IMPACT OF ELEVATION AND ASPECT ON THE SPATIAL DISTRIBUTION OF

VEGETATION IN THE QILIAN MOUNTAIN AREA WITH REMOTE SENSING DATA

X.M. Jin a, c, Y.-K. Zhang b, M.E. Schaepman c, J.G.P.W. Clevers c, Z. Su d

a

School of Water Resources and Environment, China University of Geosciences, Beijing, 100083, China

-abqf6z@r.postjobfree.com

b

Department of Geoscience, University of Iowa, Iowa City 52242, IA,USA

-abqf6z@r.postjobfree.com

c

Wageningen University, Centre for Geo-Information, Wageningen, The Netherlands

- (Jan.Clevers,Michael.Schaepman)@wur.nl

d

International Institute for Geo-Information Science and Earth Observation (ITC), Enschede, The

abqf6z@r.postjobfree.com

Commission VII, ICWG-VII-IV

ABSTRACT:

The spatial distribution of vegetation in the Qilian Mountain area was quantified with remote sensing data. The MODIS NDVI values

for June, July, August and September are the best indicators for the vegetation growth during a year in this area and thus were used in

this study. The results obtained by analyzing the NDVI data for seven years from 2000 to 2006 clearly indicated that elevation is the

dominating factor determining the vertical distribution of vegetation in the area: the vegetation growth is at its best between the

elevations of 3200 m and 3600 m with the NDVI values lager than 0.5 and a peak value of larger than 0.56 at 3400 m. The horizontal

distribution of vegetation within the zone of 3200 m and 3600 m is significantly impacted by the aspect of hillslopes: the largest

NDVI value or the best vegetation growth is found in the shady slope whose aspect is between NW340 to NE70 due to relatively

less evapotranspiration. The methodology developed in this study should be useful for similar ecological studies related to vegetation

distribution.

been used in large-scale global assessments of vegetation

1. 1 INTRODUCTION

distribution and land cover with the Normalized Difference

The vegetation cover in mountain areas is very important. Vegetation Index (NDVI) data from Advanced Very High

Vegetation cover affects local and regional climate and reduce Resolution Radiometer (AVHRR) and the Moderate Resolution

erosion. Economy of local communities and millions people in Imaging Spectroradiometer (MODIS) (Chen and Brutsaert 1997;

mountain areas depends on forests and plants. They also Defries and Townshend 1994; Defries et al. 1995; Friedl et al.

effectively protect people against natural hazards such as 2002; Loveland et al. 2000, 1999). The NDVI is an index

rockfall, landslides, debris flows, and floods (Brang et al., derived from reflectance measurements in the red and infrared

2001). Settlements and transportation corridors in alpine portions of the electromagnetic spectrum to describe the

regions mainly depend on the protective effect of the vegetation relative amount of green biomass from one area to the next

(Agliardi and Crosta, 2003). Therefore, understanding of (Deering 1978). The NDVI is an indicator of photosynthetic

distribution and patterns of vegetation growth along with the activity of plants and has been widely used for assessing

affecting factors in those areas are important and have been vegetation phenology and estimating landscape patterns of

studied by many researchers (Oliver and Webster 1986; Weiser primary productivity (Sellers, 1985; Tucker and Sellers, 1986).

et al. 1986; Stephenson 1990; Turner et al. 1992; Henebry 1993; It was designed to quantitatively evaluate vegetation growth:

Endress and Chinea 2001; Bai et al. 2004). higher NDVI values imply more vegetation coverage, lower

NDVI values imply less or non-vegetated coverage, and zero

Topography is the principal controlling factor in vegetation NDVI indicates rock or bare land.

growth and that the type of soils and the amount of rainfalls

play secondary roles at the scale of hillslopes (O Longhlin 1981; Most studies with remote sensing data were concentrated on

Wood et al. 1988; Dawes and Short 1994). Elevation, aspect, two-dimensional horizontal patterns and a few were focused on

and slope are the three main topographic factors that control the the effect of elevation on the vertical distribution of vegetation

distribution and patterns of vegetation in mountain areas in mountain areas (Franklin 1995; Edwards 1996; Guisan and

(Titshall et al. 2000). Among these three factors, elevation is Zimmermann 2000; Hansen 2000; Miller et al. 2004). The

most important (Day and Monk 1974; Busing et al. 1992). objectives of this study are two-fold: 1) to quantitatively assess

Elevation along with aspect and slope in many respects both vertical and horizontal distribution of vegetation in the

determines the microclimate and thus large-scale spatial Qilian Mountain area and its main controlling factors, i.e.,

distribution and patterns of vegetation (Geiger 1966; Day and elevation, aspect, and 2) to demonstrate the usefulness of the

Monk 1974; Johnson 1981; Marks and Harcombe 1981; Allen methodology which may be used for other environmental and

and Peet 1990; Busing et al. 1992). ecological studies.

One of the powerful tools to study the spatial distribution of

vegetation is remote sensing. Remote sensing has traditionally

1385

The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008

the NDVI values of these four months can best reflect the

2. STUDY AREA

pattern of the vegetation cover in the region.

Located in the upstream of the Heihe River basin, the Qilian

Mountain area has a steep topography with an elevation range The Digital Elevation Model (DEM) data was downloaded

from 1680 m to 5100 m (Figure 1). The intermountain basin from the Digital River Basin website (http://heihe.westgis.ac.cn)

and longitudinal valley are widely developed in the area. The and its spatial resolution is 100 m. The MODIS NDVI was

northern part of the Qilian Mountain surrounded by tributaries resampled and interpolated to have the same spatial resolution

of Heihe River to the east and west was selected to be the study as the DEM data in this study.

area (the area outlined with the bold black line in Figure 1)

because this area represents a typical mountain range and best

reflects the vegetation change with elevation. The total study 4. RESULTS AND DISCUSSION

area is 2968 km2. The climate in this area is characterized by

typical high plateau continental climate. The average annual It is well known that spatial distribution of vegetation cover is

temperature is 0.6 and the amount of precipitation increases usually affected by elevation and aspect. Most vegetation in the

with the elevation. Due to complex topography, the climate is northern part of Qilian Mountain area is distributed between the

diverse and has distinct vertical characteristics. These vertical elevations of 1800 m and 4500 m. To the best of our knowledge,

climate characteristics have important impacts on the soil however, the obvious spatial distribution and patterns have not

development and vegetation growth in the areas as they do in been studied quantitatively. We show in this study that the

many other mountains. readily available NDVI data can be used to quantify the spatial

distribution of vegetation. The range of elevations from 1800 m

The vegetation distribution in this area exhibits an obvious and 4500 m was divided into a total of 270 intervals with 10m

vertical gradient due to the climatic changes with elevation. The in each intervals. The aspect angle of 360 were divided into a

vegetation types from the low altitude to high altitude are: total of 72 intervals with 5 in each intervals. These divisions

desert-grassland vegetation (1800 2100 m), dry result a total of 19360 cells among which 19060 cells with the

shrub-grassland vegetation (2100 2400 m), mountain NDVI values larger than zero. In each cell the NDVI values

forest-grassland vegetation (2400 3400 m), sub-alpine from year 2000 to 2006 were averaged. The mean values

shrub-grassland vegetation (3400 3900 m), and cold-desert represent the general conditions of vegetation growth in

alpine meadow vegetation (>3900 m). The mountain different elevations and aspects. A contour map of the mean

forest-grassland vegetation is the main vegetation type and the NDVI values with elevation and aspect in the northern part of

main component of the Qilian Mountains ecosystem. The range Qilian Mountain was plotted in Figure 2. A Gaussian smooth

of elevations (1800 5100 m) in study area was divided into a filter was used and a low pass convolution was performed on

total of 31 intervals with 100m in each of the intervals and the the gridded data to obtain the more consistent and smooth map

aspect angle was divided into a total of 72 intervals with 5 in in Figure 2.

each of the interval.

Several observations can be made in Figure 2 regarding the

The vegetation in the Qilian Mountain area plays an important effects of elevation and aspect on the vegetation growth in the

role in the local water cycle by affecting hydrological processes, mountain area. First of all, it is clearly seen that the elevation is

e.g., evapotranspiration and runoff, and is an important the main controlling factor in the vegetation growth. The NDVI

ecological storage for water resources. Qilian Mountain value increases with the elevation and reaches its maximum

supplies water for Hexi Corridor which is the most important value around 3400 m and then decreases as the elevation

agricultural region and settlement in northwest China. The increases beyond 3400 m. The NDVI value is mostly larger

vegetation in the Qilian Mountain area significantly affects the than 0.50 (the dark green region in Figure 2) when the elevation

oasis system in the region and protects the middle and is between 3200 m and 3600 m which is the best vertical zone

downstream area of Heihe River against desertification. in terms of vegetation growth. The NDVI values are less than

0.50 when the elevation is lower than 3200 m and higher than

3600 m or the vegetation growth is poorer in these elevations

3. DATASET that in the zone of 3200 m and 3600 m.

The MODIS NDVI data, the vegetation index maps depicting Secondly, the vegetation growth in the Qilian Mountain area is

spatial and temporal variations in vegetation activities, is significantly affected by aspect. The impact of aspect on the

derived by precisely monitoring the Earth s vegetation. These vegetation growth is most significant in the vertical zone of

vegetation index maps have been corrected for molecular 3200 m and 3600 m. The best vegetation in this zone is

scattering, ozone absorption, and aerosols. The MODIS NDVI distributed between NW340 and NE70 (the darkest green area

data is based on 16-day composites and its spatial resolution is in Figure 2 with the NDVI value larger than 0.56). In other

250 m. Currently, the MODIS NDVI products have been used words, the best vegetation growth is on the shady side of the

throughout a wide range of disciplines, such as inter- and mountain where much less evapotranspiration (ET) is expected.

intra-annual global vegetation monitoring climate and The reduced ET on the shaded side is important for the

hydrologic modeling, and agricultural activities and drought vegetation growth in the Qilian Mountain area since it is

studies (Zhan et al. 2000; Jin and Sader 2005; Sakamoto et al. located in a semi-arid region. It is also observed in Figure 2 that

2005; Knight et al. 2006; Lunetta et al. 2006). In this study the a better vegetation growth occurs over a larger elevation range

NDVI values from 28 MODIS NDVI images of the 16-day on the side facing north and northeast. At the aspect of N0, for

composites of June, July, August and September in seven years example, the NDVI value of 0.50 or larger are observed over

from 2000 to 2006 were used to study the spatial distribution of the vertical zone of 600 m between the elevation range of 3100

vegetation in the northern part of the Qilian mountain area m 3700 m while at the aspect of S180 the same NDVI values

because June, July, August and September are the most are observed in a smaller range 400 m between 3200 m and

productive months of vegetation growth during a year and thus 3600 m. The much wider vertical zone with better vegetation

1386

The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008

growth on the shady side of Qilian Mountain may significantly REFERENCES

affect the local water cycle and climate.

Agliardi, F., Crosta, G. B., 2003, High resolution

three-dimensional numerical modelling of rockfalls.

Third observation made in Figure 2 is the rate of change in the

International Journal of Rock Mechanics and Mining Sciences,

NDVI values with elevation. This rate varies more gently at

40, 455-471.

lower elevations from 2000 m to 3400 m and it changes more

quickly when elevation is higher than 3400 m, implying that the

vegetation growth is more sensitive in high altitude area. On the

Allen, R B., Peet, R. K., 1990, Gradient analysis of forests of

average, for example, it takes about 300 m (roughly from 2600

the Sangre de Cristo Range, Colorado. Canadian Journal of

m to 2900 m) for the NDVI value to change from 0.3 to 0.4 at

Botany, 68, 193-201.

the lower altitude zone and only about 200 m at the higher

altitude zone.

Bai, Y., Broersma, K., Thompson, D., and Ross, T. J., 2004,

Landscape-level dynamics of grassland forest transitions in

The NDVI values corresponding to the same elevation were

British Columbia. Journal of Range Management, 55, 66-75.

averaged in order to clearly show the relationship between the

vegetation growth and elevation. A total of 221142 pairs of

Brang, P., Sch nenberger,W., Ott, E., Gardner, R.H., 2001,

NDVI and elevation were obtained based on the 28 MODIS

Forests as protection from natural hazards. In: Evans, J. (Ed.),

NDVI images of the 16-day composites of June, July, August

The Forests Handbook. Blackwell Science Ltd., Oxford, pp.

and September in seven years from 2000 to 2006. The

53-81.

relationship between the averaged NDVI and elevation for the

study area is clearly shown in Figure 3: the averaged NDVI

Busing, R. T., White, P. S., and MacKende, M. D., 1992,

increases with elevation and reaches its maximum value of

Gradient analysis of old spruce-fir forest of the Great Smokey

about 0.56 at 3400 m and then decreases as the elevation

Mountains circa 1935. Canadian Journal of Botany, 71,

increases beyond 3400 m, an clear indication that the vegetation

951-958.

growth is at its best at the elevation of 3400 m.

The effect of aspect on the vegetation growth is more clearly Chen, D., Brutsaert, W., 1998, Satellite-Sensed distribution and

demonstrated in Figure 4 where the change of the NDVI values spatial patterns of vegetation parameters over a Tallgrass

with aspect between the elevations of 3200 m and 3600 m was Prairie. Journal of the Atmospheric Sciences, 55, 1225-1238.

plotted. It is seen in Figure 4 that the NDVI value is larger than

0.55 or the vegetation growth is best in the aspect range of

NW340 to NE70 and the NDVI value is less than 0.54 or the Dawes, W. R., and Short, D., 1994, The significance of

vegetation is worse between E90 to W270 . As we discussed topology for modelling the Surface hydrology of fluvial

above, this shows that the aspect of the mountain slopes landscapes. Water Resources Research, 30, 1045- 1055.

significantly affects the vegetation growth in the study area. In

general, the vegetation coverage on the sunny side in the Day, F. P., and Monk, C. D., 1974, Vegetation patterns on a

semi-arid Qilian mountain area is less developed than that on Southern Appalachian watershed. Ecology, 55, 1064-1074.

the shady side because of more evapotranspiration in the sunny

side than in the shady side due to the differences in their solar Deering, D.W., 1978, Rangeland reflectance characteristics

radiation and higher land surface temperature. measured by aircraft and spacecraft sensor. ph.D. Dissertation,

Texas A&M University, College Station, TX, 338pp.

5. CONCLUSIONS DeFries, R., Hansen, M., & Townshend, J., 1995, Global

discrimination of land cover types from metrics derived from

The spatial distribution of vegetation in the Qilian Mountain AVHRR pathfinder data. Remote Sensing of Environment, 54,

area was quantified with remote sensing data. The MODIS 209-222.

NDVI values for June, July, August and September are the best

indicators for the vegetation growth during a year in this area Defries, R. S., Townshend, J. R. G., 1994, NDVI-derived

and thus were used in this study. Based on the results obtained land-cover classifications at a global-scale. International

by analyzing the NDVI data for seven years from 2000 to 2006, Journal of Remote Sensing, 15, 3567- 3586.

the following important conclusions can be drawn.

Endress, B. A., and Chinea, J. D., 2001, Landscape patterns of

1) Elevation is the dominating factor determining the vertical tropical forest recovery in the Republic of Palau. Biotropica, 33,

distribution of vegetation in the Qilian Mountain area: the 555 565.

vegetation growth is at its best between the elevations of 3200

m and 3600 m with the NDVI values larger than 0.50 and a Edwards, G. R., Parsons, A. J., Newman, J. A., Wright, I. A.,

peak value of larger than 0.56 around 3400 m. 1996, The spatial pattern of vegetation in cut and grazed

grass/white clover pastures. Grass and Forage Science, 51,

2) The horizontal distribution of vegetation within the elevation 219 231.

range of 3200 m and 3600 m is significantly impacted by the

aspect of hillslopes: the best vegetation growth is found in the Franklin, J., 1995, Predictive vegetation mapping: geographic

shady slope between NW340 to NE70 with the largest NDVI modeling of biospatial patterns in relation to environmental

value (>0.56) due to relatively less evapotranspiration. gradients. Progress in Physical Geography, 19, 474-499.

3) Better vegetation growth occurs over a larger elevation range Friedl, M. A., McIver, D. K., Hodges, J. C. F., Zhang, X. Y.,

on the shady than sunny side because of less ET in the former Muchoney, D., Strahler, A. H., et al., 2002, Global land cover

than in the latter.

1387

The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008

mapping from MODIS: Algorithms and early results. Remote O Loughlin, E. M., 1981, Saturation regions in catchments and

Sensing of Environment, 83, 287-302. their relations to soil and topographic properties. Journal of

Hydrology, 53, 229-246.

Geiger, R., 1966, The climate near the ground. Harvard

University Press, Cambridge, Mass. Purves, D. W., Law, R., 2002, Fine-scale spatial structure in a

grassland community: quantifying the plant's-eye view. Journal

Guisan, A., Zimmermann, N., 2000, On the use of static of Ecology, 90, 121 129

distribution models in Ecology. Ecological Modelling, 135,

147-186. Sakamoto, T., Yokozawa, M., Toritani, H., Shibayama, M.,

Ishitsuka, N., and Oho, H., 2005, A crop phenology detection

Hansen, A. J., Rotella, J. J., Kraska, M. P. V., Brown, D., 2000, method using time-series MODIS data. Remote Sensing of

Spatial patterns of primary productivity in the Greater Environment, 96, 366-374.

Yellowstone Ecosystem. Landscape Ecolog, 15, 505 522.

Sellers, P.J., 1985, Canopy reflectance, photosynthesis, and

Henebry, G. M., 1993, Detecting change in grasslands using transpiration. International Journal of Remote Sensing, 6,

measures of spatial dependence with Landsat TM data. Remote 1335-1371.

Sensing of Environment, 46, 223 234.

Stephenson, N.L., 1990, Climatic control of vegetation

Jin, S., Sader, S. A., 2005, MODIS time-series imagery for distribution: The role of the water balance. American Naturalist,

forest disturbance and quantification of patch effects. Remote 135, 649 670.

Sensing of Environment, 99, 462-470.

Titshall, L.W., O Connor, T.G., and Morris, C.D., 2000, Effect

Johnson, E. A., 1981, Vegetation organization and dynamics of of long-term exclusion of fire and herbivory on the soils and

lichen woodland communities in the Northwest Temtories. vegetation of sour grassland. African Journal of Range and

Canada Ecology, 62, 200-215. Forage Science, 17, 70 80.

Knight, J. K., Lunetta, R. L., Ediriwickrema, J., and Khorram, Tucker, C.J., and Sellers, P.J., 1986, Satellite remote sensing of

S., 2006, Regional Scale Land-Cover Characterization using primary production. International Journal of Remote Sensing,

MODIS-NDVI 250m Multi-Temporal Imagery: A Phenology 22, 3827-3844.

Based Approach. GIScience and Remote Sensing (Special Issue

on Multi-Temporal Imagery Analysis), 43, 1-23. Turner, C. L., Seastedt, T. R., Dyer, M. I., Kittel, T. G. F., and

Schimel, D. S., 1992, Effects of management and topography

Loveland, T. R., Reed, B. C., Brown, J. F., Ohlen, D. O., Zhu, on the radiometric response of a tallgrass prairie. Journal of

Z., Yang, L., et al., 2000, Development of a global land cover Geophysical Research, 97(D17), 18 855 18 866.

characteristics database and IGBP DISCover from 1 km

AVHRR data. International Journal of Remote Sensing, 21, Walthall, C. L., and Middleton, E. M., 1992, Assessing spatial

1303-1330. and seasonal variations in grasslands with spectral reflectances

from a helicopter plat form. Journal of Geophyical Research,

Loveland, T. R., Zhu, Z. L., Ohlen, D. O., Brown, J. F., Reed, B. 97(D17), 18 905 18 912.

C., & Yang, L. M., 1999, An analysis of the IGBP global

land-cover characterization process. Photogrammetric Wang, J., Price, K.P., and Rich, P.M., 2001, Spatial patterns of

Engineering and Remote Sensing, 65, 1021-1032. NDVI in response to precipitation and temperature in the

central Great Plains. International Journal of Remote Sensing,

Lunetta, R. S., Knight, J. F., Ediriwickrema, J., Lyon, J. G., and 22, 3827-3844.

Worthy, L. D., 2006, Land-cover change detection using

multi-temporal MODIS NDVI data. Remote Sensing of

Environment, 105, 142-154. Weiser, R. L., Asrar, G., Miller, G. P., and Kanemasu, E. T.,

1986, Assessing grassland biophysical characteristics from

Marks, P. L. and Harcombe, P. A., 1981, Forest vegetation of spectral measurements. Remote Sensing of Environment, 20,

the Big Thicket, southeast Texas. Ecological Monographs, 51, 141 152.

287-305.

Wood, E. F., Sinpalan, M., Beven K., and Band L., 1988,

Martin, P., 1993, Vegetation responses and feedbacks to Effects of spatial variability and scale with implications to

climate: a review of models and processes. Climate Dynamics, hydrological modelling. Journal of Hydrology, 102, 29-47.

8, 201-210.

Zhan, X., Defries, R., Townshend, J. R. G., Dimiceli, C.,

Miller, J. R., Turner, M. G., Stanley, E. H., Smithwick, E. A. H., Hansen, M., Huang, C., and Sohlberg, R., 2000, The 250m

and Dent, L. C., 2004, Spatial extrapolation: the science of global land cover change product from the Moderate Resolution

predicting ecological patterns and processes. Bioscience, 54, Imaging Spectroradiometer of NASA s Earth Observing

310-320. System. International Journal of Remote Sensing, 21,

1433-1460.

Oliver, M. A., and Webster, R., 1986, Semi-variograms for

modeling the spatial pattern of landform and soil properties.

Earth Surface Processes and Landforms, 11, 491 504.

1388

The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008

FIGURE CAPTIONS

Figure 1 The DEM (digital elevation model) map of the Qianlian Mountain area with the spatial resolution of 100 m. The area

surrounded by watershed of Heihe tributaries in east and west boundary (outlined with bold black line) was selected as the study area.

4400

0.1

4200 0.2

4000 0.56

0.54

3800

0.52

3600 0.54

6 0.50

0. 5

Elevation(m)

3400 0.45

Mean NDVI

0.54 0.40

0.54

3200 0.5

0.35

3000

0.30

2800 0.25

0.20

2600

0.15

2400

0.10

2200 0.02

0.2

2000

0.15 0.15

30-60-90-120-*** 180

180-***-***-*** 300 330 360

Aspect Figure 2 The change of the mean NDVI values with elevation and aspect in the northern part of Qilian Mountain. A Gaussian smooth

filter was used and a low pass convolution was performed on the grid data to present a more consistent and smooth map. Note: a

refiner scale (0.02) was used when the NDVI value is larger than 0.5.

1389

The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008

5000

4000

Elevation(m)

3000

2000

1000

0 0.2 0.4 0.6

Mean NDVI

Figure 3 The change of the NDVI values with elevation in the northern part of Qilian Mountain area.

Aspect

0

315 45

270 90

0.52 0.53 0.54 0.55 0.56 0.57

Mean NDVI

225 135

180

Figure 4 The change of the NDVI value with aspect for the elevation range of 3200 m to 3600 m in northern part of Qilian Mountain

area.

1390

Contact this candidate