MONITORING AND MAPPING OF THE EXTENT OF INDUSTRIAL FORESTS IN MALAYSIA By UY DUC PHAM A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of Forestry – Doctor of Philosophy 2016 ABSTRACT MONITORING AND MAPPING OF THE EXTENT OF INDUSTRIAL FORESTS IN MALAYSIA By UY DUC PHAM There are scattered studies in the international forestry sector that Industrial Forests (IFs) have been expanding as a newly-emerging Land Use and Land Cover Change (LULCC) in the tropics, especially in the Asia-Pacific region. However, these new tree plantations have not yet been well-documented; the area, along with its geography and land use dynamics, are not well known. Additionally, the drivers are not well understood, but it is widely believed that changes in tropical silviculture and increased international demand for wood and fiber are shifting to new demand centers in Asia. These trends have the potential to create global shifts in source producing areas, from long-standing IFs in North America and Europe to newer areas to the tropics.
Considerable remote sensing research and product development have been focusing on monitoring closed canopy natural forests, but less work has been done on intensively managed IFs, which involve techniques for remote characterization of the establishment, management, and rotation. Moreover, the studies to date have been geographically limited to some key areas, such as the Amazon and Indonesia, and more work needs to be done outside of these closed natural forest regions. This research is conducted in the tropical Asia-Pacific region with a focus on the new IFs in the Sabah and Sarawak states of Malaysia. This study aims to improve the knowledge base and understanding of the extent and characteristics of new IFs as a new agent of LULCC, and to develop the methods for Landsat data, in particular by using forest fractional cover (fC) and vegetation indices (VIs) analyses in time series integrated with textural, spectral, visual, and other analyses to detect and quantify IF LULCC patterns and dynamics in the country.
Results showed that the selected IFs-including acacia, rubber, and other IFs-have expanded quickly from 2000 to 2014 with a net increase of 288,547 ha at the annual mean rate of 20.1% in Sabah, and 459,898 ha at the annual mean rate of 59. The annual mean expansion rate of faster-growing, shorter-rotation acacia IFs at 28.4% in Sabah and 376.5% in Sarawak was much faster than that of slower-growing, longer-rotation rubber IFs at 13.7% in Sabah and 5.8% in Sarawak, as well as other IFs at 10.9 % in Sabah and 78. The development of IFs in both states was primarily dominated by the larger scale holdings; however, the role of the small-scale IFs in developing new IFs in the region grew through an increase of its total area and rate of change in area. The expansion of IFs in Sabah and Sarawak significantly contributed to a LULCC in the regions.
Most of these new IFs replaced disturbed natural forests (81-95%), followed by agricultural land (4-18%), and waste land (< 1%). These have caused a significant decline for the aboveground C stock in Sabah (11.5 Tg C) and Sarawak (24.7 Tg C), and resulted in an emission of 42.1 Tg CO2 in Sabah and 90.5 Tg CO2 in Sarawak over the period. The expansion of these new IFs had also led to a reduction in biodiversity in Sabah at 2.98% and in Sarawak at 2. The results also showed a possibility of developing the fC and VIs-based methods in a time series for Landsat datasets that could detect and monitor the extent, pattern, and scale of IFs in the tropics.
The accuracy for detecting the IF land using the fC-based method (with its producer’s accuracy at 83% and Kappa coefficient at 0.46) was higher than that of the VIs-based method. Among VIs, ARVI worked the best with its producer’s accuracy at 64% and Kappa coefficient at 0.4, followed by SAVI, SARVI, EVI, NDVIaf, and MSAVIaf. For both the fC-based method and the VIs-based method, the accuracy of detecting acacia and rubber IFs was better than that of other IFs in the region. In brief, this study successfully developed the fC- and VIs-based methods in multi-dated Landsat data to detect and quantify IF LULCC.
Copyright by UY DUC PHAM 2016 This dissertation is gratefully dedicated to my family, especially to my beloved wife, Lien Hoang Thi Pham, and to my adored daughter, Elise Pham or Linh Khanh Pham, who have encouraged, inspired me, and sacrificed a lot throughout my Ph. I would also like to dedicate it to my whole family, my parents, and my parents-in-law, who tirelessly support me during my entire doctorate program. Without their supports and love, I could not complete this dissertation study and my doctorate program. v ACKNOWLEDGEMENTS First and foremost, I would like to deeply express my sincere gratitude to my main advisor Professor David L.
Skole for the continuous, generous, and tireless support of my Ph. study, for his patience, motivation, kindness, and immense knowledge. His priceless guidance and comments helped me in all the time of research and writing this dissertation. Besides, I would like to sincerely thank all my guidance committee members: Prof.
Stuart Gage, Prof. Leefers, and Prof. Pascal Nzokou, for their insightful comments and encouragements. Without their guidance and supports, I could not complete this study.
On this occasion, I would like to give special thanks to Professor Richard Kobe, who has offered me an opportunity to study here, at the Department of Forestry, Michigan State University, USA. My thanks also go to my lab-mates and all staff at the Center of Global Observatory for Ecosystem Services (GOES), Michigan State University, especially to Daniel Zelenak and Hartanto Sanjaya for remote sensing and GIS technical discussions and supports, to Jay Samek and James Gray for many supports and insightful recommendations for my study, and also to Mr. James Gray and Ms. Cat Olenick for their English proofreading assistance for this dissertation.
I would also like to say thank you to all of Michigan State University’s Forestry Department Faculties, particularly to Dr. Phu Nguyen, Dr. Runsheng Yin, and Dr. David Rothstein for their early discussions, supports, and encouragement for my study; and to all my family members and friends, who have supported spiritually throughout my doctorate program.
Lastly, I would like to say that my words are never enough to say thank you to all of you for all what you have done and given to me. I really appreciate that and thank you very much again. vi TABLE OF CONTENT LIST OF TABLE ix LIST OF FIGURE xii KEY TO ABBREVIATIONS xx CHAPTER 1: OVERVIEW 1 1.2 Industrial Forest: Concepts and Definitions 5 1.3 The Development of Industrial Forests (IFs) in the Asia-Pacific Region 7 1.4 The Development of Industrial Forests in Malaysia 13 1.5 Literature Review of the Studies on Industrial Forests 16 1.1 In the Asia-Pacific Region 16 1.6 Literature Review of the Studies on the Methods Development for Detecting and Mapping Industrial Forests 21 1.7 The Significance of this Study: Problems and Rationale 23 1.8 Selection of the Study Area and Industrial Forest Systems 26 Industrial Forest Systems Studied 27 1.9 Research Questions and Objectives 29 Research Questions 29 Objectives 30 1.10 Research Methods 30 Validation 32 1.11 The Flowchart of the Study 37 CHAPTER 2: DEVELOPING THE VEGETATION INDICES-BASED INDUSTRIAL FOREST DETECTION METHOD FOR LANDSAT DATASETS 38 2.2 Acquiring and Preprocessing Images 40 2.3 Developing the Method 44 2.2 Results 45 Textural Analysis 62 Spectral Analysis 70 Visual Interpretation and Using Other Data 79 Making IF Maps 83 2.5 Discussions and Conclusions 97 vii CHAPTER 3: DEVELOPING THE VEGETATION/FOREST FRACTIONAL COVER- BASED INDUSTRAL FOREST DETECTION METHOD FOR 101 LANDSAT DATASETS 3.2 Acquiring and Preprocessing Images 102 3.3 Developing the Method 103 3.2 Results 105 Spectral Analysis 120 Textural Analysis 120 Visual Interpretation and Using Other Data 122 Making IF Maps 123 3.5 Discussions and Conclusions 130 CHAPTER 4: ASSESSING THE INDUSTRIAL FOREST LAND USE AND LAND COVER CHANGES, AND THEIR CONSEQUENCES 133 4.1 Industrial Forest Land Use and Land Cover Changes 133 4. The fC-based LULCC 133 The Pattern Indices for IF LULC Changes 140 4.
The Vegetation Indices-based LULCC 145 4.2 Assessments of the IF LULC Changes and their Consequences 154 The Consequences of the IF LULC Changes 161 4.3 Discussions and Conclusions 171 CHAPTER 5: SYNTHESIS 180 5.3 Applicability 184 APPENDIX 185 REFERENCES 236 viii LIST OF TABLES Table 1. A summary of plantation areas and the rate of their change in Malaysia by state…. The principal requirements for a sampling scheme to validate the developed methods……………………………………………………. The ways for assessing accuracy of IF maps derived from Landsat datasets………………………………………………………………….
Sequences of the vegetation cover changes based on the changes of VI values (MSAVIaf) in 30 key areas chosen to observe in Sabah and Sarawak, 2000-2014…………………………………………………. The changes of MSAVIaf values in some key areas in Sabah, 2000- 2014. The accuracy assessment results for ARVI, EVI, MSAVIaf, NDVIaf, SARVI, and SAVI-based IF land detection methods for Landsat data… 93 Table 2. The accuracy assessment results specific for acacia, rubber, and other IFs for ARVI, EVI, MSAVIaf, NDVIaf, SARVI, and SAVI-based IF detection methods for Landsat data………………………………….
The fC value changes in 30 monitored key locations in Sabah, 2000- 2014. The accuracy assessment results for the fC-based IF land detection method …………………. The accuracy assessment results specific for acacia, rubber, and other IFs for the fC-based IF detection method for Landsat data……………. The IF area expansion in Sabah, 2000-2014………………………….
The IF area expansion in Sarawak, 2000-2014………………………. The area (in ha) of large-scale and small-scale IFs in Sabah, 2000- 2014…………………………………………………………………… 138 Table 4. The percentage of large-scale and small-scale IFs in Sabah, 2000- 2014……………………………………………………………………. The area (in ha) of large-scale and small-scale IFs in Sarawak, 2000- 2014…………………………………………………………………….
The percentage of large-scale and small-scale IFs in Sarawak, 2000- 2014……………………………………………………………………. The new IF areas and their LULC replacements in Sabah, 2000-2014… 156 Table 4. The new IF areas and their LULC replacements in Sarawak, 2000-2014 159 Table 4. The above ground carbon stock values (tC ha-1/MgC ha-1) for the classified LULC types in Sabah and Sarawak (adapted from Agus et al., 2013a; Agus et al.
Comparisons of ABG stocks (Mg) of new IFs and their LULC replacements in Sabah…………………………………………………. Comparisons of C stocks (Mg) of new IFs and their LULC replacements in Sarawak………………………………………………. The number of species in the different LULC types in the study area…. The percentage of declining or increasing number of species if UF, DF and OP lands were converted into IF land……………………………… 168 Table 4.
Estimating the biodiversity loss caused by the expansion of the new IFs in the study area from 2000 to 2014 (adapted from Brook et al. The full list of Landsat scenes used for the study in Sarawak, Malaysia, 2000-2014………………………………………………………………. The date, type, and cloud coverage of Landsat scenes used for the study in Sarawak, Malaysia, 2000-2014………………………………………. The full list of Landsat scenes used for the study in Sabah, Malaysia, 2000-2014……………………………………………………………….
The date, type, and cloud coverage of Landsat scenes used for the study in Sabah, Malaysia, 2000-2014…………………………………………. The changes of ARVI, EVI, MSAVIaf, NDVIaf, SARVI, and SAVI values in 30 key areas in Sabah, 2000-2014……………………………. The result for calculating the growth rate of VIs in Sabah, 2000-2014… 207 x Table A. The GLCM_MEA, DIS, and HOM values for different LULC types in VIs, band 4 and 5 images in Sarawak, 2000-2014……………………… 213 Table A.
The GLCM_MEA, DIS, and HOM values for different LULC types in VIs, band 4 and 5 images in Sabah, 2000-2014………………………… 215 Table A. The PCA, ICA, TCA, band 4 and band 5 values for different Land Use Land Cover (LULC) types in Sabah, 2000-2014………………………. The PCA, ICA, TCA, band 4 and band 5 values for different Land Use Land Cover (LULC) types in Sarawak, 2000-2014……………………. The fC value changes and its change sequence in 30 monitored key locations in Sarawak, 2000-2014……………………………………….