Luận văn thạc sĩ: Applications of remote sensing and GIS to mapping land cover change in Son la province

<span class="text_page_counter">Trang 1</span><div class="page_container" data-page="1">

MINISTRY OF TRAINING MINISTRY OF AGRICULTURE AND AND EDUCATION RURAL DEVELOPMENT

THUY LOI UNIVERSITY

Nguyen Thi Lien

APPLICATIONS OF REMOTE SENSING AND GIS TO MAPPING LAND COVER CHANGE IN SON LA PROVINCE

Master Thesis

</div><span class="text_page_counter">Trang 2</span><div class="page_container" data-page="2">

MINISTRY OF TRAINING MINISTRY OF AGRICULTURE AND AND EDUCATION RURAL DEVELOPMENT

THUY LOI UNIVERSITY

Nguyen Thi Lien

APPLICATIONS OF REMOTE SENSING AND GIS TO MAPPING

LAND COVER CHANGE IN SON LA PROVINCE

Field of study: Disaster Mitigation

Master Thesis

Advisors: Assoc. Prof. Hoang Thanh TungDr. Vu Thanh Tu

</div><span class="text_page_counter">Trang 3</span><div class="page_container" data-page="3">

I am indebted to my respected Assoc. Prof. Hoang Thanh Tung and Dr. Vu Thanh Tu who work as lecturers in Department of Hydrology and Water resources in Thuy Loi University for their continuous guidance, advice and expedience from the proposal preparation to thesis finalization. Their constructive comments, untiring help, guidance and practical suggestions inspired me to accomplish this work

Besides, I am especially grateful to Dr. Nguyen Quoc Khanh and members in the Department of Geologycal and Remote Sensing in Vietnam Institute of Geosciences and Mineral Resources (VIGMR) who supported me in terms of the data collection and gave me useful advices for my thesis.

I remember all those who have contributed directly or indirectly to successfully completing my study.

Finally, I must express my very profound gratitude to my family for providing me with unfailing support and continuous encouragement throughout my years of study and through the process of researching and writing this thesis. This accomplishment would not have been possible without them. Thank you.

Hanoi, May 11" 2017

Nguyen Thi Lien

</div><span class="text_page_counter">Trang 4</span><div class="page_container" data-page="4">

I hereby certify the work which is being presented in this thesis entitled,

“APPLICATIONS OF REMOTE SENSING AND GIS TO MAPPING LAND COVER CHANGE IN SON LA PROVINCE” in partial fulfillment of the

requirement for the award of the Master of Disaster Management, is an authentic record of my own work carried out under supervision of Assoc. Prof. Hoang Thanh

Tung and Dr. Vu Thanh Tu. The matter embodied in this thesis has not been

submitted by me for the award of any other degree or diploma.

Nguyen Thi Lien

il

</div><span class="text_page_counter">Trang 5</span><div class="page_container" data-page="5">

The thesis “Application of remote sensing and GIS to mapping of land cover changes in Son La province" was carried out and finished in May 2017. Purpose of the thesis is to study and apply remote sensing technology and GIS in mapping changes of land cover in Son La Province. To meet the thesis requirements, the following tasks have been implemented:

Study of land cover, remote sensing and GIS theories.

Collection of satellite imagery data and statistical data for classification, intepretation of land cover maps over years. Implementation of sptial analisis in GIS for evaluation of land cover change map.

Conclusions about the results achieved and the assessment methodology.

After the implementation process, the subject has obtained some results:

Land cover map of Son La province in 1999 and 2015 with 7 types of land cover: bush, lake-river, natural forest, planted forest, bare soil, agricultural land, specialized land.

Land cover changes map in Son La province during 1999-2015. With the results achieved, remote sensing technology and GIS can be seen as

effective methods with relatively high accuracy, cost savings in classification and catalytic activity.

11

</div><span class="text_page_counter">Trang 6</span><div class="page_container" data-page="6">

GIS Geographic Information System TM Thematic mapper

ETM Enhanced Thematic Mapper

MLC Maximum Like hood classifer method

ROI Region Of Interest

</div><span class="text_page_counter">Trang 7</span><div class="page_container" data-page="7">

TABLE OF CONTENTS

INTRODUCTION... o5 5< 5< <5. HH HH 0000100 000000050909084 1

Research rationale 0... e...4.. 1 Main objective of the r€S€aTCH:...- 0 5 1112101193011 0112 ng HH ve 1 Subject and scope of the research ...- ..-- G2 3211331188 1113 111911 9111 11 1g ng ng ng 1 Structure of the research ew... ee 5 c1 9301 391 910 910v HH HH nh nh 1 CHAPTER I: LITERATURE REVIE WS... So H000 00800886 3

1.1. Overall study on applications of Remote Sensing and GIS in mapping land

cover Changes 1n the 407 ¡01 ... 3 1.2. Overall study on applications of Remote Sensing and GIS in mapping land

cover Changes 1n VietNam ...cccccesccesscesssesseeceseeeeseessnecseeceseeceaecseaeceeecseeeeseeeeseseseeeeaees 5 1.3. Approach of the r€S€aTC...- .. -- c1 TH Họ HH key 7 1.3.1. The remote sensing is used for monitoring land cover changes ... 10 1.3.2 GIS spatial analysis is used for evaluation of land cover changes... 14 CHAPTER II: RESEARCH ON THE APPLICATION OF REMOSTE

SENSING AND GIS IN ESTABLISHMENT OF LAND COVER CHANGE

MAP OF SON LA PROVINCC|E...o 5< 5< 5s s9... HH TH 000009 009660968996 17 2.1. Overview Of StUCY AT€A... 0Q TH. HH TH TH HH nh 17 2.1.1. Natural and socIo-economic CONCITIONS ... ..- -- 5 5 s13 ren 17 2.1.2. Characteristics of land cover in the study ar€a... ..- 5c «sec scssssesseeske 18 2.2. Data collection and afiaÏYS1S...- - -- «c1 311211191 9v ng 19

2.2.1. Remote sensing đafa... cece «+ 4 +31 TT HH 19

2.2.2. Land USE Status ... 0 HH HH tt 21

2.2.3. Field SUIVCY oo... cece . ... 25

2.3. Preprocess of remote SENSING IMALE ...- G5 c1 3210113111191 1 111 11 11 re 262.3.1. Radiant calIDrafIOII... - -.- 5 «cv HT TH HH 26

</div><span class="text_page_counter">Trang 8</span><div class="page_container" data-page="8">

2.3.2. Enhance image Visibility ... -- -- - <2 3111891118911 13911 19111 9 1 vn vn rry 26 2.3.3. CŒ@OIm€fTIC COTT€CfIOT1...- G13 E91111 1911 91011 TH HH HH ghi Hà 28 2.4. Image classification Im€thO... .- - --- << + 61131113111 911 91v He 30 2.4.1. Unsupervised classification methO(...- --- c6 + 1k E*EEskEsekeskeeekese 30 2.4.2. Supervised classification mmethOC...- --- << + +3 E3 EeESeeerseeereeeeeeere 31 2.5. Evaluate accuracy and image processing after classificafion... ---«-+ 36

2.5.1. Evaluate the accuracy after classification ...ccscesscesseeenseeeeeeseeeeseeeeeenenees 36 2.5.2. Image processing after classification 0.0.0... cc sceseseeseeseeneeeeeeeeeseeeeeeeeeeseeseeaeens 37 2.6. Establish land cover map and land cover change map...- ---‹ +5 «<++<<++ss2 37

CHAPTER III: RESULTS AND DISCUSSIONS... Ăn 01165 see 40 3.1. Image classification results 1n 19.,...- -- + xxx ng HH re 40 3.2. Image classification results 1n 2015 ...- --- --- 5 + 2+ x11 ng ng it 44 3.3. Established results of land cover change maD...-- s5 5+ + ++seseeereseeeesers 46

CONCLUSIONS AND RECOMMENDA TIONG... on 1 0e 51

2. RecommMendations nh ...e... 51

REFERENCES .0...csccsscsssccscsscssssenccseccscsssccsscssssescssscssscesscesessscsssssssssscesscssssssssocssoeesess 53

vi

</div><span class="text_page_counter">Trang 9</span><div class="page_container" data-page="9">

LIST OF FIGURES

Figure 1. Overview Of r€S€ATCH... G1 TH ng TH HH Hit 8

Figure 3.Spectral reflective characteristics of some of the main natural objects... 10

Figure 5. Land use map in Son La province in 2005 ... ..- --c c + k*sv+skEserseeseree 24 Figure 6. Field survey of sample points ...-- - << 11191119111 91 1993 19 vn ng rưy 25 Figure 7. Enhance image VISIDIÏIEV... -.-- -- << +1 11191 91119 1x HH ng ky 27 Figure 8. Image before and after being enhanced image qualIty...-- --- eee 28 Figure 9. Landsat 7 image in 1999 after geometric grafting and correcfion... 29 Figure 10. Landsat 8 image in 2015 after geometric grafting and correction... 30 Figure 11. Select the sample area... ...ầ.... 34

Figure 13.Use the Crosstab tool in the IDRISI software to calculate sub-carpets

fÏUC{UAfIOTNS... Gọi 39 Figure 14.Photo classification result 1n I ,...- - - + x91 9910191 91g ngư 41 Figure 15. Image classification results 1n 2 ÏŠ,...-- -- -- + ++< 1+ E**EE+eeEEseereeeerrsersee 44

Vil

</div><span class="text_page_counter">Trang 10</span><div class="page_container" data-page="10">

LIST OF TABLES

Table 1.Some interpretation signs on the fake color combination of SPOT satellite

TMA QOS... ... 5Õ... 14 Table 2. Distribution of vegetation cover and current land use status in 2011... 18 Table 3. Remote sensing image (Ìafa... ... - - -- c1 1191119111191 9 11 9111 8111 ng vn rưy 20 Table 4. Land use status of Son La province in 2005 as folÏOWS:...- 5< «<< <<++ 22 Table 5. Statistics of sample scores for each type Of COVED... cceeceeseeeteeeseeeetseeeeenenees 25

Table 6. Some interpretive patterns used in the topic ... Error! Bookmark not defined.

Table 7. Assessment accuracy according to Kappa coefficientand overall accuracy

for 1999 image classification results in Son La province ...- 5 5< £+c+sessessee 42 Table 8. Statistics of area covered by each type of cover in Son La province in 1999 .... 42 Table 9. Statistics of area covered by each type of cover in districts of Son La

PTOVINCE IN 1999. 1... ...ốố... 43 Table 10. Assessment of accuracy according to Kappa coefficientand overall

Table 11. Statistics of cover area of each type of cover in Son La province in 1999... 45 Table 12. Statistics of area covered by each kind of cover in districts of Son La

PFOVINCE IN n1... ... 46 Table 13. The number of pixel land cover change between 1999 and 20015 in Son La

PIOVINCE. eee ... 47

Table 14. Pixel ratio changes in two periods 1999 and 2015 in Son La province... 47

Table 15. Evaluate the maturation variation in stage 1999-2015...---c«++<<s+ssc<ss 49

Vili

</div><span class="text_page_counter">Trang 11</span><div class="page_container" data-page="11">

Research rationale

Land is an extremely precious natural resource, a special production material. Land

is the living environment of humans as well as creatures, the territory of population distribution, construction of economic, cultural, national security works. Today, due

to population increase, urban development, socioeconomic growth and other problems are currently having large impacts on land, especially for Son La province. Up until now on the territory of Son La province there have been 49 hydropower plants, which have considerable impact on the narrowed forest area. Standing before such pressure, land and ground decks also convulse non-stop with the development of society. This is a special resource that can be exploited for use but cannot be increased in terms of quantity. Thus the monitoring, research, management and usage of this resource effectively and reasonably are a very important matter.

Remote sensing technology is applied more and more widely in many sectors, many fields — hydrology, geology, environment to agriculture — forestry — fishery... in which land cover changes is monitored with high accuracy, from which it can help managers to acquire more materials to supervise the convulsion of Land use. This can be considered one of the solutions to the problems set out. On the other hand this method has not been applied for piloting in Son La province. Thus, the topic of “Remote sensing and GIS application for establishment of land cover changes map of Son La province” was selected to perform for my thesis.

Main objective of the research

The main objective of the research is to study remote sensing and GIS and their applications in assessment of land cover changes. A case study is then applied to Son La province of Vietnam during the 1999-2015 periods.

Subject and scope of the research

Subjects of the research are Remote Sensing, GIS and their applications in mapping land covers and in assessing of land cover changes.

Scope of the research is Son La Province in the Northern Part of Vietnam. Structure of the research

In additional to the Introduction, Conclusion and Recomendation, the thesis is

structured in 3 chapters as the followings:

</div><span class="text_page_counter">Trang 12</span><div class="page_container" data-page="12">

Chapter I: Literature review. In this chapter the author focus on researching the theoretical basis on the decks on the surface of the earth and domestic and foreign researches on the application of remote sensing and GIS in research of land cover changes and Land use. In addition in this chapter the author shall also orientate the research, and generalize the steps of data collection, image analysis and role of remote sensing — GIS in research of land cover convulsion.

Chapter IT: Research of remote sensing and GIS application in establishment of land cover changes map of Son La province. In this chapter the author focus on

researching the following issues:

- Overview of research area: The status quo of the research area land cover.

- Data collection and analysis: Remote sensing data, Land use status, field

survey data.

- Pre-processing of remote sensing image: Radiation correction, image quality enhancement, geometry correction.

- Methods of image classification, post-classification accuracy assessment. - Methods of land cover changes map establishment.

Chapter III: Results and Discussions. In this chapter the author shall present the achieved results such as:

- Land cover classification result of 1999. - Land cover classification result of 2015.

- Land cover changes map during the period of 1999-2015.

</div><span class="text_page_counter">Trang 13</span><div class="page_container" data-page="13">

CHAPTER I

LITERATURE REVIEWS

1.1. Overall study on applications of Remote Sensing and GIS in mapping land

cover changes in the World

In the world, research topics on the convulsion of land use form for analysis, assessment, and forecast of development have been applied fairly widely.

In the topic “Remote sensing based quantification of land cover and land use change for planning” (Bjorn Prenzel, 2003), the author made scientific bases of the selection of method for quantitative results in the research of land cover changes and usage based on remote sensing basis. In which depending on the situation that we use methods based on defined theory or based on experience. A noteworthy point that the author mentioned is the requirement of data when assessing convulsion: the collected data shall have the same characteristics (regarding space, spectral

resolution...) and the data shall reach certain standards in terms of cloud or fog, and the collected data shall be in the same research area

In the research “Land use/ land cover changes detection and Urban Spawl Analysis”

(M. Harika, et al., 2012) the author assessed the form of land use/ land surface at the

cities of Vijayawada, Hyderabab and Visakhapatanam at Southeast India. In addition to using remote sensing image data for interpretation, the topic also used the Markov

sequence to predict areas that may be convulsed in the future.

In the research “Monitoring Land Use Change by Multi-temporal Landsat Remote Sensing Imagery” (Tayyebi, et al., 2008), the author group used multi-time landsat image to assess the urban land convulsion in the past (the period of 1980-2000) to

make interpretations for the future (year 2020).

In the topic “Analyzing Land use/ Land Cover change Using Remote Sensing andGIS in Rize, North-East Turkey” (Selcuk Reis, 2008), the author established a landuse / land cover changes map at Rize, Northeast Turkey with 7 types of land cover.The data the author used in this topic is the image of Landsat MSS (1976) andLandsat ETM+(2000) with the respective resolution of 79m and 30m. However inthis topic, the author did not explain clearly the method but only focused on theassessment, statistics of convulsion with profound changes to the agricultural, urban,grass and forestry land, areas near the sea with low slope.

</div><span class="text_page_counter">Trang 14</span><div class="page_container" data-page="14">

In the topic “Land use/Land Cover Mapping and Change Detection in part of Eastern Ghats of Tamil Nadu using Remote Sensing and GIS” (S. JAYAKUMAR AND D.I.

AROCKIASAMY, 2003), the author used the software ERDAS IMAGINE 8.3 to

process the satellite image and establish a land cover and land use map of 1990 and 1999. The analysis of land use/ land cover changes is done based on the correlative

matrix in the Erdas Imagine software.

In the research “Land cover change detection using GIS and remote sensing techniques: A sapatio-temporal study on Tanguar Haor, Sunamganj, Bangkadesh”

(Md. Inzamul Haque,Rony Basak, 2017). This research uses satellite data in the past

and in recent times to assess the change of typical scenery in various decades. Both methods of prior and post classification are used to assess the result of change from 1980 to 2010. In the prior classification method the NDVI, CVA, NDWI indexes are used to assess the script of change. The supervised classification technique is used to create unique land cover of the surface of the earth.

In the research “Monitoring land use/cover change using remote sensing and GIS

techniques: A case study of Hawalbagh block, district Almora, Uttarakhand, India” (J.S.Rawat, Manish Kumar, 2015). This research demonstrates clearly the strong change in space and time of land cover and land use at the Hawalbagh area of Almora province, Uttarakhand, India. The Landsat satellite image of two different time periods, i.e. Landsat Thematic Mapper (TM) years 1990 and 2010 were collected at the GLCF webpage. The supervised classification method is maximally used in the ERDAS 9.3 software. Images of the research area are divided in 5 different classes which are land cover, agriculture, vacant land, construction and

In the research “Monitoring urban growth and land use change detection with GIS and remote sensing techniques in Daqahlia governorate Egypt” (Ibrahim Rizk

Hegazy, Mosbeh Rash Kaloop, 2015). In this research the author uses the image

source from the USGS Earth Explorer webpage. Image processing is done in the ERDAS software. Images are geometrically corrected, adjusted and removed of clouds. These data are stratified into regions, where the land cover has similar

spectral properties. Processed images shall be classified by both methods: supervised classification and unsupervised classification method. In the unsupervised classification method, the use of the ISODATA clustering algorithm built in ERDAS

shall classify by the necessary amount of classes and the amount of available pixels.

4

</div><span class="text_page_counter">Trang 15</span><div class="page_container" data-page="15">

In the supervised classification method, the use of the Maximum likelihood algorithm shall classify based on the sample set provided by the user based on their field knowle

shall be used for reference and understanding on the distribution of pixels. The “The classification result by the unsupervised classification method classification result divided into four classes of land cover including agricultural land, construction area, arid land and water areas determined in the research area, In the research of Li etal (2003) “Research of the change of cover in the Tarim basin ‘of China in the period of 1964 and 2000”, The author has used Landsat ETM image for the year 2000 Corona Panchromatic image for the year 1964. Using the method ‘of post-classification convulsion discovery, the author noticed a change in the area of reclaimed land from water and soil, the death of an old primeval forest surrounding ‘Tarim River and also the change in the salinity

In the research “Analysis of Urban Sprawl Pattern in Tiruchirappalli City Using Applications of Remote Sensing and GIS” (Nisha Radhakrishnan Satish Kumar Eerni -Sathees Kumar, 2013). The author used ASTER image with high resolution, collected from the national remote sensing center. Images after processing shall go through land cover classification via the ERDAS IMAGINE 9.1 software. In this research the author used the supervised classification method, and used the ‘Maximum likelihood Class

types of land cover which are construction soil, vegetation, moist soil, vacant land determined in the research area,

+ algorithm, The classification result divided into S

1.2. Overall study on applications of Remote Sensing and GIS in mapping land

cover changes in Vietnam

‘Currently there have been many research works on the convulsion of land cover with many different viewpoints, of which there are works that focus on argumentative research analysis, while others focus on methods of finding convulsion and there are works that combine both: convulsion discovery technique, result assessment and argument supplementation,

‘The topic “Research of convulsion of several forms of land use in the peri-urban areas of Tu Liem district Hanoi city on the basis of remote sensing and GIS technology application” of author Nguyen Thi Thuy Hang has solved problems such as extraction of information on convulsion of land use from multispectral and multi

and time remote sensing data through several methods of digital image analysis

</div><span class="text_page_counter">Trang 16</span><div class="page_container" data-page="16">

processing, integrating remote sensing data analysis results with other data to assess the correlation between convulsion of land use and socioeconomic phenomena.

‘The author group Nhu Thi Xuan, Dinh Thi Bao Hoa, Nguyen Thi Thuy Hang with the article “Assessment of land use convulsion of Thanh Tri district, Hanoi city in the period of 1994 - 2003 on the basis of remote sensing method combined with GIS” has analyzed, assessed the convulsion of land use in the area of Thanh Tri. This is also one of the areas largely affected by the urbanization,

In addition to using remote sensing data in the research of convulsion, author Hoang ‘Thi Thanh Huong in the topic “Research of convulsion of land use in Long Bien distr . Hanoi city during urbanization” has combined remote sensing material with the spatial analysis of the geo information system. The topic pilots the new classification method of classification by subject, method of performance on remote sensing data with very high resolution (VHR). In addition, spatial analysis is used in GIS to compare classification results with the socioeconomic data to see the interaction between them. The result shows that the remot snsing image with high spatial resolution can meet the requirements of accuracy of urban areas with fragmentation as in Vietnam.

In the topic “Establishment of vegetation map on the basis of remote sensing image analysis, processing” at the area of Tua Chua ~ Lai Chau (Hoang Xuan Thanh, 2006), the author used the supervised classification method for the Landsat image data of 2006 to classify 7 different vegetation classes with the Kapa index =0:7

In the topic “Application of remote sensing in monitoring the urban land convulsion

of Vinh city, Nghe An province” (Nguyen Ngọc Phi, 2009) the most approximate

classification method was used to divide into 5 classes of subject. The most noteworthy point of this topic is the combined use of various remote sensing images such as Landsat (1992, 2000) and SPOT (2005) to bring forth the interpreted results, while also have a comparison on the accuracy, detail between the image types. With a Kappa index of =0,9, the SPOT image data has the post-classification accuracy higher than compared to Landsat (Kapa~0.7),

In the research “Application of remote sensing and GIS in establishing the land cover map of the area of Chan May, Phu Loc district, Thua Thien Hue province” (Nguyen Huy Anh, Dinh Thanh Kien, 2012), the author has used the most approximate classification method for the Landsat TM image data at a resolution of

</div><span class="text_page_counter">Trang 17</span><div class="page_container" data-page="17">

10m, combined with field sampling to divide into 13 types of cover with relatively high accuracy

In the research “Usage of MODIS satellite image material in research of crop season, establishment of status quo and convulsion map of the Red River delta cover in the period of 2008 -2010" (Vu Huu Long, Pham Khanh Chi, Tran Hung, 2011), the author used supervised classification with the most approximate classification algorithm. The topic classfied 9 types of cover with Kapa index ~0,9. To assess the accuracy, the author used a combination of the survey sample data, field investigation and the status quo map of Land use of the most recent year

In the topic “Research of the impact of shifting agricultural land to non-agricultural land in the vicinity of Hue city, period of 2006 - 2010” (Nguyen Thi Phuong Anh et al, 2012), the author assessed the impact of the shift of agricultural land to non-agricultural land on the economic structure, social life and bring out viable solutions,

with the pilot research area being Kim Long ward. In this topic, the author only used

methods of synth

perform research. 1

is, analysis, comparison, contrasting, and statistics of data to we data is extracted via tables, without visual output by system of

In the topic “Research of change of forest vegetation at the Bach Ma national garden, ‘Thua Thien Hue province” (Dang Ngoc Quoc Hung, 2009), the author built forest vegetation maps of the years 1989, 2001, 2004, 2007 by supervised method of

remote se ing image interpretation to extract information from satellite images Erdas software was used to interpret satellite images. The change of the forest

vegetation in the periods 1989-2001, 2001-2004, 2004-2007 was analyzed and

assessed by method of stacking and analysis by Arcview 3.2 software,

he application of remote sensi in monitoring the status of land use, although popular in the world, is still not widely applied in Vietnam. This may show that, the capability of remote sensing in status monitoring is very good but the performance of this task is still difficult, especially for small areas,

1.3. Approach of the research.

In this research, the author used the method of remote sensing to extract information

of land cover on the surface of the research area, combined with field survey and ‘other materials determining the status of land use

</div><span class="text_page_counter">Trang 18</span><div class="page_container" data-page="18">

_—x". na.

Tad cover map THỂ cover map

<small>and cover changemap 1999-2015,</small>

<small>Fvaluaiag theresus of and coverng</small>

Figure I. 1. Overview of the research

‘The GIS geographical information system is used to analyze, assess the convulsion of land resources.

‘The processing of remote sensing image for analysis of land cover for the research

area adhers to the following procedure

‘The Landsat 7 remote sensing image data taken in 1999 and Landsat 8 taken in 2015

were used to analyze and establish the land cover map of 2 periods. The methods of remote sensing image analysis are very diverse. Some of the methods of image analysis can be Tisted such as manual threshold, unsupervised classification, supervised classification, Fuzzy classification or Mixing models but the two popular methods for classifying land cover currently is the unsupervised classification and supervised classification.

</div><span class="text_page_counter">Trang 19</span><div class="page_container" data-page="19">

Each method of classification uses certain algorithms. The algorithms have limitation and application in different situations (Shrestha and Alfted, 2001). The

popularly used algorithms are the Minimum Distance, Parallelepiped and Maximum

Likelihood (Richards, 1994). Among these, the Maximum Likelihoood algorithm is used the most by classifiers in works of land cover research (Keuchel et al, 2003 Shrestha and Alfred, 2001; Swain and Davis, 1978; Este etal, 1983; Schowengerdi,

1983; Sabins, 1986; Lillesand and Kiefer, 2000; Jensen, 1996). The Minimum Distance algorithm is often applied in the unsupervised classification method, while the two algorithms Maximum Likelihood and Parallelepiped are usually applied in

the supervised classification method. In addition, people also use several methods to

highlight the vegetation factor such as the vegetation index analysis method ~ NDVI

In this research, the author uses the supervised classification method and the

Maximum Likelihood algorithm to analyze the land cover of 2 periods on the research area,

“The calculation of land cover changes of two periods is done by the CROSSTAB tool (Cross-Classification) in the IDRISI 17 software. This method is the popular

9

</div><span class="text_page_counter">Trang 20</span><div class="page_container" data-page="20">

approach to calculation of land cover changes in land cover researches around the world,

1.3.1. The remote sensing is used for monitoring land cover changes

Remote sensing is a science of technology that helps to identify, measure or analyze the properties of objects or phenomena from a distance without direct contact with the abject.

‘These natural objects absorb and reflect electromagnetic waves with intensity and in

different ways, known as spectral characteristics. These characteristics conta important information that allows the grouping of natural formations into objects of the same spectral reflectance. This is useful for the interpretation of satellite imagery so the spectral reflectance characteristics of natural objects play a very important role in exploiting and applying effectively the collected information,

Spectral reflectance properties of natural objects depend on many factors such as lighting conditions, aunospheric environment and object surface, especially the objects themselves (moisture, surface roughness, vegetation, humus, surface structure, ete.). Different objects will have ability to reflect different spectra, with ceach object, the reflecting and re-absorbing varies by wavelength. The remote sensing method relies primarily on this principle to identify and detect objects and

phenomena in nature, The information about the spectral characteristics of natural

objects will help professionals select image processing methods to get optimal

channel, which contains a lot of information about the subject of study, this is the

basis to analyze and research of the properties of the object and proceed to classify them, The following is spectral reflective characteristics of some of the main natural

</div><span class="text_page_counter">Trang 21</span><div class="page_container" data-page="21">

Solar radiation on the leaf surface, in the red and blue regions is absorbed by the chlorophyll for photosynthesis, in the green and infrared regions will be reflected when exposed too much chlorophyll of leaves (when plants are healthy), When plants are weak, chlorophyll decreases, the ability to reflect the red wave predominates, so leaves are yellow (green - red combination) or red in cold weather. ‘The difference in spectral reflective characteristics of plants depends on the internal and external constituents of the plant (chlorophyll content, the structure of the dermis, the composition and structure of the epidermis, the morphology of the

leaves, efe.), growth period (tree age, growth stage, etc.) and external influences,

(growth conditions, lighting conditions, weather, geographical location, etc.) However, the spectral reflective characteristics of the vegetation cover still have the following common features: Reflected near the near-hip zone (2> 0.720qim), strongly absorbed in the red zone (1 = 0.680 - 0.270 jum).

Reflective spectral characteristics of water

‘The spectral reflectance of water also varies by the wavelength of the incoming radiation and the composition of materials in water. Water only reflects strongly in the blue wave zone and weakens when it reaches the green zones, and is destroyed at the end of the red strip. It also depends on the surface and state of the water.

Most of the sun's radiation energy is absorbed by water for the process of raising water temperature, The reflected energy of water consists of the energy reflected on the surface and the energy reflected afler scatterred with suspended materials in water. Therefore, the reflex energy of different types of water is very different,

especially clear water and turbid water. In general, the reflectivity of water is low

and decreases by increasing wavelength. Solar radiation is almost completely absorbed in infrared and near infrared waves. The turbid water is stronger than the clear water in relfection, especially in the red zone due to the scattering effects of suspended materials. The use of photographs in long wave channels allows us to interpret water objects. For example, the waterfronts will be easily interpreted on infrared and near infrared channel.

Reflective spectral characteristics of soil

‘The characteristic curve of the spectral reflectance of the terrestrial overlap gradually increases from the ullraviolet zone to infrared zone in a monotonous manner, with Tess obvious maxima and minima, The main reason is that the elements of the soil

"

</div><span class="text_page_counter">Trang 22</span><div class="page_container" data-page="22">

are complex and not as clear as in plants. Spectral reflectance depends mainly on the physical and chemical nature of the soil, organic content, humidity, structure (sand, powder and clay fraction), state, surface and mechanical components of the soil. This causes the curve to fluctuate around an average value. However, the general rule is that the spectral value of the soil increases toward the longer wavelength

Reflective spectral characteristics of rock

Rock has mass and dry structure with the same spectral reflectance curve similar to soil but the absolute value is usually higher. However, as for soil, the fluctuation of the reflectance value depends on many factors of the rock: water level, structure, ‘composition, mineral composition, surface condition, ete

In short, reflection spectrum is the most important information that remote sensing has obtained about the objects. Based on the reflection spectral characteristics (intensity, curvature in different wave bands), itis possible to analyze, compare and identify objects on the surface. Specal information is the first information, which is the premise for image analysis methods in remote sensing, pecially digital

Different objects within the same group of objects will have the same general spectral-reflection curve type, but they will be different in terms of small details on the curve or varying in magnitude of reflection value, When the nature of the object ‘changes, the spectral curve is also varied. Based on these characteristics, there are signs of interpretation for objects on satellite images as follows

<small>- Photo elements:</small>

+ Brightness (fone): The sum of light reflected by the object's surface, which is a very important interpretative sign for object identification.

+ Texture: texture is interpreted as the number of repetitions of tone changes caused by the aggregation of many distinct characteristics of individuals. For example, the fine structure typical of loose sediment, the coarse structure typical of magmatic rocks, the strip structure characteristic of metamorphic sedimentary rocks. Accordingly, it is possible to distinguish between different rocks and their relative

+ Pattern: A very important element that represents the arrangement of objects according to a certain law in space. A particular type of terrain would inelude the

</div><span class="text_page_counter">Trang 23</span><div class="page_container" data-page="23">

arrangement in accordance with a specific rule of natural objects which are ‘components of such terrain forms, For example, concentrated urban areas of houses, streets, and trees that make up a distinctive pattern of urban structure, Rice paddies have distinctive patterns that differ from orchards,

+ Shape: The external characteri typical for each object, the external image of the object. Shape is the first factor that helps the analysts distinguish between different objects, For example, the horseshoe lake is a narrow river, bright broom type is the coastal sand dunes.

+ Size: The size of an object is determined by the ratio of imagery and size measured ‘on the image. Based on this information we can also di inguish the objects on the

+ Shadow: the obscured part, no sunlight or light from active source, so there is no light reflected back to the receiver. Shadows are usually represented by black tone on black and white and dark to bla

structure of objects. However, the shadow is also the part where the information

on color image. The shadow can reflect on the tht of the object. Shadow is an important element that creates the characteristic about the object is not available or very little, so the amount of information needs to bè added in this zone.

+ Site: site is also a very important clement to distinguish objects, With the s sign, but in different sites it may be different objects. For example, the mudflats ‘cannot be found on the mountain slope, although some of the characteristics of the image look very much li is sign. The mudflats are distributed only on both sides

Of streams and rivers and are light color, On the mountain slopes the bright patches

are the discharge cones, sliding zones or shifting cultivation areas.

+ Color: Color of the object on the fake color combination (FCC) allows the

interpreter to distinguish many objects with similar color tone characte

black and white images. The common color combinations used in LANDSAT and

ties on

SPOT images are blue, green and red representing the basic elements of pink to reddish-green plants, light to dark green water, dark red to dark brown mangrove forests, pink to yellow bare land with winter crops, ete.

In addition to the above three fake-color combinations, we can create many other different fake color combinations using optical method (using color filters) or digital image processing techniques. Therefore, when interpreting the objects on fake color

B

</div><span class="text_page_counter">Trang 24</span><div class="page_container" data-page="24">

images, there must be early orientations on fake color combinations, thus avoiding confusion.

- Geotechnical elements

Elements of topography, vegetation, land use status, river network, cracks system and linear elements, combinatorial of interpretation elements, Following are some <small>interpretive signs:</small>

Table 1. 1 Some interpretation signs on the fake color combination of SPOT satellite images

No. Object Sign

1 | Cities Blue (dark or light, even white depending on the density of the building works)

‘Traffic (road, Dark blue or grey.

S5 | Perennials Red — pink

Source: Snidy on rational use of suburban land in Thanh Tri district, Hanoi with the support of remote sensing and geographic information system - Dinh Thi Bao Hoa, 2007.

So in visual judgments, itis needed to capture and distinguish between interpretative signs, the work requires the interpreter to have a good knowledge in order to

properly combine knowledge in image interpretation and only then accurate results

can be released.

1.3.2 GIS spatial analysis is used for evaluation of land cover changes ‘Ther © are many concepts about GIS but generally they are in two directions:

<small>= The concept of GIS as a map database controlled by computer graphics techniques</small> with the functions of importing, organizing, displaying, querying the map information stored in the database.

</div><span class="text_page_counter">Trang 25</span><div class="page_container" data-page="25">

- The concept of GIS as a geographic information system including functions of input, analysis and display, and able to modelization the information layers ‘organized in a database to create thematic maps.

Regardless of GIS's concepts, it both has to meet the requirements of a four-part information system:

1, Computers and peripherals capable of performing software input, output and

processing functions.

2. A software capable of entering, storing, adjusting, updating and organizing spatial information and attribute information, analyzing, transforming information in a database, displaying and presenting information in different forms, with different 3. With a database containing spatial information (geographic information) and attribute information, organized according to a specific specialized intention

4. User ‘with specialized expert knowledge.

In the land cover change study, GIS plays an important role in gathering and analyzing database. It is aimed to synthesize, systematize and unify the data sources for monitoring and evaluating land cover variation.

“The strength of a GIS is expressed through spatial analysis. Spatial analysis is often used to generate additional geographic information using existing information or developing spatial structures or relationships between geographic information. In the

ground cover variation analysis, we often use the following techniques:

Creating additional geographic information overlapping the data layer or creating a

buffer pool: Overlay is a common technique in spatial analysis. Multiple data layers

are overlapped by an algebraic or logical operation to obtain a new data, Buffering is

determining the area within a certain radius as compared to an object or point.

‘Typically, the length of the buffer zone radius is determined by the effect of the point ‘or path to the surrounding area,

Linking Technique: Linking multiple spatial analysis techniques together to get the results needed.

In addition, to search for data that satisfies a set condition, the author also use the spatial query function of GIS, There are two types of spatial query that are attribute data query (it means to find a spatial distribution or a region that satisfies certain

1s

</div><span class="text_page_counter">Trang 26</span><div class="page_container" data-page="26">

attribute conditions) and geographic data query (search all data satisfying a given ‘geographical condition such as position, shape or intersection, etc.)

Searching and analyzing data depend on the ability to link these two data types. The ‘greater the linking capacity, the better the search and analysis of data. Users can access table data through the map or can create maps through the table data, To access and display this data, the computer needs to store both tabular and graphical data in an organized and searchable format

</div><span class="text_page_counter">Trang 27</span><div class="page_container" data-page="27">

CHAPTER II

RESEARCH ON THE APPLICATION OF REMOSTE SENSING AND GIS IN ESTABLISHMENT OF LAND COVER CHANGE MAP

OF SON LA PROVINCE

2.1, Overview of study area

2.1.1. Natural and socio-economic conditions

Son La is a mountainous province in Northwest Vietnam, with an area of 14,12 ket, accounting for 4.27% of Vietnam's total area, ranking third among 63 provinces

and cities. Geographic coordinates: 20°39 '- 2202 North latitude and 10311 "=

10502" East longitude. The border: bordered Yen Bai, Dien Bien, Lai Chau in the

north; Phu Tho and Hoa Binh to the east; bordered Dien Bien province in the west bordered Thanh Hoa province and Huaphanh province (Laos) in the south; and

Luangprabang Province (Laos) in the southwest, Son La has a national boundary of

250 km, the border with other provinces is 628 km. The province has 12

administrative units (I city, 11 districts) with 12 ethnic groups.

Son La has subtropical moist montane climate, cold dry non-tropical winter, hot and humid summer, heavy rain, Due to the deep and strong geographical divided terrain, many sub-climates are formed, allowing for the development of a rich agro-forestry

production, Moe Chau plateau is suitable for temperate plants and animals.The area

along the Da River is suitable for tropical evergreen forests, Statistics show that the

annual average temperature of Son La tends to increase over the past 20 years with

an increase of 0.5 ° C - 06° C, the average annual temperature in Son La is at 21.1 ° ©, Yen Chau 23 ° C; Average annual rainfall tends to decrease (city is currently at 1,402 mm, Moc Chau 1,563 mm); Average annual humidity also declines. The <rought in winter, the hot and dry west wind in the last months of the dry season and at the beginning of rainy season (March - April), are the factors affecting the agricultural production of the province. Salt fog, hail, flash floods are unfavorable

Son La is still one of the poor provinces of the country. There are 11 districts and 1 city in the whole province, in which there are 5 poor districts under the Government's program for Rapid and Sustainable Poverty Reduction (Program 30) In recent years, the introduction of Son La hydropower projects has contributed significantly to the overall change of infrastructure, creating a motive force for Son

0

</div><span class="text_page_counter">Trang 28</span><div class="page_container" data-page="28">

La province to develop. In mountainous districts, agriculture is still their strength With the propaganda of knowledge, farmers have boldly renewed crops and cultivation methods, bringing productivity increased 4-5 times.

2.1.2. Characteristics of land cover in the study area

According to the results of collecting and compiling documents of the Northern Geological Mapping Division in 2013, the forest distribution in Son La province accounts for 45.49% of the provincial area in 2011, while the area of bare soil accounts for 33.78%. Specifically, the area of Son La province has the following

types of vegetation cover:

- Special-use forest - protection forest: occupying 15% of Son La province. forest type with green state all year round, leaves are often broad-leaf, small to medium leaf size. The soil in the Forest is usually moist, thick, yellow or brown, Due to the high deforestation at present, this forest type is mainly secondary forest, the soil is strong degraded, secondary vegetation is no longer wood but converted into bamboo forest.

- Plantation forest - zonation forest (production forest): occupying 22.59% of the

area of Son La province, is a forest type with many layers, thick trees, closed ‘canopy, the highest plant only 20-30m. Few additional trees and parasites. The soil is thin and dry. Some areas are further degraded, becoming deciduous for vegetation cover is bush with sim, mua, sam and finally hay.

- Shrub - grassland: occupying 8% of the (otal area of Son La province, including mangroves, scrublands and grasslands. In this type of forest, thick and scattered

woods are degraded by human impact.

Table 2. 1 Distribution of vegetation cover and current land use status in 20111

No, | TyPe of land cover Distributed area Area proportion

</div><span class="text_page_counter">Trang 29</span><div class="page_container" data-page="29">

No, | T9Pe of lind cover Distributed area Area proportion <small>cover (km2) 6)</small> 5 [River-take 36612494

6 [Bare soil 46279892 33.78)

- Rice - crops: occupying 15.8% of the area of Son La province, this is a human ecological system created and maintained by humans for food and other essential

products for the their life. The characteristics of the agro-ecological system is it

usually has only one species, intercropping a maximum of 2-3 species (maize-beans, sh, long-term industrial crops, vegetables) leading unbalanced nutrient use in

the soil, In case of extensive cultivation, the productivity is not high compared to the productivity of natural ecological systems. Morevover, this ecosystem is very fragile, vulnerable to natural disasters, If intensive, it will degrade the environment, cause soil erosion and fading.

- Bare soil, rivers and lakes, and residential areas: accounting for about 38.6% of the total area of Son La province. In which vacant and bare land is too large, so it is

necessary to plan investment in afforestation, growing fruit trees and industrial plants

in a reasonable manner, contributing to minimize the damage caused by the

2.2, Data collection and analysis 2.2.1. Remote sensing data

Within the scope of the study, the author has collected Landsat 7 and Landsat 8 image data for the study area from 1992 to 2015, with a resolution of 30m taken

from hup://earthexplorer.usgs.gov , Path | Row: 127 / 45,127 / 46, 128/45,

128/46, The details are shown in Table 3

</div><span class="text_page_counter">Trang 30</span><div class="page_container" data-page="30">

Table 2. 2 Remote sensing image data

Province Image type Seenes| Date | resotution | ©"_cloud

Son La province is covered by four image scenes: PI28R45, PI28R46, P]27R45 and PI26R46. But the scene of P128R45 occupies most of the area (> 90%), so when 4 Mosaics are mixed together, the Landsat8 and Landsat7 images of Son La province ‘will be taken time of fly shooting as time of P128 / R45 scene.

20

</div><span class="text_page_counter">Trang 31</span><div class="page_container" data-page="31">

igure 2. 1 Scene P128-R45 covers 90% area of Son La province 2.2.2. Land use status

Son La province has a total area of 1.405.500 ha. The results of the survey on the soil mapping in scale of 1 / 100,000 of Son La province show that:

Son La is a mountainous province with strong divided terrain, there are 85% (about

1.2 million ha of land) with slope above 25°, low vegetation cover, adverse weather

climate features, much rain, seasonal focus. Most of the area of annual trees are cultivated on sloping land, strongly eroded soil. This is a pressing problem in the use such as water of land, damaging land ssources and affecting other resoure:

resources, vegetation and ecological environment,

21

</div><span class="text_page_counter">Trang 32</span><div class="page_container" data-page="32">

Table 2. 3 Land use status of Son La province in 2005 as follows:

No. ‘Type of land Area(ha) Strueture(%)

Total natural area 1.405.500 100,00 1__| Agricultural land group 672.485 4185

1 | Land for agriculture produetion 191828 28,53

11 | Annual crop land 158.855 82,81

LI1 | Land of rice fields 15.430 971

112 | Land of kaingin 123.084 7148

1.1.3. | Land for planting other annual erops 20441 1281

1.2 _| Mixed garden soil 6933 3.61 13 _| Land for perennial crops 2.584 177 1.4 _| Land for grass planting for livestock 1832 0.96 15 _| Land with water surface for aquaculture 1624 085 2 |Forestryland 480.657 7174 2.1 | Land with natural forest 439.592 91.46 2.1.1 | Forest land for production 50.297 nas 2.1.2 | Protective forest land 322.149 73.28 2.1.3. | Special-use forest land 67.146 1528 2.2 _| Land with planted forest 41.087 854 2.2.1. | Forest land for production 8364 2038 2.2.2 | Protective forest land 32646 79/53 2.2.3. | Special-use forest land 37 009 2.3 _| Land for nursing seedlings 18 0

IL | Non-agricultural land group 30.276 215

1 _| Residential land 6.033, 19.93 Lt | Rural land 438 7.30

2

</div>