of shifting sand accounts for about
5–20 the total grassland area, then the grassland was regarded as moderately desertified. If vegetation cov-
erage was reduced to 10–30 and area of shifting sand accounts for about 20–50 the total grassland area,
then the grassland was regarded as severely deserti- fied. If vegetation coverage was reduced to less than
10 and area of shifting sand accounts for more than 50 the total grassland area, then the grassland was
regarded as very severely desertified. Grazing density for studied sandy grassland was about five to eight
sheep or sheep equivalents before a treatment of en- closure Zhao, 1992.
In 1992 we chose open and level interdunal sandy grassland as the observation site; the vegetation at this
site has been slightly desertified due to poor environ- mental management. Dominant plants are graminoids,
accompanied by some legumes and forbs; shrubs and semi-shrubs are few. In the middle of this site, an area
of 5.24 ha was fenced with cement piles and barbed wires as the grazing experiment field. This field was
further divided into four plots according to grazing sheep numbers, i.e., Plot A, ungrazed 0; Plot B,
lightly grazed three sheep, Plot C, moderately grazed six sheep, and Plot D, overgrazed nine sheep. Plot
A was 0.74 ha 200×37 m and each of the other three plots had an area of 1.5 ha 200×75 m. The long sides
of these plots were set along the dominant wind di- rection of growing season to guarantee sufficient fetch
for meteorological measurements. Grazing was started on 1 June and ended on 30 September each year for 3
years from 1992. From 1991 to 1994 micrometeoro- logical measurements were conducted at the unfenced
grassland Plot E, which was located roughly at the middle of the experimental field before enclosure, at
each of the experimental grazing plots, and at a mobile sand dune Plot F. The dune was about 1 km to the
northeast of the grazing plots. The fetch at the dune measurement site was more than 800 m. These mea-
surements were each made on intermittent clear days generally 3–5 days in growing season across each of
the study years, and the measurements were not con- ducted simultaneously. The monthly mean precipita-
tion during the measurements is shown in Table 1.
Wind profiles and wind directions were mea- sured with photo-electric cup anemometers Makino,
AF750S, Tokyo, Japan, accuracy ±3 at 1–10 m s
− 1
and a photo-electric anemoscope Makino, VF016, Tokyo, Japan, accuracy ±3. Air temperatures
and relative humidity were measured with type-T thermocouple thermometers ventilated, fan speed
3.5 m s
− 1
, accuracy ±0.1
◦
C and electrostatic ca- pacitive humidity sensors that were shielded by white
plastic covers against direct sunshine North Hightech Co. Ltd., HU-1A, Sapporo, Japan, accuracy ±3 at
25
◦
C and in 30–90 relative humidity. The wind, air temperature, and humidity sensors were mounted on a
6 m mast. Solar radiation and reflected solar radiation were measured with thermal-charged pyranometers
Iio, S-SR2, Tokyo, Japan, accuracy ±3. Net radi- ation was measured with a ventilated net radiometer
EKO, CN-11, Tokyo, Japan, accuracy ±3 at −15 to 40
◦
C. Soil heat flux was measured with soil heat plates EKO, CN-9, Tokyo, Japan, accuracy ±5
at −20 to 120
◦
C. Soil temperatures were measured with copper-constantan thermocouple sensors. Mea-
surement height or depth minus above dune surface or grass community surfaces were 5.0, 2.5, 1.4, 0.7,
0.3 m for the wind profiles, 5.0 m for wind direction, 1.4 and 0.3 m for air temperature and humidity, 1.3 m
for solar radiation, 1.2 m for net radiation, 1.0 m for reflected solar radiation, −0.01 m for soil heat plates,
Table 1 Monthly mean precipitation mm from April to September during
the measurements and monthly mean precipitation from 1981 to 1991
Months 1981–1991
1991 1992
1993 1994
April 20.2
27.7 4.5
12.7 0.0
May 36.6
19.6 57.4
10.6 64.3
June 74.1
137.1 65.3
56.1 22.5
July 108.7
111.6 143.2
123.8 196.9
August 71.4
8.8 64.2
112.0 116.2
September 53.4
143.3 7.5
2.9 2.1
Sum 344.2
420.4 337.6
305.4 402.0
128 S.G. Li et al. Agricultural and Forest Meteorology 102 2000 125–137
− 0.01, −0.05, −0.1, −0.2 and −0.5 m for soil tem-
peratures, respectively. Data were digitally recorded once every 2 min in a data logger North Hightech Co.
Ltd., IDL3200, Sapporo, Japan. Wind data and radi- ation data were averaged before recording, and other
data were instantaneous values at recording time. The hardness of soil surface at the grazing experimental
plots was measured in August 1994 with a hardness meter Yamanaka-07202301, Fujiwara Scientific Co.
Ltd., Tokyo, Japan. A portable power generator was placed ca. 50 m leeward of the mast for power supply.
The sensors for micrometeorological measure- ment were calibrated in 1990 at National Insti-
tute of Agro-Environmental Sciences NIAES. The photo-electric cup anemometers were cross calibrated
in a wind tunnel of NIAES. The platinum resistance thermometers and the type-T thermocouple ther-
mometers were cross calibrated in comparison with an enclosed scale type thermometer certificated by Japan
Meteorological Agency, and the correction factor of each sensor was determined to remove systematic
difference among the thermometers. The instrumen- tal offsets of the pyranometers and radiometer were
also examined at the roof floor of NIAES. Correction factor of each sensor was determined through each
calibration curve. The same procedure was made each year in Naiman Station of Desertification Research of
Chinese Academy of Sciences or NIAES before field measurements.
2.2. Methods The heat budget over both the plant communities
both on the unfenced grassland and the experimental grazing plots and the surface of the sand dune is given
by
R
n
= H + lE + G
1 where R
n
is net radiation W m
− 2
, H is sensible heat flux W m
− 2
, lE is latent heat flux W m
− 2
l is latent heat of vaporization of liquid water and E is evapo-
transpiration rate, and G is heat flux to and from the soil W m
− 2
. In general, R
n
is positive in daytime and negative at night. The sign convention for H, lE and G
adopted here is such that each of the vertical fluxes H, lE and G is positive upwards and negative downwards.
According to the Bowen ratio method Thom, 1975, H =
R
n
− G
1 + β
− 1
2 lE =
R
n
− G
1 + β 3
where β =
H lE
= γ
δT δe
= γ
T
a1
− T
a2
e
1
− e
2
4 in which β is the Bowen ratio, γ is the psychometric
constant, T
a1
and T
a2
are the air temperatures at two levels, and e
1
and e
2
are the corresponding water vapor pressures at the two levels. Thus if the air temperatures
and water vapor pressures are known, and R
n
and G are obtained by direct measurement, then H and lE
can be calculated from Eqs. 2 and 3, respectively. Under neutral conditions, the vertical wind shear
or wind-speed gradient over uniform, level ground or over a plant community of uniform height can be de-
scribed by Thom, 1975,
du dz
= u
∗
kz − d 5
or uz =
u
∗
k lnz − d
z
o
6 where u is the mean wind-speed at height z; u
∗
is the friction velocity u
∗
= τρ
0.5
, τ is the surface shear stress; k is von Karman’s constant 0.41; d is the
zero-plane displacement, usually 0.6–0.7 of the plant height Grace, 1977; and z
o
is called the roughness length of the surface, often equaling about 10 of the
length of surface protuberances Grace, 1977. u
∗
, d, and z
o
were computed from the measured wind pro- files under almost neutral atmospheric stability abso-
lute values of the gradient Richardson number were less than 0.05 by means of the least squares method
reiteration calculation Harazono et al., 1992:
lnz − d = a × uz + b 7
u
∗
= k
a 8
z
o
= expb
9 The roughness length of the dune surface was calcu-
lated using the relation put forward by Chamberlain 1983:
S.G. Li et al. Agricultural and Forest Meteorology 102 2000 125–137 129
z
o
= 0.016u
2 ∗
g 10
where g is the gravitational constant. 2.3. Data processing
The following criteria were adopted for data se- lection and analysis. 1 Most data were observed in
fine days or in a bit cloudy days. 2 Wind directions were mainly southwest to south 180–240
◦
. In case of the ungrazed plot, wind angles were within 194–232
◦
. 3 Data obtained when wind speed was smaller than
0.5 m s
− 1
were rejected. 4 Missing meteorological data were filled via interpolation between earlier and
later measurements. 5 Mean values of 30 min in day- time were used for analysis.
3. Results