Assessing the effects of climate change on forest cover in dai tu district thai nguyen province

KUA TOPIC TITLE: ASSESSING THE EFFECTS OF CLIMATE CHANGE ON FOREST COVER IN DAI TU DISTRICT, THAI NGUYEN PROVINCE BACHELOR THESIS REMOTE SENSING GLOBAL VARIATIONS: EFFECTS OF CLIMAT

1

THAI NGUYEN UNIVERSITY

UNIVERSITY OF AGRICULTURE AND FORESTRY

KENNETH JOSHUA ZARATE KUA

KENNETH JOSHUA Z KUA

KENNETH JOSHUA Z KUA

KENNETH JOSHUA Z KUA

TOPIC TITLE:

ASSESSING THE EFFECTS OF CLIMATE CHANGE ON FOREST COVER IN DAI TU DISTRICT, THAI NGUYEN PROVINCE

BACHELOR THESIS

REMOTE SENSING GLOBAL VARIATIONS: EFFECTS OF

CLIMATE CHANGE PARAMETERS ON FOREST COVER AND VEGETATION IN DAI TU DISTRICT, THAI NGUYEN PROVINCE

REMOTE SENSING GLOBAL VARIATIONS: EFFECTS OF

CLIMATE CHANGE PARAMETERS ON FOREST COVER AND VEGETATION IN DAI TU DISTRICT, THAI NGUYEN PROVINCE

Study Mode: Full-time

Major: Environmental Science and Management

Faculty: International Programs Office

Thai Nguyen, 20/11/2017

ii

DOCUMENTATION PAGE WITH ABSTRACT

Thai Nguyen University of Agriculture and Forestry

Degree Program Bachelor of Environmental Science and Management

Student name Kenneth Joshua Zarate Kua

Student ID DTN1454290056

Thesis Title Assessing the Effects of Climate Change on Forest Cover in

Dai Tu District, Thai Nguyen Province

Supervisor Th.S Nguyễn Văn Hiểu

List of Figures 1 List of Tables (if necessary) 2 List of Abbreviations 3

1.1 Research rationale 4 The unpredictable and changing environment awdawdawdawdawdawdawdawdawdawdawdawdawdawdawdadddddddawdawdaawdadwawdawdawdawdawd been a serious topic all around the world, drawing the interests of intellectual humans to investigate its influence in different aspects

(Ravindranath 2008, p 1) The effects of climate change are predominated by rising temperatures, varying precipitation patterns and sea level increase, these impacts are capable to disturb different kinds of ecosystems and worst, damaging natural resources (such as forests, fertile lands and minerals) The inevitable losses of natural resources are most likely threat

of this thesis is to assess the effects of climate change on forest cover in Dai Tu district, Thai Nguyen province Landsat 5 TM images of 10th June 1993 and 10thJune 2004, and Landsat 8 OLI image of 6th June 2017 of Dai Tu district were utilized for supervised classification by using ArcGIS software Cross-tabulation change matrices were established to assess the land-cover changes for the 1st period (1993 – 2004) and the 2nd period (2004 – 2017) The results from the land-cover change analysis showed that, from the first period, the forest cover had decreased

by 10.43% of the study area While, the second period had decreased by 12.53% of

the study area These changes were a byproduct from the expanding agricultural areas and some human interventions (such as urbanization and mining activities) that resulted to deforestation Moreover, regression analysis was performed to investigate the relationships between the mean values of vegetation indices (NDVI and FAPAR) and climate change parameters (SMI and LST) including the forest cover data that were extracted from the land-cover classification The result of the analysis proves that, climate change parameters have significant relationships to the changing forest cover (r² = < 0.80) of Dai Tu district

Keywords: climate change; forest cover; remote sensing; Landsat;

ACKNOWLEDGEMENT

Firstly, I humbly acknowledging my God, "Jesus Christ", who is the “Son of God” that I believe in Without His constant provision of love and grace, I might not have had the positive outlook to keep and press toward especially while working on with my thesis

I am using this opportunity to consider everyone who supported me throughout my life and academics I may not include you all here, but I would like to say, “thank you very much!”

This piece of work couldn’t be possible without the help and support of some dedicated and considerate people:

I'd like to show my sincere gratitude and appreciation to my thesis supervisor Dr Nguyễn Văn Hiểu for offering his research center for me to work on Also for the immense support and valuable recommendations

I am acknowledging the Advanced Education Program (AEP) of Thai Nguyen University of Agriculture and Forestry (TUAF) and staffs for building, teaching, encouraging and inspiring me throughout my college life, which helped me to have a brighter future

Many thanks to my good friends (Anne, Katleen, Ekang, Tina, Carlo, Colleene, Jelo, Real, Nicole, Anh Kiet, and Kuya Jose) for the positive vibes that helped me a lot emotionally during the majority of my tiring days

I greatly appreciate the members of GeoInformatic Research Center (GIRC) for the cares and concerns, which made me feel comfortable and special while doing my research

I am deeply fascinated to mention my beloved brothers and sisters in Jesus Christ the Refiner’s Fire (JCRF) church and the Refiner’s Christian School (RCS) Thank you for all, without you, I might not have achieved a higher purpose

Words can’t express my deepest thankfulness to Mishel Rañada, for the unceasing support and compliments that boost me to do my best Many thanks, Mishel, for the insights, which you have shared for the betterment of my thesis

I am grateful beyond reasonable doubt and willingly dedicating this thesis to my family (Mommy Vec, Daddy Bong, Kuya Kien, Kezia Baby, Ate April, Tita Cherry, Tita Ester, Tito Eddie, Tito Edison, Tita Lau, Tita Leoni, Lola Paking) for the support not merely financial but also in lots of different aspects

The Researcher, Kenneth Joshua Zarate Kua

TABLE OF CONTENTS

List of Figures 1

List of Tables 2

List of Abbreviations 3

PART I INTRODUCTION 5

1.1 Research Rationale 5

1.2 Research Objectives 8

1.2.1 Main Objective 8

1.2.1 Specific Objectives 8

1.3 Research Questions and Hypothesis 9

1.4 Scope and Limitations 10

1.5 Definition of Terms 11

PART II LITERATURE REVIEW 16

2.1 Land-Use and Land-Cover (LULC) 16

2.2 Land-use research studies 17

2.3 Remote sensing and GIS techniques for LULC change 18

2.4 Forest vegetation monitoring using RS and GIS techniques 19

2.5 Remote sensing climate change effects on forest vegetation 20

PART III METHODOLOGY 23

3.1 Materials 23

3.1.1 Time and place of research 23

3.1.2 Remotely sensed study area 23

3.1.3 Software used 23

3.1.4 Satellite data used 23

3.2 Methods 25

3.2.1 Satellite image pre-processing 25

3.2.2 Supervised classification 26

3.2.3 Accuracy assessment 26

3.2.4 Change rate analysis 27

3.2.5 Vegetation indices and climate change parameters 27

3.2.6 Establishing relationship 29

PART IV RESULTS 30

4.1 Study area 30

4.1.1 Geography 30

4.1.2 Topography 31

4.1.3 Hydrology 31

4.1.4 Climate and weather 31

4.1.5 Socio-economic activities 32

4.1.6 Population 32

4.2 Land-cover analysis 33

4.2.1 Land-cover classes 34

4.2.2 Land-cover maps 34

4.3 Land-cover area proportion 35

4.4 Accuracy Assessment results 38

4.5 Land-cover change analysis 38

4.5.1 Land cover change cross-tabulation 38

4.5.2 Land-cover gain-loss 40

4.6 Visualization of vegetation indices and climate change parameters 41

4.6.1 NDVI maps 41

4.6.2 FAPAR maps 42

4.6.3 SMI maps 43

4.6.4 LST maps 44

4.7 Linear relationships 45

Part V DISCUSSIONS AND CONCLUSIONS 47

Part VI RECOMMENDATIONS 49

Part VII REFERENCES 49

APPENDIX A 57

APPENDIX B 58

APPENDIX C 59

APPENDIX D 60

APPENDIX E 61

APPENDIX E 62

APPENDIX F 63

APPENDIX G 64

LIST OF FIGURES Figure 1: The overall methodological framework for assessing the effects of climate

change on forest cover 25

Figure 2: Maps and locations for Dai Tu district, Thai Nguyen province, Vietnam 30

Figure 3: Land-cover classification maps for years 1993; 2004; and 2017 34

Figure 4: Illustrates the proportion of land-cover classes by area (km²) and percentage (%), in year 1993 35

Figure 5: Illustrates the proportion of land-cover classes by area (km²) and percentage (%), in year 2004 36

Figure 6: Illustrates the proportion of land-cover classes by area (km²) and percentage (%), in year 2017 36

Figure 7: Comparison of land-cover proportion by percentage (%) years 1993; 2004; and 2017 37

Figure 8: Land-cover gain – loss in km² for the 1st period (1993 – 2004) and 2nd period (2004 – 2017) 40

Figure 9: NDVI maps of Dai Tu district in years 1993; 2004; and 2017 41

Figure 10: FAPAR maps of Dai Tu district in years 1993; 2004; and 2017 42

Figure 11: SMI maps of Dai Tu district in years 1993; 2004; and 2017 43

Figure 12: LST maps of Dai Tu district in years 1993; 2004; and 2017 44

Figure 13: Graphical relationship between (a) FC and SMI, (b) FC and LST, (c) NDVI and SMI, (d) NDVI and LST, (e) FAPAR and SMI, (f) FAPAR and LST 46

LIST OF TABLES

Table 1 Details of the satellite data used in the study 24

Table 2 Illustrates the characteristics of Landsat bands that were used for calculating vegetation indices and climate change parameters 28

Table 3 Land-cover classes definitions and the criteria used to identify classes 33

Table 4: Land-cover classes conversion in area (km²) from 1993 – 2004 period 38

Table 5: Land-cover classes conversion in area (km²) from 2004 – 2017 period 39

Table 6: Statistical relationship between vegetation indices and climate change parameters in Dai Tu district in years 1993; 2004; and 2017 45

LIST OF ABBREVIATIONS

AVHRR Advanced Very High-Resolution Radiometer

FAO Forest and Agriculture Organization

FAPAR Fraction of Absorbed Photosynthetically Active

Radiation

GIS Geographic Information System

MODIS Moderate Resolution Imaging Spectrometer

NDVI Normalized Difference Vegetation Index

NOAA National Oceanic and Atmospheric

Administration

REDD Reducing Emissions from Deforestation and

forest Degradation

RS Remote Sensing

SPOT Système Pour l'Observation de la Terre

UNFCCC United Nations Framework Convention on

Climate Change

USGS United States Geological Survey

PART I INTRODUCTION 1.1 Research Rationale

The unpredictable and changing environment has been a serious topic all around the world, drawing the interests of various scientists, citizens, and policymakers to investigate its influence on different aspects (Ravindranath and Ostwald, 2008) Shako (2015) reportedly demonstrated the climate change parameters, such as temperature, precipitation, rainfall, soil moisture, vegetation cover, sea level, sunshine hours, atmospheric pressure, wind velocity, etc Slight changes in these parameters affect each other directly or

indirectly (Palmate et al., 2014) These effects are capable to disturb different kinds of

ecosystems and worst, damaging natural resources (e.g forests, fertile lands, minerals, etc.) The inevitable losses of natural resources are unquestionably a threat to human survival According to the United Nations Framework Convention on Climate Change (UNFCCC, 2006), demonstrates proven prediction of some catastrophic events of climate change, which are subsequent droughts and heavy rainfall conditions, decreased in the terrestrial forest, loss of biodiversity, food and water scarcity that can result in increased risk of hunger

Forest occupies one-third of the Earth’s surface and serves as an essential resource for Earth’s inhabitants Furthermore, Forests give protection for the natural disasters (e.g floods, landslides, tsunamis, etc.), preserve the quality of the soil, provide habitats for animals, increase the biodiversity, progress the economic growth (producing raw materials such as woods and medicines), and functions globally as a prevention for climate change

by means of lessening global warming through carbon sequestration (Baumann et al., 2014; Kim et al., 2014)

Unfortunately, according to Food and Agriculture Organization (FAO, 2012), forests have been continuously and rapidly depleting worldwide Recent studies claim that forest depletion has been a serious issue regarding global variations To prove that, recent

report from Chakravarty et al (2012), demonstrates that world forest cover lost from 1990

to 2000 was approximately 0.20% and from 2000 to 2010 was approximately 0.13% She also outlines that North and South Africa were leading countries that had the highest rates

of deforestation from 1990 to 2010 with average approximately to 0.62% - 0.66% Moreover, FAO has shown that since 1990, the total amount of forest that had been lost was equivalent to 129 million hectares, which are approximately the size of South Africa

It is widely known that deforestation described as clearing out massive Earth’s forests and potentially damages the quality of the land Deforestation has a lot of negative impacts on the environment and to the diverse ecosystems It is the primary cause of soil erosion that leads to loss of habitats for many species and sedimentation of water bodies

(Chakravarty et al., 2012)

For many years until now, degradation of the forest has been widespread due to human interventions (intentional) and natural factors (unintentional) (FAO, 2012) Human interventions to the forest comprise of expansion in agricultural area, urban development, commercial logging, illicit cutting, grazing, construction of dams/reservoirs and barrages,

etc (Torahi and Rai., 2011; Ghebrezgabher et al., 2014) On the other hand, natural factors

consist of climate change (e.g forest fires, hurricanes, and droughts), pests and diseases, etc Furthermore, eliminating trees in the forest can damage the forest canopy structure, which blocks the sun-rays and keeps the moisture of the soil The decrease in forest canopy can result to increase in heat that can be harmful to plants and animals and dry out the soil moisture content, which leads to deficiency in available water in the soil Consequently, it will be hard for trees to uptake water, which can result to wilting Former forests became

barren deserts because of this occurrence (Singh, 1989; Ghebrezgabher et al., 2014; Nyssen

et al., 2004)

Forest monitoring has increasingly become a vital factor in environmental planning FAO (2012) uses the term “National Forest Inventory” (NFI) as the collection of forest analyzed data including field measurements and remote sensing data It is also mentioned

as the thorough process of evaluating forest data for appropriate interpretation and preparation Countries that are members of Reducing Emissions from Deforestation and forest Degradation (REDD+) program are responsible to report their forest data, which are requirements for REDD+ reporting

Change detection is defined as a process of identifying and monitoring the differences in the state of an object or phenomenon by observing it at different times (Singh, 1989) Remote Sensing (RS) and Geographic Information System (GIS) techniques for monitoring forest cover are one of the most important tools due to the increasing population growth and human interventions to the forests Remote sensing research is increasingly becoming widespread due to the environmental issues (e.g climate, land and

forest change) that the Earth is facing The remarkable features of remote sensing include its fast ability to provide precise and useful data, broad range, capability to scope inaccessible areas, repetitive monitoring of dynamic changes, quick data processing using

software, etc (Singh, 1989; Ozdogan et al 2010; Polidori, 2011)

Dai Tu district (located about 100 km north of Hanoi) is a mountainous area covering 57,618 ha in the northwest of Thai Nguyen province Together with the lack of easily accessible and reliable data has shown the need for high-resolution remote sensing analysis for the region Therefore, the purpose of this paper is to extract and analyze the forest cover of Dai Tu district over the past three decades and establish a linear regression with vegetation indices and climate change parameters

1.2 Research Objectives

1.2.1 Main Objective

The primary objective of this study was to assess the effects of climate change on forest cover in Dai Tu district, Thai Nguyen province by using remote sensing and GIS techniques

1.2.2 Specific Objectives

The specific objectives of this study correspond to:

1 To Identify land-cover classes within Dai Tu district and their corresponding areas (km²) and spatial distribution

2 To prepare classified maps of Dai Tu district for years 1993, 2004, and 2017

3 To assess land-cover change during 1993 – 2004 (1st period) and 2004 – 2017 (2nd period)

4 To prepare maps for vegetation indices (NDVI and FAPAR) and climate change parameters (SMI and LST) in Dai Tu district for further investigation and visualization

5 To establish the relationship between forest cover, vegetation indices and climate change parameters

1.3 Research Questions and Hypotheses

This thesis is designed to address the following questions:

1 What are the land classes within Dai Tu district and their changes in areas (km²) during 1993 – 2004 (1st period) and 2004 – 2017 (2nd period)?

2 Does mining activities in Dai Tu district expanded?

3 Does expansion in agricultural areas had caused deforestation?

4 Is there a reduction or expansion of forest coverage in the study area within the given times?

5 Does GIS methods (integrated with this study) prove beyond reasonable doubt its capabilities of spatial analysis of the forest cover change?

6 Does remote sensing image manipulation applicable for locating, identifying and quantifying forest cover change?

7 Does climate change effect negatively on forest cover?

Alternative Hypothesis: Climate change parameters (independent variables) have

significant linear relationships between vegetation indices and forest cover data

(dependent variables) Therefore, r² is not equal to zero (r² ≠ 0)

Null Hypothesis: Climate change parameters (independent variables) don’t have

significant linear relationships between vegetation indices and forest cover data

(dependent variables) Therefore, r² is equal to zero (r² = 0)

1.4 Scope and Limitations

The main limitation of this research is the actual validation of the remote sensing data, field work is usually limited because of time, cost and difficulty in reaching some places This study only considered the use of remote sensing images and Google Earth software for analyzing changes This thesis also suffered from lack of clear clouds and haze for satellite images in the interested area In result, chosen years were limited due to some unfavorable disturbances Moreover, due to lack of fund to afford higher resolution images, Landsat series freely provided by the United States Geological Survey (USGS) satellites images had been used This research clearly consists of certain limitations, nevertheless, images without clouds and haze in the study area had been chosen to observe, which are Landsat 5 TM images of 10th June 1993 and 10th June 2004, and Landsat 8 OLI image of

6th June 2017 of Dai Tu district Landsat 5 and 8 provided by the USGS are high-resolution images with 30m spatial resolution, which are suitable for this investigation Furthermore, this study only analyzed Soil Moisture Index (SMI) and Land Surface Temperature (LST) for climate change parameters, because of lack or no available data for the study area

1.5 Definition of Terms

These following definitions were established for the purpose of clarification and further understanding of the given terms of the study

Land-cover is widely recognized as a remote sensing data, which can be examined of how

much area of land is covered by forests, wetlands, agriculture, impervious surfaces, water bodies and other land types

Land-use reflects how people use and interact with a certain land (e.g development,

recreational, conservation, agricultural, etc.)

Forest-cover consists of vegetation or tree cover more than 5 m in height with more than

two species, and the canopy or crown ranges from 10% to 40% for open forest and above 40% for closed forest

Change detection is defined as a process of identifying and monitoring the differences in

the state of an object or phenomenon by observing it at different times

Satellite image pre-processing is referred as an image restoration and rectification, which

is intended to correct for the sensor and platform specific radiometric and geometric distortions of data Satellite image pre-processing examples are geometric correction,

radiometric correction, atmospheric correction, topographic normalization, etc

Radiometric correction is an image pre-processing technique, which is necessary due to

variations in image illumination and viewing geometry, atmospheric conditions, and sensor noise and response

Supervised classification is a process of selecting sample pixels in an image that are

representative of specific classes and then apply the image processing software to use these sample pixels as references for the classification of all other pixels in the image

Maximum likelihood classifier is one of the most popular methods of classification in

remote sensing, in which a pixel with the maximum likelihood is classified into the corresponding class

Universal Transverse Mercator (UTM) is a conformal projection that uses a

2-dimensional Cartesian coordinate system to give locations on the surface of the Earth

WGS84 is an Earth-centered, Earth-fixed terrestrial reference system and geodetic datum

It is also based on a consistent set of constants and model parameters that describe the Earth's size, shape, and gravity and geomagnetic fields

Climate change parameters are key factors in measuring climate change, such as

temperature, precipitation and biomass

Accuracy assessment is known as an approach in image classification, which usually

examines the precision level between the classified image and the reference image (the original image)

Confusion matrix or also known as error matrix is recognized as a tool for accuracy

assessment A confusion matrix cross-tabulation can be described as a table that includes

a section of statistics prepared in rows and columns, which symbolizes the number of pixels (that are assigned to the reference image to be analyzed and compared to the classified image) that represent a particular type of class

Producer’s accuracy is described as a percentage of correctness determined by looking

on the classified image and predicting if pixels for every classes are correctly placed from the reference image

User’s accuracy is described as a percentage of correctness determined by looking on the

classified image and predicting if pixels for every classes are positioned in the same area

as if using a map to identify a location

Kappa Coefficient is described as a percentage of correctness between estimated model

and the real truth For further comprehension, in case the pixel statistics contained in a confusion matrix produce a result considerably much better than choosing a random pixel The Kappa Coefficient formulation is shown in the Appendix G, Equation 1

Overall accuracy or also known as the average accuracy is the overall accuracy of every

class quantified through the percentage of every test sample for that class Therefore, the overall accuracy is usually greater than the value of Kappa coefficient

NDVI or Normalized Difference Vegetation Index (NDVI) is spectral index that can be

examined by means of remote sensing methods and indicate perhaps the observed area

contains high quantity of vegetation or not NDVI calculation is shown in the Appendix G, Equation 2

FAPAR or the Fraction of Absorbed Photosynthetically Active Radiation (FAPAR) is

widely-known as the fraction of the arriving solar radiation from the Photosynthetically Active Radiation spectral region that can be consumed by a photosynthetic organism, basically explaining the light consumption throughout an integrated vegetation canopy This kind of biophysical distinction is definitely associated with the primary productivity

of the natural photosynthesis and several models apply it to calculate the intake of carbon dioxide of plants FAPAR equation is demonstrated in Appendix G, Equation 3

LST or the Land Surface Temperature (LST) is commonly the radiative complexion

temperature of the land surface, as calculated on the way to the remote sensor LST can be described as the combination of vegetation and bare soil temperatures LST affects the division of energy between soil and vegetation, and as well as determining the surface air

temperature LST formulation is indicated in Appendix G, Equation 4

SMI or the Soil Moisture Index (SMI) considers the water that can be found in the upper

10cm of soil SMI is regarded as an indicator of drought and soil moisture content The function of SMI is founded on the scientific parameterization of the association of LST and NDVI The equation for SMI is presented in Appendix G, Equation 5

Vegetation Index is a spectral transformation of two or more bands designed to enhance

the contribution of vegetation properties and allow reliable spatial and temporal

inter-comparisons of terrestrial photosynthetic activity and canopy structural variations NDVI and FAPAR are examples for vegetation index

Linear relationship is a statistical term used to describe the relationship between a

variable and a constant Linear relationships can be expressed in a graphical format where the variable and the constant are connected via a straight line or in a mathematical format where the independent variable is multiplied by the slope coefficient, added by a constant, which determines the dependent variable

PART II LITERATURE REVIEW 2.1 Land-Use and Land-Cover (LULC)

Land is one of the basic element and a primary resource to support human activities (Young, 1998) Due to the increasing population of human and the progression of technology, humans are labeled as the most powerful instrument when it comes to shaping the environment On a global scale, the majority of land-cover are influenced by human activities (Frimpong, 2011)

The idea of “land-use” was first applied by British geographer named Stamp (1948) Stamp explained that “land-use” is how humans interact to a particular land Therefore, the term land-use became known as a human activity or a land that reflects human activities For example, development, recreational, management, conservation, agricultural and other activities Furthermore, Stamp developed “Land Utilization Survey”

It was performed in the 1930s with the concept of “a field-to-field analysis of the whole nation, covering every acre and tracking its use”

Later on, FAO (1998) identifies land-use as “preparations, activities, and inputs humans perform in a certain land-cover type to create, change or maintain it” Furthermore,

Lambin et al (2006) define land-use as the manipulation of humans because of their

purpose to utilize a land Thus, these claims molded the term “land-use” and established

an understanding of what describes “land-cover” Land-cover is widely recognized as a remote sensing data, which can be analyzed of how many parts of a land is covered by forests, wetlands, agriculture, impervious surfaces, water bodies and other land types

2.2 Land-use research studies

Land-use study can be utilized for the purpose of examining “human interventions

to the terrestrial ecosystems” Colonization of human being to different ecosystems (e.g forests, landscapes) in order to manipulate them can certainly be examined through interpersonal and economic activities, which affect the ecosystems or by inspecting the modifications to those ecosystems (Krausmann, 2001) Moreover, land-use studies likewise employed for environmental science studies (Fischer-Kowalski and Haberl, 2007) concerning the recognized environmental issues (climate change, deforestation, the decrease of biodiversity, and so on)

Land-use change models are approaches to assist the investigation of the causes and effects of land use transformations Land-use models are capable to support land use planning and policy Various land use models are existing, formulated from distinctive

disciplinary backgrounds (Verburg et al., 2004)

O'Connell et al (2007) study regarding the connection in agricultural land-use

management and flooding in the United Kingdom (UK) Because of the “run-off” problem

in the local agricultural systems of UK, they created a model approach, which used to delineate back the downstream of run-off onto its sources

Moreover, Tong and Chen (2002) examined the hydrological effects of land-use to the watershed in Miami River Basin They established the statistical and spatial approach

to examine the factors that affect the watershed In result, statistical analysis had shown a

significant relationship between land-use and in-stream water quality, such as nitrogen, phosphorus, and fecal coliform

2.3 Remote sensing and GIS techniques for LULC change

Past studies by way of assessing LULC change that includes excessive efforts shown the needs for the support regarding the advancement of technologies especially using satellite sensors, to assist the long-run investigation of LULC change As outlined by

Miller et al (1998), remote sensing and GIS provide the most accurate methods to examine

and analyze different patterns of modifications in a land by having the scope to observe these transformations in numerous and different times Satellite data turned out to be the primary tool to measure LULC change with the capabilities to observe them repetitively within a short-intervals of time (Mas, 1999)

According to the case study of Hieu (2014) on “Land use changes assessment using spatial data: a case study in Cong river basin - Thai Nguyen City - Viet Nam”, several forest areas in Vietnam had changed for various purposes For instance, urbanization (e.g establishing new industrial parks, public areas, mining), agriculture activities and other activities associated with socio-economic purposes

Yang (2001) illustrates that the information about land-use change is necessary for updating land cover maps and the supervision for natural resources Based on the summarization of the approaches on change information extracted from the remotely sensed data, the study encourages the method of change detection based on remote sensing

information and model approach He states that, the foundation for research on how the change relations of natural and human activity have a connection on each other

2.4 Forest vegetation monitoring using RS and GIS techniques

Forests at a global scale are experiencing a different state of deforestation Remote sensing and GIS techniques have shown potential capabilities to monitor and detect forest changes in a spatial and temporal scale (Coppin and Bauer, 1996) An additional remarkable quality of RS data is that it provides a way of quickly discovering and interpreting different forest types, a job that would end up being tedious and time-consuming applying the traditional ground surveys (Canada Centre for Remote Sensing Tutorials, 2008) Data are obtainable at different scales and settles to fulfill regional as well

as local preferences Species detection can be carried out by way of multispectral, hyperspectral, as well as air photo data interpretation These imageries and the extracted data can be integrated into a GIS to further examine the slopes, possession boundaries, and

so on

Miwei (2009) examined short-lived vegetation located in Poyang Lake by using Moderate Resolution Imaging Spectrometer (MODIS) satellite imagery The analysis examined the variation Area of Ephemeral Vegetation (AEV) by studying time compilation

of MODIS imagery and inspecting how these differences relate to variations in hydrological conditions

Adia et al (2007) analyzed the spatial-temporal change detection concerning

vegetation cover of Jos, Nigeria as well as its neighboring areas The investigation utilized Landsat images Thematic Mapper and Enhanced Thematic Mapper (TM and ETM+) intended for observing vegetation reflectance, band 4, 3 and 2 (false color composite) of

TM and ETM+ were stacked to develop change maps of the vegetation cover for the corresponding dates and identify the pattern of change

As outlined by Fung and Chan, (1994) Satellite RS is known as a prominent method

to generate LULC maps as well as to examine vegetation cover A research conducted in

northern California in courtesy of May et al (1997), mentioned that TM appeared to be

more efficient than SPOT (Système Pour l'Observation de la Terre, a French owned and operated satellite) when it comes to separating bushes out from meadows, nevertheless neither of the two satellites (TM nor SPOT) data have been proficient in separating meadow types

2.5 Remote sensing climate change effects on forest vegetation

Consequently, climate change incorporates a remarkable problem on the forest vegetation structure Forest vegetation phenology comprises an effective bio-indicator of weather conditions as well as anthropogenic influences and an essential element for representing vegetation-climate relationships

The National Oceanic and Atmospheric Administration-Advanced Very Resolution Radiometer (NOAA-AVHRR) provides with high temporal resolution of 1

High-kilometer (1 km) The satellite also features an excellent job pertaining to discovering live wildfires and mapping burned scars employed in diverse biomes It consists of essential aspects meant for generating day-by-day information concerning NDVI, which is often accustomed to mapping forest vegetation

Palmate et al (2004) analyzed the consequences of climate variations on forest

cover and vegetation Vegetation parameter was identified by way of determining NDVI, and climate change parameters were incorporated with annual temperature and rainfall data for years 1998, 2000, 2002, 2009 and 2001 (pre-monsoon and post-monsoon) of the Betwa river basin, a tributary of River Yamuna in Central India The study proved that during pre-monsoon period temperature was noticeably related to forest cover During post-monsoon period rainfall showed a strong response to forest cover and temperature displayed weak response to vegetation in the Betwa river basin

Satellite RS is definitely an effective instrument to evaluate the primary phenological situations determined by monitoring vital variations upon temporal trajectories concerning forest biophysical variables such as Normalized Difference Vegetation Index (NDVI) and Leaf Area Index (LAI), which usually consists of time-series data that includes decent time resolution, throughout homogeneous area, cloud-free while not troubled by atmospheric and geometric impacts and dissimilarities in sensor features

(e.g calibration, spectral responses) (Zoran et al., 2014) Zoran et al (2014) examined

about the influences on the forest ecosystem of Cernica- Branesti positioned in the Eastern region of Bucharest town, Romania, by way of determining NDVI and LAI factors

North-extracted from MODIS Terra and NOAA AVHRR satellite data during the 2000-2013 period She mentioned that noticeable decrease in NDVI and LAI were commonly noticed

at the time of heat wave and drought occurrences during 2003, 2007 and 2012

PART III METHODOLOGY 3.1 Materials

3.1.1 Time and place of research

The research and writing of manuscript was conducted in GeoInformatic Research Center (GIC) for four months, from June 16th, 2017 to September 20th, 2015

3.1.2 Remotely sensed study area

Dai Tu district, Thai Nguyen province, Vietnam

3.1.3 Software used

1 ArcGIS software by Environmental Systems Research Institute (ESRI)

2 ENvironment for Visualizing Images (ENVI) software by Exelis

3 Google Earth by Google Inc

4 Microsoft Excel by Microsoft Inc

3.1.4 Satellite data used

The satellite data that were used in this study were obtained from USGS, Earthexplorer website (https://earthexplorer.usgs.gov) All scenes were cloud free in the area of interest, geo-referenced at Universal Transverse Mercator, World Geodetic System

84 (UTM, WGS84) The “Standard Terrain Correction (Level 1T) provided by USGS were automatically orthorectified using ground control points (GCP) and digital elevation model

(DEM) Two (2) Landsat 5 Thematic Mapper (TM) (acquired on June 8, 1993, and June 9, 2004), and one (1) Landsat 8 Operational Land Imager (OLI) (acquired on June 6, 2017) were used for the analysis

Table 1 Details of the satellite data used in the study

clarification

Satellite Sensor Path/Row

Spatial Resolution (M)

Date of Acquisition Sources

3.2 Methods

Figure 1: The overall methodological framework for assessing the effects of

climate change on forest cover

3.2.1 Satellite image pre-processing

Since, Landsat level 1T (L1T) products provided by the USGS were orthorectified and terrain corrected (applied with GCP and DEM), therefore, some image pre-processing were no longer needed except for radiometric calibration This is due to some atmospheric conditions, such as water vapor (humidity), aerosols (dust, smog from volcanoes, etc.),

Data Acquisition (Landsat Data)

Image Pre-processing (FLAASH

Climate Change Parameters

Establishing Relationship Accuracy Assessment Forest Cover Data

Soil Temperature Soil

Moisture

atmospheric thickness (pressure), and atmospheric disturbances Radiometric calibration was performed using the Fast Line-of-sight Atmospheric Analysis of Spectral Hypercube (FLAASH) atmospheric correction using Environment for Visualizing Images (ENVI) 5.1

software (Singh et al., 2013; Pryor, 2009; Smith, 2015)

3.2.2 Supervised classification

A supervised classification was established by using ArcGIS to assign 20 truth polygons for each class This was guided by further visualization of the study site using Google Earth, NDVI and FAPAR results The technique was used to improve the clarification of each land type for a better classification accuracy “Maximum likelihood”

ground-classifier in ArcGIS was used to generate the classified image (see Figure 3) (Churches et

al., 2014; Batar et al., 2017; Ghebrezgabher et al., 2016; Torahi and Rai, 2010; Jia et al.,

2014)

3.2.3 Accuracy assessment

In this study, a cross-tabulation confusion matrices were established to assess if the classified images are accurate Moreover, Kappa coefficient (or known as Cohen's Kappa) was developed to measure the agreement between the classified and the reference image

For years 1993, 2004, and 2017; a total of 200 (40 pixels from each class) testing pixels were placed at random parts of the study area The testing pixels were compared to the reference image to assess if the pixels were interpreted correctly to test the accuracy of the classified image Therefore, the producer’s accuracy, user’s accuracy, overall accuracy, and kappa coefficient were computed for the accuracy assessment of the final land-cover

maps produced (see Appendix C, D and E) (Churches et al., 2014; Batar et al., 2017;

Ghebrezgabher et al., 2016; Torahi and Rai, 2010; Jia et al., 2014)

3.2.4 Change rate analysis

Land-cover change analysis was derived for 1993 - 2004 (1st period) and 2004 -

2017 (2nd period) to produce cross-tabulation confusion matrices by the help of ArcGIS software This matrix was used to elaborate the useful details on the land-cover changes (changes in km²) and the change rates between the two years Land-cover changes of the 1st and 2nd period were also manipulated to produce gain-loss bar graph (see Figure 8)

(Batar et al., 2017; Ghebrezgabher et al., 2016; Torahi and Rai, 2010)

3.2.5 Vegetation indices and climate change parameters

Certain bands in Landsat 5 and 8 were used for calculating the vegetation indices and climate change parameters (see Table 2) For calculating the vegetation indices and climate change parameters, ArcGIS “raster calculator” was used to generate computations SMI and LST were categorized as climate change parameters and FAPAR and NDVI were categorized as vegetation indices ArcGIS was also used to prepare maps for NDVI (see Figure 9), FAPAR (see Figure 10), SMI (see Figure 11) and LST (see Figure 12) (Palmate

et al., 2014)

Spectral (Landsat bands)

Landsat 5: NIR (Band 4), Red (Band 3), Blue (Band 1), TIRS (Band 6)

Near Infrared (NIR) - Emphasizes biomass content and shorelines

Red - Discriminates vegetation slopes

Blue - Emphasizes biomass content and shorelines

Landsat 8: NIR (Band 5), RED (Band 4), BLUE (Band 2), TIRS 1 (Band 10)

Thermal Infrared Sensor (TIRS) - Thermal mapping and estimated soil moisture

Vegetation Indices

Normalized Difference Vegetation Index (NDVI) NDVI = 𝑁𝐼𝑅−𝑅𝐸𝐷

𝑁𝐼𝑅+𝑅𝐸𝐷

Fraction of Absorbed Photosynthetically Active Radiation (FAPAR)

(SMI)

SMI = (Tsmax−Ts)

( Tsmax−Tsmin)

Table 2 Illustrates the characteristics of Landsat bands that were used for

calculating vegetation and climate change parameters

climate change on forest cover (Batar et al., 2017)

PART IV RESULTS 4.1 Study area

4.1.1 Geography

Dai Tu is a mountainous district in the northwest of Thai Nguyen Province, bordering Dinh Hoa district to the north, Pho Yen district and Thai Nguyen city to the south, Phu Luong district to the east, and Tuyen Quang province to the northwest and Phu Tho province to the southeast (see Figure 2) Dai Tu district is located at a Latitude of 21° 37’ 49’’ North and Longitude of 105°38’ 28’’ to the East It encompasses 570 square kilometers (km²) of land

Figure 2: Maps and locations for Dai Tu district, Thai Nguyen province, Vietnam