Advertisement

Vegetation change detection research of Dunhuang city based on GF-1 data

  • Zhaopeng Zhang
  • Zengyuan Li
  • Xin Tian
Open Access
Research Article

Abstract

This study selected the Dunhuang city with the unique landscape pattern, its oasis surround by desert, for studying the vegetation change. Based on two period GaoFen-1 images, combined with field survey data, the dimidiate pixel model and fractional vegetation coverage estimation model were applied to detect the changes of local vegetation coverage from July 2013 to July 2015. Analysis showed that during the study years, the Grain for Green project had a remarkable effect on the changes. The vegetation types of intermediate high and very low grades increased by 0.66 and 6.78 km2 respectively. The low vegetation coverage decreased by 23.87 km2. The vegetation coverage types of intermediate grade increased by 88.97 km2 because of the planted forest, which accounted for 10.84% of the study area. The vegetation coverage types of high grade reduced by 72.47 km2. This change effectively prevents the spread of desert and lays a good foundation for the ecological construction of Dunhuang city.

Keywords

Dunhuang city GF-1 Vegetation coverage Grain for green 

1 Introduction

Oasis refers to a green land with weeds in the desert. Dunhuang oasis, is one of the main bases for human settlement and production in the arid and semi-arid regions of Northwest China. Fractional Vegetation Cover (FVC) is the basic index of ecological system, which is affected by the changes of natural environment and human activities. In order to understand the current situation, trends, driving forces, and environmental impacts of the vegetation cover in desert oasis, using remote sensing date, numerous studies have been performed (Yang et al. 2010; Liu et al. 2013; Zhang et al. 2015). However, there are few research based on the Chinese high resolution satellite constellation. Based on two period GaoFen-1 (GF-1) images, the dimidiate pixel model and vegetation coverage estimation model were applied to detect the changes of vegetation covers from 2013 to 2015 in Dunhuang oasis. The goal is of this study is to provide a reference for accurately understanding the situation of regional ecological change and evaluating the effectiveness of the ecological projects.

2 Materials

2.1 Study area and data

Dunhuang city of the northwest of Gansu province, is located at the ancient Silk Road. The territory is surrounded by mountains at three aspects. The Danghe river is the only water source for the survival of the city. GF-1 images (two for July 2013 and two for July 2015) covering Dunhuang city were obtained from China Center for Resources Satellite Data and Application. GF-1 is a kind of high resolution earth observation satellite with independent property rights in China (Fig. 1). Data pre-processing was conducted by the ENVI software (version 5.4.1).
Fig. 1

Geographical position of study area

2.2 Image data preprocessing

The RPC file of GF-1 images in 2013 and the corresponding DEM from Geospatial Data Cloud were used to make the geographic correction. After that, the GF-1 images in 2015 were processed to match with the GF-1 images from 2013 by means of an image-to-image matching method provided by the ENVI Imagine software. The root mean square (RMS) control of each control point is less than 1 pixels, and the total root mean square error (RMSE) is 0.49 (Fig. 2).
Fig. 2

GF-1 images of the study area (Left: 2013a, Right: 2015a)

3 Methods

3.1 NDVI and vegetation coverage

The normalized difference vegetation index is known to be an index of vegetation conditions (Carlson and Ripley 1997). This index is expressed as:
$${\text {NDVI}} = \left( {\text {NIR - R}} \right)/\left( {\text {NIR - R}} \right)$$
(1)
NIR and R represent the reflectivity of the near infrared and red bands, respectively (Ye et al. 2009).
Combined with the content of the study, the NDVI gray scale images of different time images are obtained by using NDVI. It is necessary to transform the vegetation index into the degree of coverage. In a certain extent, FVC can make up for the disadvantage of NDVI, in which NDVI is difficult to distinguish the low coverage vegetation and the coverage is easy to be saturated. On the contrary, so as to effectively stretch the range of vegetation information. The vegetation cover can be divided into two types, pure and mixed pixels by the different structure characteristics. A mixed pixel includes two kinds of information of soil and vegetation. Assume that the vegetation index is NDVIveg, and the vegetation index is NDVIsoil, and the mixed pixel vegetation index is:
$${\text{NDVI}} = {\text{FVC}} \times {\text{NDVIveg}} + \left( {1 - {\text{FVC}}} \right) \times {\text{NDVIsoil}}$$
(2)
The vegetation coverage model is:
$${\text{FVC}} = \left( {{\text{NDVI}} - {\text{NDVIveg}}} \right)/\left( {{\text{NDVIsoil}} - {\text{ NDVIveg}}} \right)$$
(3)

In the above formula, NDVIveg represents the normalized difference vegetation index of the whole vegetation cover, NDVIsoil represents the normalized difference vegetation index of bare soil.

The key of the mixed pixel decomposition method is the acquisition of NDVIveg and NDVIsoil. In this paper, NDVIsoil and NDVIveg parameters were generated by using the NDVI threshold of land cover type. The following expressions are used to calculate the NDVIsoil and NDVIveg:
$${\text{NDVIsoil}} = {{b}}1*{{n}}1 + {{b}}2*{{n}}2 + {{b}}3*{{n}}3 + {{b}}4*{{n}}4 + {{b}}5*{{n}}5$$
(4)
$${\text{NDVIveg}} = {{b}}1*{{n}}6 + {{b}}2*{{n}}7 + {{b}}3*{{n}}8 + {b}4*{{n}}9 + {{b}}5*{{n}}10$$
(5)
where b1, b2, b3, b4 and b5, are forest land cover document, agricultural land cover document, urban land use mask files, water Mask documents and other land use mask file respectively; n1 and n6, n2 and n7, n3 and n8, n4 and n9, n5 and n10, are the NDVI minimum and maximum value of corresponding each mask file respectively expressed. The Eqs. 23 were applied to conduct the transformation of gray scale map of vegetation cover (Zhang et al. 2017).

3.2 Classification of vegetation coverage

According to the criteria of classification in desertification survey of China, and the characteristics of Dunhuang oasis vegetations, the coverage vegetation in the study area were divided into five grades, (1) very low coverage vegetation, FVC from 0.00001 to 0.1; (2) low coverage of vegetation, FVC from 0.1 to 0.3; (3) Intermediate coverage of vegetation, FVC from 0.3 to 0.5; (4) middle and high coverage vegetation, FVC from 0.5 to 0.7 and (5) high coverage of vegetation, FVC larger than 0.7. The detailed classification schemes are shown in Table 1. Normally, the transformation of the low level vegetation to the high level is called the vegetation growth, and the transformation of the higher vegetation to the lower one is named as the vegetation degradation. According to the above classification criteria, the vegetation coverage map of the different periods is obtained (Fig. 3). Through field survey and land use map can know vegetation classification results is right. Then, the vegetation area and its changes from 2013 to 2015 were obtained by difference operation.
Table 1

Classification scheme used in this study

Classification

Ranges

Instruction

Very low coverage vegetation

0.00001–0.1

Intensity of desertified land, bare soil, ground flowing sand, gobi, water bodies, etc.

Low coverage vegetation

0.1–0.3

Sparse forest land, semi fixed sandy land, low yield grassland and seriously degraded land, etc.

Intermediate coverage vegetation

0.3–0.5

Fixed sandy land, light sandy land, low density forest land and cultivated land, etc.

Middle and high coverage vegetation

0.5–0.7

Middle and high yield grassland, forest land and cultivated land

High coverage vegetation

> 0.7

High yield grassland, dense forest land, good cultivated land and so on.

Fig. 3

The FVC classification figures (Left: 2013a, Right: 2015a)

4 Dynamic analysis of vegetation cover

4.1 Overall change of vegetation coverage

From 2013 to 2015, the vegetation coverage of Dunhuang oasis was provided in Table 2 that showed a great changes among FVC to the sustainable development of the Oasis Economy. In these five levels of vegetation cover types, compared with the situation in 2013, the coverage of vegetation increased 10.84%, very low vegetation coverage and high coverage vegetation remained almost unchanged, only increased by 0.83% and 0.08% respectively, and the low vegetation coverage and high vegetation coverage were reduced by 2.92%, 8.83%.
Table 2

The percentage of various types of vegetation to the total area (%)

Year

Very low

Low

Intermediate

Middle and high

High

2013

1.05

50.95

15.26

18.73

14.01

2015

1.88

48.03

26.10

18.81

5.18

4.2 The mutual changes of different vegetation cover types

As shown in Tables 3 and 4, it is obvious that very low coverage vegetation had 93.32% positive transfer, the total change area accounted for 78.22% of the total vegetation area of the initial state. Low coverage vegetation had 15.64% of the transfer. In the vegetation cover degree, there were 33.20% transfer, in which the negative transfer to 15.91%, positive transfer to 17.29%, but its total area increased by 88.97 km2. In the middle and high vegetation coverage, there were 46.47% negative transfer and 8.40% positive transfer. High coverage vegetation transfer is relatively large, there were 79.65% occurrence of negative transfer.
Table 3

Various types of vegetation changes in circumstances (km2)

Year

Background

Very low

Low

Intermediate

Middle and high

High

2013

1002.8554

15.441676

394.1752

214.159766

154.350288

42.4747

2015

1002.9178

8.664360

418.046336

125.188544

153.692192

114.947648

Table 4

Transform situation matrix of vegetation (%)

2013–2015

0–0.00001

Very low

Low

Intermediate

Middle and high

High

Background

0.000

0.000

0.000

0.000

0.000

0.000

0–0.00001

99.927

0.877

0.119

0.045

0.018

0.014

Very low

0.011

5.804

3.487

0.118

0.043

0.032

Low

0.054

92.080

84.356

15.749

5.566

4.123

Intermediate

0.008

1.152

10.997

66.802

40.863

18.770

Middle and high

0.000

0.071

0.778

13.242

45.110

56.706

High

0.000

0.017

0.264

4.043

8.400

20.355

*0–0.00001 is a special treatment for the classification, in order to remove background values and not allow them to participate in the classification

5 Analysis on the factors of vegetation cover change

5.1 Natural factors

Natural factors were mainly analyzed from two aspects, namely, precipitation and vegetation.

Dunhuang city is characterized by little precipitation but large evaporation. The average amount of evaporation is 60 times of precipitation, and there almost no conversion among precipitation, soil surface water and ground water. Precipitation is one of the most important factors affecting vegetation coverage. The precipitation from 2013 to 2015 in Dunhuang city is shown in Table 5, which showed that both the annual precipitation and the summer precipitation in 2013 and 2015 were very small. As a result, the water supply for agriculture and forestry mainly came from groundwater. Water stress is the mainly problem of Dunhuang city and because of the different farming conditions, the local areas change greatly from 2013 to 2015.
Table 5

Statistical table for precipitation information

Year

Time ranges

Precipitation(mm)

The days of precipitation (d)

Observation times (d)

Data sources

2013

From January 1st to December 31st

48

43

1458

The data set is provided by the RP5 website

From May 1st to July 31st

32

24

368

2014

From January 1st to December 31st

58

37

1460

2015

From January 1st to December 31st

33

42

1459

From May 1st to July 31st

20

22

368

Grassland and desert are the main vegetation types of the Dunhuang city. The main types of plants are Populus euphratica, Tamarisk and Manaplant Alhagi, which can resistant to drought, wind and sand and adapt to extreme drought climate conditions. In particular, the oasis fringe area, its vegetation types are more barren, and the vegetation species structure is simple with extremely low FVC. Which is consistent with that shown in Fig. 3, the oasis surrounding is mainly the extremely low vegetation coverage.

5.2 Social factors

In combination with various factors, social factors were also main factors affecting the vegetation cover change in Dunhuang oasis. Social factors were mainly analyzed from these aspects, namely, population, economic, government industrial structure conduction and so on.

As the artificial forests increase, the effects of Grain for Green project are significant. According to the information from official website of the Forestry Bureau of the Dunhuang City. In recent years, the amount of afforestation in Dunhuang is 7.12 million hectares, 14.5 million hectares of afforestation from the project 0, and 160 million acres natural vegetation were protected and managed. Only in 2015, the afforestation area is beyond 453 hectares, 1180 hectares of economic forests being newly planted, and 91.2 km of farmland shelterbelts were built and improved. As a result of large amounts of farmlands were converted to artificial forests, the medium vegetation coverage increased largely and the high vegetation coverage dereased.

Dunhuang was named as national comprehensive demonstration area of desertification prevention and control in 2013. 190 hectares sands were got under control in Mingsha Mountain Montenegro, Mogao Grottoes and other areas. The total area of the ban on desertification is 40,612 hectares (This detailed statistics comes from the Dunhuang City Propaganda Department). This government behavior leads to a decrease of low vegetation coverage and the increase of intermediate vegetation coverage in Mingsha Mountain, located in southeast of the study area.

With the rapid development of economy, the local industrial structure has been constantly adjusted. The Dunhuang City is an indispensable part for the “Silk Road Economic Zone” of “the Belt and Road” (B&R). As a result, “the Belt and Road” strategy not only brought more opportunities for the development, but also promoted its industrial restructuring, especially for the service industry. With the optimization and upgrading of industrial structure, the service industry, which is dominated by tourism, has became to be the main force for city’s economic development. The income from the tourism and cultural industries is shown in Fig. 4.
Fig. 4

Two types of economic changes from 2011 to 2016 (Top: tourist industry, Bottom: culture industry)

During the period of the China 12th Five-Year Plan, which is scheduled from 2011 to 2015, the three industries of the city were fully developed, with the primary, second and service industries increasing by 3.7, 18.1 and 15.5%, respectively. These three industrial structures tend to be optimized from 21.2: 26.5: 52.3 in the end of the 11th Five-Year Plan (2010) to 16.2: 30.3: 53.5 in the middle of the 12th Five-Year Plan (2013) and then to 13.3: 28.3: 58.4 at the end of the 12th Five-Year Plan (2015). The proportion of the first and secondary industry decreased by 2.9% and 2.0%, the proportion of the tertiary industry increased by 4.9%. The city’s three industrial structures were shown in Table 6.
Table 6

The proportion of industry of Dunhuang city (%)

Year

Primary industry

Second industry

Service industry

Total

Notes

2006

27.6

19.1

53.3

100

The industry status of Dunhuang city were obtained by the Gansu provincial bureau of statistics and the Dunhuang municipal government network

The 12th Five-Year Plan: from 2011 to 2015

The 11th Five-Year Plan: from 2006 to 2010

2010

21.2

26.5

52.3

100

2011

19.5

28.7

51.8

100

2012

17.8

32.4

49.8

100

2013

16.2

30.3

53.5

100

2014

14.8

29.1

56.1

100

2015

13.3

28.3

58.4

100

2016

13.7

25.4

60.9

100

With a slight increase of the local population, the employment structure was gradually optimized. The census data published by Dunhuang City Government Network showed that by the end of 2012 and the end of 2016, the number of household population is 141,969 and 143,800, respectively. However, with the gradual optimization of the employment structure, the labor force has been transferred from the first and second industries to the tertiary industry, and as a result, some of them moved from rural to urban areas. According to the 12th Five-Year Plan Economic and Social Development Review for Dunhuang city, published by the Dunhuang government network, the employment structure of the three industries in this city changed from 41.8:15.4:42.8 in 2010 to 37.8:14.9:47.3 in 2015. The reduction of labor force in the primary industry has a direct impact on the reduction of cultivated land area. This is also an imply for the increase of intermediate vegetation coverage and the decrease of high vegetation coverage in Table 2.

6 Conclusions

In this study, based on GF-1 NDVI information and dimidiate pixel model, changes of vegetation coverage between 2013 and 2015 in the Dunhuang oasis were analyzed. The results showed that the implement of Grain for Green project has largely affected the changes. According to the changing static information, the middle and high vegetation coverage, and extremely low vegetation coverage increased 0.66 and 6.78 km2, respectively, and low vegetation coverage reduced by 23.87 km2, and the intermediate vegetation coverage increased by 88.97 km2, and high vegetation coverage area decreased by 72.47 km2.

In summary, the social factors are the most significant factors affecting the changes of the Dunhuang. In the light of the present vegetation coverage situation of the Dunhuang oasis, the protection work for the oasis should be continually conducted in order to keep the oasis stability. Meanwhile, the local water saving agriculture and Grain for Green should also be further developed.

Notes

Acknowledgements

This study was supported by the Fundamental Research Funds for the Central Non-profit Research Institution of Chinese Academy of Forestry (CAFYBB2017QC005) and the National Natural Science Foundation of China (21-Y20A06-9001-17/18).

References

  1. Carlson TN, Ripley DA (1997) On the relation between NDVI, fractional vegetation cover and leaf area Index. Remote Sens Environ 62(3):241–252CrossRefGoogle Scholar
  2. Liu XF, Yanu Y, Ren ZY, Lin ZH (2013) Changes of vegetation coverage in the loess plateau in 2000–2009. J Desert Res 33(4):1244–1249Google Scholar
  3. Yang LN, Niu SW, MA LG, Li YX (2010) Characteristics and cause analysis of land use/cover change in Dunhuang City. Chin J Ecol 29(4):766–775Google Scholar
  4. Ye GX, Li WQ, Tian Y (2009) Dynamic changes of vegetation cover in typical oasis of arid areas based on NDVI—a case study in Cele Oasis. J Arid Land Resour Environ 23(9):128–133Google Scholar
  5. Zhang H, Qu JJ, Zhang KC (2015) Vegetation cover information extraction technology for Dunhuang oasis based on remote sensing images. J Desert Res 35(2):229–234Google Scholar
  6. Zhang ZP, Li ZY, Li QH, Hao RX (2017) Dynamic analysis on the vegetation coverage changes of Minqin Oasis based on GF-1 from 2013 to 2015. J Southwest For Univ 37(2):163–170Google Scholar

Copyright information

© The Author(s) 2018

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors and Affiliations

  1. 1.Surveying and Mapping Institute of Science and TechnologyXi’an University of TechnologyXi’anChina
  2. 2.Institute of Forest Resources Information TechniquesChinese Academy of ForestryBeijingChina

Personalised recommendations