Relationship between Landscape Pattern and Environmental Indices

in Ebinur Lake Wetland National Natural Reserve

Fei Zhang

1,2,*

, Yushanjiang Ayinuer

1,2

, Hsiang-te Kung

and Ping Shi

Key Laboratory of Smart City and Environmental Modeling of Higher Education Institute, College of Resources and

Environment Sciences, Xinjiang University, Urumqi, 830046, PR China;

Key Laboratory of Oasis Ecology, Xinjiang University, Urumqi 830046, China;

Department of Earth Sciences, The University of Memphis, Memphis, TN 38152, USA;

School of Foreign Languages, Jining Medical University, Jining, Shandong 272067 Urumqi, 830046, PR China.

Email: zhangfei3s@163.com

Keywords: Land use/cover, remote sensing, CA-Markov model, simulation and forecast, ecological environment

quality index, Ebinur Lake Wetland Nature Reserve

Abstract: The study of spatial-temporal changes of the landscape is important in revealing the mechanisms and laws

of landscape succession, and to search the relationship between human activities and the environment. This

paper presents a research of land use/cover change (LUCC) from 1998 to 2006 and from 2006 to 2014; it

also simulates and forecasts land use/cover types of Ebinur Lake Wetland National Nature Reserve

(ELWNNR) by using the CA-Markov model. The purposes of this study were determined to the

characteristics of landscape patterns in different altitudes. Results showed that: (1) LUCC was “three plus

three minus” in ELWNNR. The bared lakebed and desert obviously expanded; land use/cover dynamic

degree was 9.5% from 1998 to 2006 for bared lakebed; and 34.1% from 2006 to 2014 for desert. The

transform area of land use/cover type would smaller from 2022 to 2030; (2) There were significant

correlations between environment quality indices and the landscape indicators; the composition and spatial

structure of the landscape affected the spatial distribution of environment quality indices; (3) The R

multiple linear stepwise regression model of environmental quality indices and landscape indices in the

study area was 0.7, which suggested significant correlation between these two indices. It proved that it is

feasible to estimate the environmental quality indices of the study area by landscape indices. The quantified

effects of landscape structure variations on this index helped to better understand the effects of land use on

it and they are important for policy making and land use planning.

1 INTRODUCTION

Land use/cover change (LUCC) is one of the most

important consequences of global change; the

impact of LUCC on the watershed’s water cycle has

always been an essential part in the study about

global change and hydrology (Xia et al., 2006).

Establishing a simulating model is an effective and

reliable measure to describe, explain and

reappearance of LUCC in the past, as well as

forecast LUCC in the future to formulate

countermeasures (Han et al., 2011; Veldkamp et al.,

2001). The study of LUCC is of great significance

for understanding the stability of environment and

the sustainable utilization of water and land

resources (Wang et al., 2015; Chen et al., 2015; Tsai

et al., 2015). Reasonable land use/cover-landscape

changes can produce favorable environmental

effects, while excessive reclamation of ecosystems

can lead to a series of environmental problems, such

as desertification, vegetation degradation, reduction

of water area and so on (Brun et al., 2015; Hansen et

al., 2007; Ye et al., 2015). Therefore, the

environmental effects, which are caused by land

use/cover-landscape changes, have become an

important topic in this field. Furthermore, the

application of statistics (Moghadam et al., 2015;

Zhao et al., 2015), GIS and RS technology (Chen et

al., 2015), landscape ecology models (Wang et al.,

132

Zhang, F., Ayinuer, Y., Kung, H-t. and Shi, P.

Relationship between Landscape Pattern and Environmental Indices in Ebinur Lake Wetland National Natural Reserve.

In Proceedings of the International Workshop on Environment and Geoscience (IWEG 2018), pages 132-138

ISBN: 978-989-758-342-1

2009; Zeng., 2014; Paroissien., 2015) provide the

practical and theoretical framework.

The landscape pattern indices are objectively

used to evaluate the environment, because it is easy

to obtain and almost undisturbed by human factors.

Currently, the academic circles are mainly focus on

the changes in land use/cover-landscape patterns and

the effects of natural factors on environment

(Camacho et al., 2015; Vadrevu et al., 2015; Conrad

et al., 2015; Bing et al., 2015; Terra., 2014);

meanwhile there is less research on the correlation

analysis, which is based on landscape pattern and

environment quality. The ELWNNR is the most

representative wetland desert in China. Affected by

geography, climate and human activities, this

ecosystem is fragile; water and soil resources are

scarce, and the unreasonable distribution of

landscape patterns and human interference have

caused a series of negative environmental effect on

ecology (Lu et al., 2014). Consequently, research

based on the regional landscape pattern will be

crucial for monitoring the indicators and

implementing managing strategies for sustainable

development. Therefore, the purpose of the study is

to: i) analyze the current and future changes in land

use/cover in ELWNNR; ii) quantitatively study the

relationship between land use/cover patterns and the

environment to provide theoretical support for

conservation purposes and sustainable development

of agriculture in the region.

2 MATERIAL AND METHODS

2.1 Study Area

The study area has a total surface of about 2670.85

(Figure 1). This region represents the lowest

depression and water-salt collection center in the

western Junggar basin. It is in the hinterland of the

Eurasian continent and integrates wetlands and

deserts, representing one of the key areas in that

basin (Xie et al., 2010). The average annual

temperature of the ELWNNR is 8°C, characterized

by low rainfall, and greater evaporation than

precipitation. With the typical temperate and arid

continental climate, the ELWNNR is a typical

ecologically fragile zone. The lake area is rich in

habitat types and is one of the few concentrated

areas of desert species in the inland desert (Wang et

al., 2015). At present, the environment of the region

has been seriously damaged and the balance of

ecosystems has been seriously affected.

Figure 1: Location map of ELWNNR.

2.2 Data Acquisition and Methods

Landsat 7 TM and Landsat OLI images from the

ELWNNR were used in this study, that were

acquired in September 1998, July 2006 and

September 2014 (data from the U.S. Geological

Survey, URL: http://earthexplorer.usgs.gov/website).

The track number of the satellite images is 146/29.

Images were pre-processed, performing geometric

correction, radiation correction and image

enhancement under the support of ENVI 5.1

software. Then, remote sensing images were clipped

to center in the study area. Supervised classification

and visual interpretation were used to classify each

of the land use/cover types, that were divided into

six categories: water body, forest grassland, dry lake

bed, saline alkali land, desert, and others. Then, that

classification was verified by combining a

topographic map, a DEM and Google Earth. The

land use/cover classification map was developed for

the three years 1998, 2006 and 2014 and the overall

accuracy of the three remote sensing images were

above 89%.

2.3 The Dynamic Simulation of Land

Use/Cover Types

In this study, the classified images from 1998, 2006

and 2014 were selected in this study as input into the

model to calculate matrices of conversion areas and

conversion probabilities. A transition areas matrix

expresses the total area (in cells) expected to be

changed in the next time. Also, a transition

probability matrix determines the probability of a

Relationship between Landscape Pattern and Environmental Indices in Ebinur Lake Wetland National Natural Reserve

133

pixel in a land use class to change into another class

during that time. The prediction of future land use

changes can be calculated based on conditional

probability formula by using the following equation:

and

(1)

Where S(t) is the state of the system at time t; S(t

+ 1) is the state of the system at time (t + 1); Pij is

the matrix of transition probability in a state (Al-

sharif and Pradhan, 2013). The output of CA–

Markov algorithm is the predicted land cover map

for mentioned future year.

2.4 Dynamic Degree of LUCC

The rate of LUCC is an important index to reflect

the intensity of regional land use change. A dynamic

degree indicates the stability of the land use type and

is represented in percentage. Land use dynamic

models are usually divided into both single and

comprehensive degree. The single land use dynamic

can be formulated as follows:

121

100%

Att







(2)

Where k represents the dynamic degree of a

certain land use and % is the unit, At1 and At2

represent the area of various land use types at the

beginning of the and the end of the study; t1and t2

are the initial and final stages of the change.

2.5 The Selection and Calculation of

Landscape Indices in ELWNNR

Landscape indices are widely used in the analysis of

many landscape patterns. It can provide a lot of

information as well as it can reflect the

characteristics of its composition and spatial

configuration (Wu, 2007). In this research, Frag

stats 4.2 were used to compute landscape pattern

indices in the study area, PLAND (Landscape types’

percentage), PD (patch fragmentation), mean patch

size (MPS), Largest Patch Index (LPI), ED (edge

density), area - weighted mean patch highest-

resolution dimension (AWMPFD), COHESION

(plaque connection degree), AI (aggregation degree

of plaques) were selected on the class level to assist

with the exploration of the landscape characteristics

in this study; meanwhile, number of patches (NP),

ED (edge density), Landscape Shape Index, (LSI),

Largest Patch Index (LPI), COHESION (plaque

connection degree), AI (aggregation degree of

plaques) and SHDI (Shannon diversity Index),

CONTAG (spreading Index) were selected on the

landscape level to obtain the differences among the

landscape metrics in the different land use/cover

classifications for the entire ELWNNR. The

concepts, calculation methods and ecological

significance of these indices were summarized by

Wu (2007).

2.6 Environment Quality Indices

The overall status of a regional environmental

quality is comprehensively the total area

environment quality indices Et (Liu et al., 2003); it

is obtained by considering the area ratio of each land

use/cover type and the weighted sum of relative

ecosystem service values; its expression is:





tii

EACTA

(3)

Where Ai is the land use/cover type area of Class

i in the study period T; Ci is the service value of the

relative ecosystem of land use/cover type of Class i;

n is the number of land use/cover types in the study

area; TA is the total area of the region.

3 RESULTS AND ANALYSIS

3.1 Spatio-Temporal Change of Land

Use/Cover Pattern in Ebinur Lake

The most significant change can be observed in the

water area. According to the statistical annual book

from 1998 to 2006, the precipitation has increased

(Figure 2). In 2004, the government implemented

the increase water project of the Ebinur Lake, which

eventually led to the increase of the water body from

1998 to 2006 (Bai, 2010). However, the extension of

the area occupied by water, showed a decreased

from 2006 to 2014. This fact was partly due to the

massive reclamation of land and the rapid increase

of the cultivated land. Meanwhile, Bortala, Jinghe

and Kuitun rivers were used for irrigation and

IWEG 2018 - International Workshop on Environment and Geoscience

134

resulted in the decrease of water in Ebinur Lake and

increase of bared lake bed. Furthermore, the desert is

expanding as a consequence.

Figure 2: Land use/cover of ELWNNR in 1998, 2006 and

2014.

From 1998 to 2006, the area of salinized land,

desert and others has decreased mainly by the land

use/cover passive attitudes (-5.8% and -4.0%,

respectively) (Figure 3). The government carried out

a water increase project of Ebinur Lake in 2004 and

the area of water body increased and the area of dry

lake bed decreased as a consequence. Compared

with 2004, the area of bared lakebed increased and

the area of water body decreased in 2006. However,

compared with 1998, the area of water body and dry

lake bed expanded to a certain extent, because of

artificial water increase project. Therefore, forest,

grassland, bared bed, and water body showed an

expansion. Until 2014, the bared lakebed, water

body, and other land species showed a decreasing

trend. However, forest, grassland, salinized land,

and desert are constantly expanding, in which the

dynamic degree of desert is 34.1%. From 2014 to

2022, the bared lakebed, water body, and other areas

show a trend to decrease in extension. The bared

lakebed had a drastic decrease (-5.2%). The area of

forest, grassland, salinized land, and desert increased

instead; meanwhile the dynamic degree of forest and

grassland was 5.2%. In 2030, these changes were

expected to be limited (Figure 3).

The overall LUCC in the ELWNNR showed to

distinct trends (Figure 3): the increase of forest,

grassland, saline land and desert, and the decrease of

dry lake bed, water body, and other lands. After the

establishment of the ELWNNR, human activities

have been relatively reduced and natural grasslands

have been restored; it is mainly attributed to the

national implementation of projects, such as

returning farmland to forests, returning pasture to

grassland, and recognizing the importance of the

environment. However, the water area of Ebinur

Lake continues to shrink, because of an increase in

cultivated land and a large amount of water

resources used for irrigation; consequently, resulting

the desertification and salinization of bare lake-bed.

Meanwhile, the strong wind in Alashankou has

accelerated the expansion of saline land. Therefore,

it is urgent to manage the saline land in the western

part of ELWNNR.

Figure 3: Dynamic degree of land use/cover types of the

ELWNNR from1998 to 2030.

3.2 The Correlation Analysis between

Landscape Indices and

Environmental Quality Indices

The correlation between landscape indices and

environmental quality indices based on class level

and landscape level were shown in Figure 4 and

Table 1. From the landscape level, the environment

quality indices of forest-grassland are positively

correlated with landscape type percentage, patch

fragmentation, maximum patch indices, boundary

density, area weighted average patch fractal

dimension, and patch connectivity. However, there

was a significant and negative correlation between

environment quality indices of forest-grassland with

the average patch area, indicating that the effect of

forest and grass land on the environment was higher

than other land use/cover types.

Relationship between Landscape Pattern and Environmental Indices in Ebinur Lake Wetland National Natural Reserve

135

0.003

0.909

0.9880.349

0.868 0.259

-0.069

0.67-0.053-0.965

0.754

-0.317

0.241

0.6170.9930.961

0.936

0.627

-0.2290.64

0.7170.8540.514

0.988

0.9540.1130.722

0.7170.1080.765

0.991

0.363

0.962 0.714

0.7110.8880.555

0.94

0.1750.961

-0.619

0.996

0.9930.907

-0.624

0.9990.994

Others

Salinized

land

Bared

lakebed

Water

Desert

Forest and

grassland

COHESION AI

AWMPFDMPS

LPIPD

-1.0

-0.75

-0.50

-0.25

0.0

0.25

0.50

0.75

1.0

PLAND

0.999

Figure 4: Pearson correlation coefficients between

environment quality indices and landscape pattern indices

on class level.

The environmental quality indices of the

salinized land were positively correlated with the

percentage of landscape type, the maximum patch

indices, the average patch area and the patch

connectivity; and it was negatively correlated with

the patch fragmentation. The environmental quality

indices of the bared lakebed have a positive

correlation with the percentage of landscape types,

the maximum patch indices, the area-weighted

average plaque fractal dimension, and the patch

connectivity. The large amount of cultivated land

area rapidly increased due to the human demand for

crops and the use of water from Boltara, Jinghe and

Kuitun Rivers. As a result, the lake surface shrank,

and the area of bared lakebed increased. Therefore,

it should pay attention to the protection of forest

grassland and the prevention and control of salinized

land.

At the landscape level, the environmental quality

indices were negatively correlated with the number

of patches, boundary density, landscape shape index

and Shannon diversity index. There is a significant

and high positive correlation among patch index,

patch connectivity, plaque aggregation and spread

index.

Since the regional environmental quality indices

were influenced by the spatial pattern of landscape

pattern indices, multivariate correlation analysis

(Table 2) was performed based on single factor

analysis. Key landscape indices were pointed among

numerous landscape indices that affect the

environmental quality indices. The R

value of

multiple linear stepwise regression model of

environmental quality indices and landscape indices

(p<0.05) was greater than 0.7, indicating that the

value of landscape indices and the value of

environmental quality indices in the study area were

significantly correlated. Landscape indices can be

then used to estimate the environmental quality

indices of the study area.

Table 1: Pearson correlation coefficients between environment quality indices and landscape pattern indices on class level.

Environment indices

Landscape pattern indices

NP LPI ED LSI CONTAG COHESION SHDI AI

（E）

-0.96 0.78 -0.911 -0.911 0.91 0.667 -0.792 0.909

Table 2: Relationship between landscape indices and environment quality indices.

Environment

Indices

Landscape pattern

indices

Regression equation R

RMSE Sig.

Forest and grassland

PLAND, LPI E

Forest and grassland

=-0.913PLAND+2.146LPI-

4.323

0.888 0.0027 0.032

Salinized land

PLAND, PD E

Salinized land

=2.344PLAND-0.374PD-13.557 0.797 0.0676 0.035

Bared lakebed

LPI, AI E

Bared lakebed

=0.037LPI+0.128AI-1.082 0.76 0.3 0.01

Water

LPI, AI E

Water

=2.02LPI-56.687AI+252.918 0.83 0.053 0.04

Desert

PLAND, ED E

Desert

=2.234PLAND+0.007ED-13.181 0.78 0.08 0.001

Others

COHESION, AI E

Others

=257.226COHESION-2.774AI-

1176.746

0.793 0.0767 0.028

IWEG 2018 - International Workshop on Environment and Geoscience

136

4 DISCUSSION AND

CONCLUSIONS

The CA-Markov model has always been a common

model for predicting LUCC. The CA model has a

strong capability in simulating the spatial-temporal

characteristics of complex systems. That is why it

has been extensively used as a spatially dynamic

model in LULC research (Adhikari and Southworth

2012). This model can be understood as a dynamic

and relatively simple spatial system, in which the

state of each cell of the matrix depends on the

previous state of the cells enclosed inside a defined

neighbourhood, in accordance with a set of

transition rules. Therefore, the CA model is capable

enough to predict the spatial distribution of the

LULC pattern and its dynamics because it adds the

spatial properties of LULC. Because human factors

are the most important reason for LUCC, the

simulated results of CA-Markov model are highly

uncertain. Therefore, we will try to explore new

methods and make the results practical in future

studies. The CA-Markov model effectively

combined the advantages of the Markov model and

the CA model, improving the simulation accuracy.

This research built a CA model spatial filter of 5

pixels×5 pixels, but it did not compare spatial filters

of different sizes. Therefore, future research could

be focused on the effect of spatial resolution.

This study can reflect the LUCC of the

ELWNNR in 1998-2014 and is closely related to the

landscape pattern and environment situation, which

can be used to provide reference for environment

protection in the ELWNNR. Results show that

regional landscape pattern and environment change

are key component to develop policies in the

ELWNNR and control environmental pollution. The

environment should improve CONTAG, SHDI and

CONNECT indices at the landscape and class levels,

but affect CONNECT, NP, PD, and ED indices. The

entire study is based on remote sensing

interpretation and requires accurate data that should

be further improved in subsequent studies.

Based on Landsat TM, in 1998 and 2006, and

Landsat OLI remote sensing images, from 2014, this

research studies current and future changes of land

use/cover-landscape patterns and establishes a

quantitative expression of landscape pattern and

environmental quality indices. We can conclude that:

(1) The LUCC in the ELWNNR shows a trend with

“three increases and three decreases” in the

descriptor indices used in this study. From 1998 to

2006, the expansion of dry lake bed was notorious.

In 2014, the desert continued to expand, with a

dynamic degree of 34.1%. From 2014 to 2022, the

dry lake bed, water bodies, and other areas have a

trend to decrease. By 2030, the land use/cover type

conversion area will be smaller. (2) The landscape

indices and the environmental quality indices in the

study area are significantly correlated, proving that

the composition of the landscape and the spatial

structure of the land use have a great impact on the

regional environmental quality and the landscape

indices can be used to estimate the environmental

quality.

ACKNOWLEDGEMENTS

We are grateful for the financial support provided by

the Natural Science Foundation of Xinjiang Uygur

Autonomous Region, China (2016D01C029), the

authors wish to thank the referees for providing

helpful suggestions in improving this manuscript.

REFERENCES

Adhikari S, Southworth J 2012 Simulating forest cover

changes of Bannerghatta National Park based on a CA-

Markov model: a remote sensing approach Remote

Sens. 4(10) 3215

Al-sharif AAA, Pradhan B 2013 Monitoring and

predicting land use change in Tripoli Metropolitan City

using an integrated Markov chain and cellular

automata models in GIS Arab. J. Geosci. 7(10) 4291

Bing Z, Gao J., 2015. Research on the Impact of Human

Activities on the Landscape Pattern in Jiuzhaigou

Nature Reserve. Aasri International Conference on

Circuits and Systems.

Brun C, Cook A R, Lee J S H, et al. 2015 Analysis of

deforestation and protected area effectiveness in

Indonesia: A comparison of Bayesian spatial models

Global Environmental Change 31 285

Bai, X., 2010. Study on the wetland ecological

vulnerability and its driving system in Ebinur lake in

XinJiang. East China Normal University.

Chen L, Yang X, Chen L, et al. 2015 Impact assessment

of land use planning driving forces on environment

Environmental Impact Assessment Review 55(6) 126

Chen S F, 2015 Relationship between land use and soil

erosion in the subtropical mountain areas of north

Guangdong province Journal of Arid Land Resources

and Environment 29(2) 80

Camacho C, Pérez-Barahona A, 2015 Land use dynamics

and the environment Journal of Economic Dynamics&

Control 52(52) 96

Relationship between Landscape Pattern and Environmental Indices in Ebinur Lake Wetland National Natural Reserve

137

Conrad C, Rudloff M, Abdullaev I, et al. 2015 Measuring

rural settlement expansion in Uzbekistan using remote

sensing to support spatial planning Applied Geography

62 29

Duan Z, Zhang F, Kong X, 2005 Method for information

mining of land-use change and its application

Transactions of the Chinese Society of Agricultural

Engineering 21(12) 60

Han C J, Wu K N, Liu D Y, et al. 2011 On Cropland

Prediction of Zhengzhou Based on Markov Model in

Multi-Projects Soils 43(3) 453

Hansen A J, Defries R, 2007 Ecological Mechanisms

Linking Protected Areas to Surrounding Lands

Ecological Applications 17(4) 974

Lei B, Zhu K W, Li J H, et al. 2016 Study on the Influence

of Land Use Change on Landscape Pattern and Eco-

environmental Status Environmental Impact

Assessment 38(3) 87

Liu Y, Gao J, Yang Y, 2003 A holistic approach towards

assessment of severity of land degradation along the

Great Wall in northern Shaanxi Province, China

Environmental Monitoring & Assessment 82(2) 187

Lu C Y, Sun Q Y, Li H, et al. 2014 Estimation of

groundwater recharge in arid and semi-arid areas based

on water cycle simulation Shuili Xuebao 45(6) 701

Moghadam B K, Jabarifar M, Bagheri M, et al. 2015

Effects of land use change on soil splash erosion in the

semi-arid region of Iran Geoderma 241–242 210

Paroissien J B, Darboux F, Couturier A, et al. 2015 A

method for modeling the effects of climate and land

use changes on erosion and sustainability of soil in a

Mediterranean watershed (Languedoc, France) Journal

of Environmental Management 150 57

Tsai Y, Zia A, Koliba C, et al. 2015 An interactive land

use transition agent-based model (ILUTABM):

Endogenizing human-environment interactions in the

Western Missisquoi Watershed Land Use Policy 49

161

Terra T N, Santos R F D, Costa D C, 2014 Land use

changes in protected areas and their future: The legal

effectiveness of landscape protection Land Use Policy

38(2) 378

Vadrevu K P, Justice C, Prasad T, et al. 2015 Land

cover/land use change and impacts on environment in

South Asia Journal of Environmental Management

148 1

Veldkamp A, Lambin E F, 2001 Predicting land-use

change Agriculture Ecosystems & Environment 85(1-3)

Wang Y, Shao M, Zhang C, et al. 2015 Choosing an

optimal land-use pattern for restoring eco-

environments in a semiarid region of the Chinese

Loess Plateau Ecological Engineering 74(5) 213

Wang K, Wang H J, Shi X Z, et al. 2009 Landscape

analysis of dynamic soil erosion in Subtropical China:

a case study in Xingguo County, Jiangxi Province Soil

& Tillage Research 105(2) 313

Wang L, Ding J L, 2015 Vegetation index feature change

and its influencing factors and spatial-temporal process

analysis of desert grassland in the Ebinur Lake Nature

Reserve, Xinjiang Acta Prataculturae Sinica 24(5) 4

Wu J G, 2007

Landscape Ecology—Pattern,

Process, Scale and Grade. Beijing: Higher

Education Press. (in Chinese)

Xie X, Wang H W, Tashpolat T, 2010 Study on

Landscape Pattern Change in Ebinur Lake Region

based on RS and GIS Journal of Desert Research 30(5)

1166

Xia J, Zuo Q T, 2006 Advances in International

Hydrological Science Research Advances in Earth

Science 21(3) 256

Ye X, Liu G, Li Z, et al. 2015 Assessing Local and

Surrounding Threats to the Protected Area Network in

a Biodiversity Hotspot: The Hengduan Mountains of

Southwest China Plos One 10(9) e0138533

Zhang Y, Zhang F, Wang J, et al., 2016 Evaluation of the

Landscape Patterns Vulnerability and Analysis of

Spatial -Temporal Patterns in the Typical Region of the

Ebinur Lake Journal of Catastrophology 31(3) 222

Zhao G S, Liu J Y, Kuang W H, et al. 2015 Disturbance

impacts of land use change on biodiversity

conservation priority areas across China during 1990-

2010 Acta Geographica Sinica 69(5) 1640

Zeng, Y. N., 2014 Simulation of land-use changes and

landscape ecological assessment in eastern part of

Qinghai Plateau Transactions of the Chinese Society of

Agricultural Engineering 30(4) 185

IWEG 2018 - International Workshop on Environment and Geoscience

138