Sie sind auf Seite 1von 10

Pedosphere 20(3): 342–351, 2010

ISSN 1002-0160/CN 32-1315/P


c 2010 Soil Science Society of China
Published by Elsevier Limited and Science Press

Using Remote Sensing and GIS Technologies to Estimate Grass


Yield and Livestock Carrying Capacity of Alpine Grasslands
in Golog Prefecture, China∗1

YU Long1,2,∗2 , ZHOU Li1 , LIU Wei1 and ZHOU Hua-Kun1


1 Northwest Institute of Plateau Biology, Chinese Academy of Sciences, Xining 810001 (China)
2 Graduate University of the Chinese Academy of Sciences, Beijing 100049 (China)
(Received November 10, 2009; revised January 12, 2010)

ABSTRACT
Remote sensing data from the Terra Moderate-Resolution Imaging Spectroradiometer (MODIS) and geospatial data
were used to estimate grass yield and livestock carrying capacity in the Tibetan Autonomous Prefecture of Golog, Qing-
hai, China. The MODIS-derived normalized difference vegetation index (MODIS-NDVI) data were correlated with the
aboveground green biomass (AGGB) data from the aboveground harvest method. Regional regression model between the
MODIS-NDVI and the common logarithm (LOG10) of the AGGB was significant (r 2 = 0.51, P < 0.001), it was, there-
fore, used to calculate the maximum carrying capacity in sheep-unit year per hectare. The maximum livestock carrying
capacity was then adjusted to the theoretical livestock carrying capacity by the reduction factors (slope, distance to water,
and soil erosion). Results indicated that the grassland conditions became worse, with lower aboveground palatable grass
yield, plant height, and cover compared with the results obtained in 1981. At the same time, although the actual livestock
numbers decreased, they still exceeded the proper theoretical livestock carrying capacity, and overgrazing rates ranged
from 27.27% in Darlag County to 293.99% in Baima County. Integrating remote sensing and geographical information
system technologies, the spatial and temporal conditions of the alpine grassland, trend, and projected stocking rates could
be forecasted for decision making.
Key Words: grassland degradation, grass production, MODIS production, overgrazing rate, restoration

Citation: Yu, L., Zhou, L., Liu, W. and Zhou, H. K. 2010. Using remote sensing and GIS technologies to estimate grass
yield and livestock carrying capacity of alpine grasslands in Golog Prefecture, China. Pedosphere. 20(3): 342–351.

INTRODUCTION

Grasslands have provided fundamental goods and services to humankind for millennia. In many
mountainous regions of the Tibetan Plateau, especially in the source region of the Yellow River, pas-
toralists (livestock herders) have benefited from and maintained alpine grassland functions through
sustainable land use practices. In recent decades, however, the increasing demand for natural resources
and animal products to cope with sharply rising human populations has placed tremendous pressures
on grassland ecosystems. Serious grassland degradation is endangering the fragile environment of this
region (Zhou et al., 2003). Degraded grassland occupies approximately one-third of the area of all the
investigated grasslands in the source region of the Yellow River, and the area of degraded grassland is
increasing. Many independent research programs have concluded that the current productivity of the
plateau grasslands is approximately 30% lower than the productivity measured only two decades ago
(Zhou et al., 2005). At the same time, overgrazing caused by an increase in the population of humans
and domestic livestock in the source region of the Yellow River (Zhou et al., 2003) is considered to
∗1 Supported by the National Basic Research Program (973 Program) of China (Nos. 2009CB421102 and 2005CB422005-
01), the Second Scheme of CAS Action Plan for the Development of Western China (No. KZCX2-XB2-06-02), and the
National Key Technology R&D Program of China (No. 2006BAC01A02-01).
∗2 Corresponding author. E-mail: yuhulong@yahoo.com.cn.
GRASSLAND ESTIMATION USING REMOTE SENSING AND GIS 343

be the major cause of alpine grassland degradation. So, it is important for the local government and
pastoralists to know their grassland health conditions and how much livestock their grassland can carry
for sustainable utilization of natural resources.
One of the most important indicators for grassland health is vegetation productivity, which is also
one of the most important parameters for calculating the livestock carrying capacity (Holecheck, 1988;
Hunt et al., 2003; Hunt and Miyake, 2006). On the Tibetan Plateau, some studies about the alpine
grassland productivity have been carried out at small sampling area using traditional methods (Li et
al., 2001) or using remote sensing data to monitor the dynamics of grassland yield at large areas (Li
et al., 1998; Yang et al., 2007). Moreover, some researchers have reviewed both the grassland climate-
ecological productivity and their ability for calculating theoretical livestock carrying capacity (Yang and
Yang, 2000). However, there are few studies that have reviewed both the grassland productivity and
the livestock carrying capacity and their relationship in the source region of the Yellow River, especially
using remote sensing method (Qian et al., 2007). Furthermore, usable grassland area used in calculating
carrying capacity in most of those studies was based on outdated grassland cover data because of the
harsh environmental climate, inconvenient traffic condition, and limited district finance.
National Aeronautics and Space Administration (NASA)’s Moderate-Resolution Imaging Spectro-
radiometer (MODIS) data are updated and are free of charge and have improved spatial and spectral
resolutions, better sub-pixel geometric registration with superior calibration, cloud screening, and atmo-
spheric correction (Friedl et al., 2002). Therefore, they are widely applied in all kinds of remote sensing
monitoring, such as rangeland production (Reeves et al., 2006; Xu et al., 2007), land use, and land
cover (Tang and Zhang, 2002; Langner et al., 2007). Cheng (2006) studied the validation of the derived
MODIS-normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) prod-
ucts including single-day MODIS, level 2 (gridded) daily MODIS surface reflectance (MOD09), 16-day
composited MODIS (MOD13), which was performed using multisensory data from MODIS, Landsat
Thematic Mapper (TM), and field radiometer, for a rice-planting region in southern China. Result
showed that the 16-day composited MODIS reflectance matched well with the ground measurement
reflectance. The Terra MODIS 250 m resolution imagery may be sufficient to examine the rangeland
conditions over relatively large land areas.
Owing to their importance in socio-economics, culture, ecology, and environmental quality, grassland
ecosystems have become one of the most active subjects of research and have attracted considerable
attention from ecologists both nationally and internationally. The main aims of this study were to i)
estimate the grass yield in the grassland of Golog Tibetan Autonomous Prefecture (Golog Prefecture),
Qinghai, China using MODIS remote sensing data and ground samping data, ii) map land use and land
cover of Golog Prefecture to provide “up-to-date” usable grassland area information, and iii) determine
the proper livestock carrying of Golog Prefecture.

MATERIALS AND METHODS

Study area

Located in southeast of Qinghai Province, the Golog Tibetan Autonomous Prefecture (Fig. 1) con-
sists of six counties: Maqên, Baima, Gadê, Darlag, Jigzhi, and Madoi, covering an area of 7.64 × 106 ha,
and both the Yangtze River and Yellow River originate here. Moreover, what takes place in the upper
reaches of the catchment has far-reaching effects on downstream areas. Golog Prefecture can be divided
into two approximate topographic sectors. In the northwest, high grasslands at an average elevation
from over 4 000 m up to 5 000 m stretch to the source region of the Yellow River. In the southeast, the
A’nyêmaqên and Bayan Har mountains enclose a lower but more uneven topography ranging from an
elevation as low as 2 500 m to over 4 000 m. The climate of Golog Prefecture is dominated by southeast
monsoon and high atmosphere pressure from Siberia, with severe, long winters and short, cool summers.
The mean annual air temperature varies from −4.1 ◦ C at Madoi to −1.3 ◦ C at Darlag, and the mean
annual precipitation ranges from 326 mm at Madoi to 700 mm at Jigzhi, and approximately 80% of
344 L. YU et al.

precipitation falls in the short growing season from May to September. Over the last 40 years, the
climate has tended to be warmer than recorded previously according to the results of a recent survey
(Zhou et al., 2005).

Fig. 1 Location of the Golog Tibetan Autonomous Prefecture, Qinghai Province, China.

The main types of grassland are the alpine meadow and the alpine dry steppe. The alpine meadow,
one of the most important grassland types in the Golog Prefecture, can be divided into four subtypes:
typical alpine meadow, alpine swamp meadow, alpine steppe meadow, and alpine shrub meadow (Zhou
et al., 2005). The soils include alpine meadow soil, peat soil, alpine grassland soil, alpine desert soil,
subalpine meadow soil, chestnut soil, bog soil, and aeolian sandy soil. Because of high altitude and cold
weather, the soils are generally not developed well in the source region, and soil is thin with a depth
of generally about 30–50 cm. There are many pebbles in the soil. Because of the bad weather and the
high altitude, these soils are hard to restore once they are destroyed (Liu et al., 2006).

Satellite data

The MODIS vegetation indices (VI) are robust spectral measures of the amount of vegetation present
on the ground and can provide new options for regional land cover mapping that are less labor-intensive
than landsat and have higher resolution than previous 1 km advanced very high resolution radiometer
or the current 1 km global land cover product. MODIS/Terra L3 16-day global 250 m VI products
(MOD13Q1) were acquired between July and September 2005 from the NASA Earth Observing System
(EOS) Data Gateway, it contains two indices, the NDVI and a new EVI. A new compositing scheme,
which reduces angular, sun-target-sensor variations, is adopted. It employs the constrained view max-
imum value composite (CV-MVC) and an option to use bidirectional reflectance distribution function
models (MODIS, 1999). Each MOD13Q1 product includes VI quality information in addition to com-
posited surface reflectance bands 1–3 and 7 (red, NIR, blue, and MIR (2 105–2 155 nm)). MOD13Q1
products were designed to provide consistent spatial and temporal comparisons of global vegetation
conditions in support of change detection and phenologic and biophysical interpretations (Justice et al.,
2002). The data in grass growing season (from May to September 2005) were downloaded from the EOS
Data Gateway with bands 1 and 2 in 250 m resolution and bands 3 and 7 in 500 m spatial resolution
at nadir. Landsat TM (TM 5) data (path 133, row 36) of 16 September, 2004 were acquired from the
Remote-Sensing Satellite Ground Station, Chinese Academy of Sciences, which were used as ancillary
GRASSLAND ESTIMATION USING REMOTE SENSING AND GIS 345

data in MODIS data classification.

Grassland classification

From 2001 to 2005, more than 200 representative grass cover samples were collected from the whole
prefecture, and most of them were used as ground truth in grassland classification. Due to their high
spatial resolution, Landsat TM and ETM+ were also used as “ground truth” to collect regions of interest
at large scale. Using MVC method, the MVC of the MODIS NDVI, EVI, red, NIR, blue, and MIR
was first generated in the growing season in 2005, and then this six layers were stacked as grassland
classification file (GCF). Finally, assisted by “ground truth” collected from field and Landsat images,
the MODIS GCF was classified using supervised classification and maximum likelihood decision rules
to generate maps of recent grassland cover (Wang and Tenhunen, 2004).

Grass yield collection and estimation methods

From the end of July to the middle of September in 2005, more than 150 ground truth data were
collected from the large-scale field campaigns organized by the Grassland Monitor and Management
Center of Golog Prefecture, according to the sampling procedure in the Handbooks of Chinese Grassland
Resource and Eco-environment Monitoring Technology. Among each sampling site, aboveground green
biomass (AGGB), species composition, dominant species, vegetation height, canopy cover, site slope,
aspect, grassland type, soil texture, etc. were surveyed and located with a global positioning system
(GPS). AGGB (kg ha−1 ) were obtained at the peak of grass growing season when the green biomass was
dominant. Ground truth data were selected based on grassland type and accessibility and to provide an
even geographical distribution across the six counties of Golog Prefecture. For each site, a transect was
established perpendicular to and started at least 50 m from fence lines to avoid the influence of roads
or channeled livestock trails that typically run along the fence lines. Both palatable and unpalatable or
poisonous herbages were separated and placed in paper bags. AGGB in 1981 came from the Committee
of Agricultural and Animal Husbandry Regionalization of Golog Prefecture (CAAHRGP) (Zhou et al.,
2005). The natural data have originated from observation of all those countries in Golog Agricultural
Regionalization Committee. This regionalization has been implemented for years and is rather authority.
To establish model for grass yield estimation, according to the sampling time, latitude and longitude
of the sample areas, the geographical information system (GIS) method was used to calculate the MODIS
MVC NDVI mean value within the 250 m × 250 m of the sample areas in the time-corresponding NDVI
distribution diagram (3–4 pixel mean values in general). The average weight of fresh grass within the 250
m × 250 m pixel was calculated and transformed based on LOG10. Finally, on the basis of constructing
the database of the NDVI and the transformed average weight of corresponding sampled fresh grass,
the regression equations of NDVI and AGGB were determined (Xu et al., 2007). Ordinary least squares
regression was used to evaluate the statistical significance and the accuracy of the regression models by
Data Processing System software (Tang and Feng, 2006).

Livestock carrying capacity

The number of livestocks in 2005 was collected by the Animal Husbandry Bureau of Department of
Agriculture, Golog Prefecture, while the number in 1981 came from CAAHRGP. It included numbers
of the domestic Tibetan sheep, goat, yak, cattle, horse, donkey, and mule, and all the livestocks were
then conversed into sheep units (SU) (the conversion ratio is goat:horse:yak:cattle:mule:donkey:Tibetan
sheep = 0.8:6:4:5:5:5:1).
Digital elevation data of the Shuttle Radar Topography Mission (SRTM) were used to determine
slope, where the median slope for each 250 m × 250 m grid cell was determined from 90-m grid cells.
Data of county boundaries, lakes and waters (1:1 000 000); soil erosion data at a scale of 1:250 000
of Golog Prefecture were obtained from the Data Center for Resources and Environmental Sciences,
346 L. YU et al.

Chinese Academy of Sciences. All geospatial and remotely sensed data were displayed as an Albers
Conical Equal Area projection using a spherical earth of radius 6 378 245 m. The latitude of the first
and the second standard parallels was 25◦ N and 47◦ N, and the central meridian was 105◦ E.
To ensure the sustainable development of the alpine grassland ecosystem, remotely sensed and other
geospatial data were used to determine the suitable stocking rates. The proper grazing capacity was
calculated as follows:
−1
Ausy = Yy × Ey × Hy × Ius × Dy−1 (1)

Yy = Yym × (1 + Gc ) × Ry−1 (2)

where Ausy is the proper grazing capacity expressed in SU in a specified grassland area and period of
grazing, Yy is the usable grass yield in a grazing season (kg ha−1 ), Ey is the yearly rangeland utilization
rate (55%), Hy is the conversion coefficient for calculation of standard hay (1.0), Ius is the daily intake
of one SU (4.0 kg SU−1 ), Dy is the grazing period (365 days), Yym is the yield of aboveground standard
hay, Gc is the regrowth rate of forage (10%), and Ry is the annual variation rate of forage yield. The
calculation of grazing capacity was based on the fresh forage yield.
According to the proper grazing capacity and other adjustments, the proper carrying capacity was
calculated using the following equation:

CC = Ausy × (1 − DW) × (1 − DS) × (1 − DE) (3)

where CC is the adjusted proper carrying capacity, and DW, DS, and DE are the reduction rates of
available grass yields considering distance to water, slope class, and soil erosion gradients, respectively
(Table I).
TABLE I
Reduction rates and area percentages of the adjustment factors for livestock carrying capacity

Factor Reduction Area


%
Distance to water
0–2 km 0 55.93
2–4 km 50 38.59
> 4 km 100 5.49
Slope
0◦ –5◦ 0 30.90
5◦ –15◦ 30 35.60
15◦ –25◦ 50 23.11
25◦ –35◦ 70 8.98
> 35◦ 100 1.41
Soil erosion
Light to moderate 10 98.11
Serious to extremely serious 0 1.89

RESULTS AND DISCUSSION

Grassland classification

MODIS-derived grassland cover was divided into five types: alpine dry steppe and alpine steppe
meadow (12.86%), typical alpine meadow (52.49%), alpine swamp meadow (6.99%), alpine shrub meadow
(also including sparsely forested woodland and shrub lands) (9.06%), and nonvegetated land (including
river, lake, reservoir, bottomland, town built-up area, village, sands, gobi, bare land, gravel, and other
unused land) (18.59%). The overall accuracy (87.8%) and Kappa coefficient (0.85) have been calculated
based on the ground truth and the Landsat data, respectively.
GRASSLAND ESTIMATION USING REMOTE SENSING AND GIS 347

Grass yield of grassland in Golog

Strong relationships were found between either MVC-NDVI or MVC-EVI and fresh grass yield (Y ,
kg ha−1 ). The optimal regression model was: Y = 101.303NDVI+2.575 (r2 = 0.5, P < 0.001, degree
of freedom = 37), the mean relative estimation error was approximately 18.1%, and the measured
and calculated precision was approximately 72%, which could satisfy the rough estimation of grass
production for pastoral administration (Xu et al., 2007). The optimal regression model corresponded
well with a previous study carried out in Qinghai Province in 2003 by Dong et al. (2005). The same
ranking of the NDVI and EVI was obtained considering the dry biomass of edible and inedible forage
parts as independent variables in the regression analysis. A decrease of 12%–19% of the determination
coefficient was found in this case. Further analysis was, therefore, conducted on fresh biomass-NDVI
relationships. Hence the usable grass yield (Yy ), which is used for calculating livestock carrying capacity,
was estimated by Yy = kY , where k is 0.68, being the ratio of edible forage yield to total grass yield;
and Y is the fresh grass yield.
From 1981 to 2005, the areas of productive grasslands (Classes I, II, III, and IV) decreased by
12.54% while the low-productivity grassland (VII and VIII) increased by 16.68% (Fig. 2). The average
fresh grass yield decreased from 3 605 to 2 954 kg ha−1 in Golog, decreasing dramatically in most of
the counties, especially in Darlag and Baima counties (Fig. 3). Field survey showed that total numbers
of species decreased, and dominated forage species were replaced by unpalatable weed and poisonous
species in the degraded grassland, and heavily degraded grassland were becoming denuded landscape,
indicating that the biodiversity and forage quality in this region are declining. These results showed
that the grassland conditions were becoming worse over the past two decades, which is very important
for grassland management, pasture grazing, and grassland ecosystem studies in this region.

Fig. 2 Difference between percentages of fresh grass yield levels (I = > 12 000 kg ha−1 , II = 9 000–12 000 kg ha−1 , III
= 6 000–9 000 kg ha−1 , IV = 4 500–6 000 kg ha−1 , V = 3 000–4 500 kg ha−1 , VI = 1 500–3 000 kg ha−1 , VII = 750–1 500 kg
ha−1 , VIII = < 750 kg ha−1 ).

Fig. 3 Average fresh grass yields of Golog Prefecture and its six counties in 1981 and 2005.

The average annual precipitation of Golog Prefecture was 544.2 mm in 1981 and was 632.5 mm in
2005 (Qinghai Meteorological Bureau). Pearson product moment correlation analysis showed there were
no significant relationships in yearly precipitation (P = 0.68) from 1980 to 2005. Many studies have
been carried out about vegetation cover and terrestrial net primary productivity in relation to climatic
factors in our region. Results indicate that generally there is a significant correlation between vegetation
cover change and temperature while there is no significant correlation between vegetation cover change
and precipitation in the same period (Piao et al., 2004). In alpine meadow areas, precipitation was not
correlated with NDVI, presumably a consequence of vegetation activity limited by low temperature. The
net primary production was also insensitive to the precipitation. Therefore, temperature is considered
348 L. YU et al.

as the leading factor driving vegetation cover change in our study areas (Zhou et al., 2006).

Livestock carrying capacity

In practice, the evaluation of grazing capacity for the whole year in the pastoral areas of Golog is
based on the usable yield in the whole seasons because it was difficult to distinguish between the cold
(winter and spring) and warm seasons (summer and autumn) grazing lands. The prevailing nomadic
activity, comparatively flat terrain, and grazing habits and characteristics of local Tibetan sheeps and
yaks contribute to the uniqueness of the Tibetan grazing systems. Besides the domestic livestock, there
are many kinds of wild lives, such as wild Tibetan yak (Poephagus mutus), wild ass (Equus kiang),
plateau pika (Ochotona curzoniae), Himalayas marmot (Marmota himalayana robusta), gazelle (Pro-
capra picticaudata), and caterpillar (Gynaephora qinghaiensis) living on the forage in the study areas.
Because of their numbers, habitat, food and daily intake are difficult to determine, the wild life carrying
capacity was not calculated, so the above calculated grazing capacity is somewhat high than the proper
livestock carrying capacity norm. Only Maqên and Baima counties have mixed farming systems, the
main crops are highland barley, pea, spring wheat, potato, and some oil plants, and total forage supplies
is less than 2.7%, so these areas were not included in estimating the livestock carrying capacity.
Calculated maximum livestock carrying capacity (Ausy ) ranged from ≤ 0.5 SU in steppe meadow to
> 1.0 SU in the typical alpine meadow and alpine shrub meadow, and usable grassland area that can
support an SU in 1981 was 0.61 ha while it was 1.13 ha in 2005. Though the actual stock numbers
reduced by 99.24 × 104 SU, the theoretical grazing capacity reduced by 253 × 104 SU from 1981 to
2005 (Table II). The grass capacity was gradually declined by long-time overgrazing in Baima and Gadê
counties whose overgrazing rate reached 88.18% and 49.17%, respectively, in 2005 while in the Madoi
County, the livestock surplus rate increased from 44.26% to 56.13% due to high reducing rate of actual
livestock numbers, indicating that grazing balance was a very important indicator for grassland health.
Because the alpine grassland ecosystem is fragile, it is hard to recover in a short time if it begins to
degrade seriously, so we should adjust the maximum theoretical grazing capacity to a more proper level
to sustain the fragile ecosystem.

TABLE II

Actual stock numbers (AS) and theoretical grazing capacity (TS) of Golog Prefecture in 1981 and 2005

Site 1981 Maximum theoretical value in 2005 Adjusted sustainable value in 2005

AS TS DSa) Rb) AS TS DS R AS TS DS R
× 104 SUc) % × 104 SU % × 104 SU %
Maqên 138.28 142.35 −4.07 −2.86 105.73 103.62 2.11 +2.04 105.73 57.74 47.99 +83.11
Baima 94.05 90.32 +3.73 +4.12 120.91 64.25 56.66 +88.18 120.91 30.69 90.22 +293.99
Gadê 105.54 84.85 +20.69 +24.38 102.10 68.44 33.66 +49.17 102.10 37.42 64.68 +172.85
Darlag 114.23 198.18 −83.95 −42.36 92.58 117.82 −25.24 −21.42 94.32 74.11 20.21 +27.27
Jigzhi 123.76 129.45 −5.69 −4.40 94.32 85.29 9.03 +10.59 92.58 47.06 45.52 +96.71
Madoi 87.50 156.99 −69.49 −44.26 48.13 109.72 −61.59 −56.13 48.13 79.85 −31.72 −39.72
Golog 663.01 802.14 −139.13 −17.34 563.77 549.14 14.63 +2.66 563.76 326.87 236.89 +72.47
a) Differencein stock number, as calculated by DS = AS – TS.
b) Ratioof DS to AS: “+” indicates overgrazed and “−” indicates surplus.
c) SU = sheep unit.

The reduction factors for maximum theoretical carrying capacity in Table I showed that in Golog
Prefecture, the slope was moderate, water was densely covered, and serious and extremely serious soil
erosion areas were small. The reduction rates of maximum grazing capacity adjusted by these factors
were 33.50%, 29.12%, and 0.76%, respectively, indicating that slope and water were the most important
factors affecting the livestock grazing. From the adjusted sustainable carrying capacity calculated for all
six counties in Golog Prefecture (Table II), it could be seen that most of the areas were suitable for graz-
GRASSLAND ESTIMATION USING REMOTE SENSING AND GIS 349

ing and most counties had overgrazing problems, and the overgrazing rate ranged from 27.27% in Darlag
to 293.99% in Baima, with an average of 72.47%. Taking the adjustment factors into consideration,
this result indicated that the actual carrying capacity greatly exceeded the proper theoretical carrying
capacity. However, the grazing land area was unevenly distributed in cold seasons and warm seasons. In
the cold seasons, the grazing land area was smaller than that in the warm seasons, and furthermore, the
theoretical carrying capacities differed considerably in cold and warm grasslands (Wang et al., 2001). In
addition, grazing period was different between the cold grassland and the warm grassland, the former
being 7 months and the latter being 5 months. Hence, overgrazing is common in the winter grasslands
(Zhou et al., 2003, 2005). The seasonally unbalanced distribution and irrational use of the grasslands,
i.e., heavy overgrazing in cold grasslands but lying idle in some warm grasslands, had caused serious
degradation of the alpine meadow vegetation.
Field survey showed that aboveground plant biomass, the percentages of graminoids, sedges, and
dominant shrub decreased with increasing grazing intensity, and the percentage of forbs increased with
the grazing density (Wei et al., 2005; Zhou et al., 2003). Some of the unpalatable plant species are
grazed by the livestock, which would result in declining of body health of livestock. At this time, rodent
pests damage vegetation by excessive excavating and gnawing and ejecting soil to the ground, while
wind and water erosion appear simultaneously (Wei et al., 2007). Grasslands are covered gradually by
shifting sands or sandy materials, and their grazing values are considerably cut down. Heavily degraded
grasslands are not suitable for grazing and are unable to sustain the pastoralist economy. In summary,
when degradation changes from light to heavy, the aboveground biomass, plant coverage, proportion
of palatable herbage, and soil hardness as well as soil moisture and soil organic matter content will
decrease (Wei et al., 2005).
Long-term overgrazing in this region will result in the eventual overuse of the vegetation. When the
area can no longer support the disturbance of the animal grazing, it crashes, leading to the inevitable,
massive die-off of large numbers of livestock and wild animals. The remainder starts to recover slowly
at first on account of the poor vegetative cover and low available plant production resulting in the
extremely low carrying capacity. Once the vegetation has recovered to such an extent that it attains its
previous carrying capacity, animal numbers start building up again. The whole cycle of animal number
build-up and the consequent overgrazing resumes.
The government officials and ecologists in Qinghai Province and China have realized the on-going
degradation of grassland ecosystem in the source region of the Yellow River. Most of the studies showed
that the degradation is mainly caused by human activities, especially overgrazing (Liu et al., 2006). Our
results showed that the health of alpine grassland ecosystem was declined and overgrazing was still se-
rious in Golog Prefecture. So, it would be important and urgent to perform some effective management
of grassland ecosystem and livestock. Fortunately, some effective management practices, e.g., rodent
and ruderal weed control, reseeding, establishment of artificial grassland, rational management of grass-
land, optimizing livestock structure, ecological emigration policy, fencing natural grassland, warm pen
construction, establishment of perennial pastures and use of solar energy as well as some integrated
countermeasures for the restoration of degraded grasslands, have been developed in this region (Zhou et
al., 2005). Harnessing of ecological principles and biological processes, the grazing pattern, grazing den-
sity, sheep and yaks feedlot fattening, sustainable livestock production system are approached according
to serial pilot studies in the region (Zhao and Zhou, 2005).

Advantages and limitations of the modified model

Our method provided both the quality of grass and the quantity of livestock rapidly, which can
help improve decision making by the government managers, optimize their management practices, and
increase grassland productivity and profitability. They may also be used to understand what variables
may be controlling actual stocking rate; infer the “average” stocking rate that would be expected
for a given site according to an empirical regional pattern, which might be a better tool than the
350 L. YU et al.

costly trial-and-error process just described; indicate potential cases of overgrazing or underutilization
of particular habitats; provide an indicator of production capacity that may guide policy makers to make
decisions about the spatial distribution of different policies; and stimulate further research on potential
relationships at finer scales, which would have strong impacts on management decisions at that scale
(Oesterheld et al., 1998).
However, there existed several limitations in our study. For instance, forest was classified as alpine
shrubs because of their similar reflectance in MODIS images, which would result in overestimate of
forage biomass and therefore the corresponding theoretical livestock carrying capacity. The edible
forage coefficients were uniform, which would underestimate aboveground edible biomass of high-quality
palatable forage biomass and overestimate that of high unpalatable forage biomass. The reduction
factors for adjustment of the theoretical capacity were still empirical.

CONCLUSIONS

Using remote sensing and GIS technologies, grass yield and proper livestock carrying capacity were
estimated in alpine grassland in the source region of the Yellow River. The average fresh mass of
grassland decreased from 3 605 to 2 954 kg ha−1 in this region, and the areas of productive grassland
decreased by 12.54% while the low-productivity grassland increased by 16.68%. The livestock carrying
capacity reduction was closely related with slope, distance to water, and vegetation coverage, and most
of the areas were suitable for grazing. Most counties in the study area had serious overgrazing problems,
with overgrazing rate ranging from 22.27% in Darlag to 293.99% in Baima with an average of 72.47%.

ACKNOWLEDGEMENTS

The authors would like to thank Mr. CAI La-Jia, GUO Yuan-Hu, and XU Wei in the Grassland
Monitoring Station, Golog Prefecture, Qinghai Province and Dr. LIU Han-Wu in the Northwest Institute
of Plateau Biology, Chinese Academy of Sciences, for their help with field work, and the two anonymous
reviewers for their valuable comments and suggestions. Thanks also to the Data Center for Resources
and Environmental Sciences, Chinese Academy of Sciences for providing the geospatial data.

REFERENCES
Cheng, Q. 2006. Multisensor comparisons for validation of MODIS vegetation indices. Pedosphere. 16(3): 362–370.
Dong, Y. P., Wu, H. X. and Rong, Y. P. 2005. Remote Sensing Technology for Grassland Resource Monitoring (in
Chinese). Chemical Industry Press, Beijing.
Friedl, M. A., McIver, D. K., Hodges, J. C. F., Zhang, X. Y., Muchoney, D., Strahler, A. H., Woodcock, C. E., Gopal,
S., Schneider, A., Cooper, A., Baccini, A., Gao, F. and Schaaf, C. 2002. Global land cover mapping from MODIS:
algorithms and early results. Remote Sens. Environ. 83(1): 287–302.
Holecheck, J. L. 1988. An approach for setting the stocking rate. Rangelands. 10(1):10–14.
Hunt, E. R., Everitt, J. H., Ritchie, J. C., Moran, M. S., Booth, D. T., Anderson, G. L., Clark, P. E. and Seyfried, M.
S. 2003. Applications and research using remote sensing for rangeland management. Photogramm. Eng. Rem. S.
69(6): 675–693.
Hunt, E. R. and Miyake, B. A. 2006. Comparison of stocking rates from remote sensing and geospatial data. Rangeland
Ecol. Manage. 59(1): 11–18.
Justice, C. O., Townshend, J. R. G., Vermote, E. F., Masuoka, E., Wolfe, R. E., Saleous, N., Roy, D. P. and Morisette, J.
T. 2002. An overview of MODIS Land data processing and product status. Remote Sens. Environ. 83(1): 3–15.
Langner, A., Miettinen, J. and Siegert, F. 2007. Land cover change 2002–2005 in Borneo and the role of fire derived from
MODIS imagery. Glob. Change Biol. 13(11): 2329–2340.
Li, J. L., Liang, T. G. and Chen, Q. G. 1998. Estimating grassland yields using remote sensing and GIS technologies in
China. New Zeal. J. Agr. Res. 41(1):31–38.
Li, Y. N., Zhou, H. K. and Shen, Z. X. 2001. The association analysis of herbage yield and meteorological factors in
Alpine Meadow. Acta Agrestia Sin. (in Chinese). 9(3): 232–238.
Liu, L. S., Zhang, Y. L., Bai, W. Q., Yan, J. Z., Ding, M. J., Shen, Z. X., Li, S. C. and Zheng, D. 2006. Characteristics of
grassland degradation and driving forces in the source region of the Yellow River from 1985 to 2000. J. Geogr. Sci.
16(2): 131–142.
GRASSLAND ESTIMATION USING REMOTE SENSING AND GIS 351

Oesterheld, M., DiBella, C. M. and Kerdiles, H. 1988. Relation between NOAA-AVHRR satellite data and stocking rate
of rangelands. Ecol. Appl. 8(1): 207–212.
Piao, S. L., Fang, J. Y., Ji, W., Guo, Q. H., Ke, J. H. and Tao, S. 2004. Variation in a satellite-based vegetation index in
relation to climate in China. J. Veg. Sci. 15(2): 219–226.
Qian, S., Mao, L. Y., Hou, Y. Y., Fu, Y., Zhang, H. Z. and Du, J. 2007. Livestock carrying capacity and balance between
carrying capacity of grassland with added forage and actual livestock in the Qinghai-Tibet Plateau. J. Nat. Resour.
(in Chinese). 22(3): 389–397.
Reeves, M. C., Zhao, M. and Running, S. W. 2006. Applying improved estimates of MODIS productivity to characterize
grassland vegetation dynamics. Rangeland Ecol. Manage. 59(1): 1–10.
Tang, J. M. and Zhang, S. W. 2002. Application research of MODIS data in monitoring land use change. Remote Sens.
Technol. Appl. (in Chinese). 17(2): 104–107.
Tang, Q. Y. and Feng, M. G. 2006. Data Processing System: Experimental Design, Statistical Analysis, and Data Mining
(in Chinese). Science Press, Beijing.
Wang, G., Qian, J., Cheng, G. and Lai, Y. 2001. Eco-environmental degradation and causal analysis in the source region
of the Yellow River. Environ. Geol. 40(7): 884–890.
Wang, Q. and Tenhunen, J. D. 2004. Vegetation mapping with multitemporal NDVI in North Eastern China Transect
(NECT). Int. J. Appl. Earth Observ. Geoinf. 6(1): 17–31.
Wei, X. H., Li, S., Yang, P. and Chen, H. S. 2007. Soil erosion and vegetation succession in alpine Kobresia steppe meadow
caused by Plateau Pika—A case study of Nagqu County, Tibet. Chinese Geogr. Sci. 17(1):75–81.
Wei, X. H., Yang, P., Li, S. and Chen, S. H. 2005. Effects of over-grazing on vegetation degradation of the Kobresia
pygmaea meadowand determination of degenerative index in the Naqu Prefecture of Tibet. Acta Pratacult. Sin. (in
Chinese). 14(3): 41–49.
Xu, B., Yang, X. C., Tao, W. G., Qin, Z. H. and Liu, H. Q. 2007. Remote sensing monitoring upon the grass production
in China. Acta Ecol. Sin. 27(2): 405–413.
Yang, Y. L., Qiu, X. F. and Yin, Q. J. 2007. Study on Monitoring System of Qinghai Grassland Output Based on MODIS
EVI Data. Meteorol. Mon. (in Chinese). 33(6): 102–106.
Yang, Z. L. and Yang, G. H. 2000. Potential productivity and livestock carrying capacity of high-frigid grassland in China.
Resour. Sci. (in Chinese). 22(4): 72–77.
Zhao, X. Q. and Zhou, H. K. 2005. Eco-environmental degradation, vegetation regeneration and sustainable development
in the Headwaters of Three Rivers on Tibetan Plateau. Bull. Chinese Acad. Sci. (in Chinese). 20(6): 471–476.
Zhou, H. K., Zhao, X. Q., Tang, Y. H., Gu, S. and Zhou, L. 2005. Alpine grassland degradation and its control in the
source region of the Yangtze and Yellow Rivers, China. Grassl. Sci. 51(3): 191–203.
Zhou, H. K., Zhou, L., Zhao, X. Q., Liu, W., Li, Y. N., Gu, S. and Zhou, X. M. 2006. Stability of alpine meadow ecosystem
on the Qinghai-Tibetan Plateau. Chinese Sci. Bull. 51(3): 320–327.
Zhou, H. K., Zhou, L., Zhao, X. Q., Liu, W., Yan, Z. L. and Shi. Y. 2003. Degradation process and integrated treatment
of “black soil beach” type degraded grassland in the source regions of Yangtze and Yellow Rivers. Chinese J. Ecol.
(in Chinese). 22(5): 51–55.

Das könnte Ihnen auch gefallen