M. Mahmudur Rahman


ISBN: -978-3-938807-73-6
Veröffentlicht: Feb. 2008, 1. Auflage, Einband: Broschur, Seiten 260, Format B5 17,6 x 25,0 cm, Gewicht 0.48 kg
Lieferzeit: 2 - 6 Werktage
Verfügbarkeit: Auf Lager
39,80 €


Mehr Ansichten


Short Abstract

The issue of climate change is often discussed by the environmentalists and carbon emission is one of the main drivers of this threat. Increasing accumulation of atmospheric CO2 is the consequence of human activities including fossil-fuel emission and deforestation. Our civilization would not be able to drastically cut fossil-fuel burning unless an alternative source of energy is discovered. On the other hand, global carbon emission can be reduced by halting deforestation. This research work contributes to integrate remote sensing and terrestrial sample-based forest inventory data for estimation of carbon pool in a forest ecosystem using an alternate regression technique. Regression has been used to estimate forest biomass (carbon stocks) from remotely sensed data over decades. The investigation applied dummy variables in regression analysis and the dummies were set from the optimal stratification of forest-land. The finding increased the correlation that often used as an indicator of precision in regression modelling. Tropical forest region of south-eastern Bangladesh was selected as test-site, which had been affected by deforestation in the past decades. Finally, the study estimated the amount of carbon released with the result of forest cover change in the region. The technique proposed in this study can be extended to the deforestation hotspots to estimate carbon emission and would help to understand the terrestrial carbon dynamics and global climate change.


Author’s Biography

M. Mahmudur Rahman was born in 1972 in Mymensingh, a northern city of Bangladesh. He obtained B.Sc. (Hons.) at Institute of Forestry (later changed to Institute of Forestry and Environmental Sciences), University of Chittagong (Bangladesh). He started his professional career as Assistant Conservator of Forests (ACF) at Bangladesh Forest Department in 1995. After a few months he changed his job to Bangladesh Space Research & Remote Sensing Organization (SPARRSO) and started working as a researcher. He completed M.Sc. (2000) and PhD (2004) from Faculty of Forest, Hydro- and Geosciences, Dresden University of Technology (Germany). Then he returned back to SPARRSO and applied remote sensing techniques for various applications in Bangladesh. Currently he is working at Center for Environmental Remote Sensing (CEReS), Chiba University (Japan) as a post-doctoral researcher funded by Japan Society for the Promotion of Science (JSPS). He presented his research-work at many international conferences and published peer-reviewed articles in scientific journals. Mr. Rahman has devoted himself to carry out investigation for several environmental issues of the Earth using both optical and microwave remote sensing technology.

Der Herausgeber:
Prof. Dr. techn. habil. Elmar Csaplovics leitet den Lehrstuhl Geofernerkundung am Institut für Photogrammetrie und Fernerkundung der Technischen Universitaet Dresden

TU Dresden, Institut für Photogrammetrie und Fernerkundung, Helmholtzstraße 10, 01062 Dresden

Prof. Dr. Elmar Csaplovics: "In loser Folge sollen in dieser Schriftenreihe wissenschaftliche Arbeiten, die am Lehrstuhl für Fernerkundung bearbeitet und betreut werden, in ansprechender Form veröffentlicht werden. Wir glauben, dass dadurch das Spektrum von Literatur zum Themenkreis projektorientierter angewandter fernerkundlicher Forschung nachhaltig bereichert werden kann und wünschen uns demgemäß eine kritikfreudige Schar von Leserinnen und Lesern." >Professur Geofernerkundung


Nature, to be commanded, must be obeyed
(Francis Bacon, Novum Organum, 1620)

We hand over to the community of interested readers the third volume of our series on remote sensing and applied geoinformation analysis. As its forerunners it aims to publicise an outstanding PhD thesis which was conducted and completed at our research unit. The work appears in a slightly adapted version suitable for attracting a representative spectrum of students and professionals in environmental sciences and environmental management in general and in integrated monitoring of forest environments in particular.

In Africa and monsoon Asia, only about one-third of the forests of 10 millennia ago still remain. … Temperate forests had declined sharply in previous centuries and could rebound because of the confluence of three factors: slow population growth, less land required for farming because of yield improvements, and the emergence of overseas sources of supply for timber. In this last respect, the stabilisation of forest area in temperate lands promoted the deforestation of the tropics
(John R.McNeill, Something new under the sun - an environmental history of the twentieth century world. Norton, London, p.229f.).

Tropical deforestation is one of the most striking driving factors of environmental change. Its sources are spread all over the tropical eco-climatic zone and can be detected by the puzzled patterns of land use and land cover changes in regional and local scales. Their impact heavily affecting the mechanisms of global climatic change. To model the latter needs an operational monitoring scheme of these changes with effective accuracies both in space and in time. Earth observation via remote sensing satellites proved to serve as one of the more valuable technologies to provide such data, not to forget the needs for a well-structured field sampling methodology as well as the in-depth monitoring of socio-economical and socio-ecological background information.

Mahmud Rahman is scientific officer at the Bangladesh Space Research and Remote Sensing Organization (SPARRSO). He got a leave for his PhD studies at the University of Dresden and was awarded a PhD grant by the Daimler-Benz-Stiftung from 2000 till 2002. He received furthermore short-time grants of honour provided by the Association of the Patrons and Friends of the University of Dresden several times. Currently he is a JSPS Postdoctoral Research Fellow at the Centre for Environmental Remote Sensing (CEReS), Chiba University, Japan.

The scientific activities of Mahmud Rahman deal with the severe problems of tropical deforestation, carbon release and related social dynamics. His studies focus on a region in Southern Chittagong, Bangladesh. The tools for his investigations are mainly provided by various levels of earth observation and monitoring by remote sensing satellites and by methods of in-situ observations. Mahmud has provided extended research on the methodology of assessment of forest cover change in the context of tropical forests in Southern Chittagong. He has carefully prepared a systematic analysis of related patterns of carbon release, as well as of socio-ecological and socio-economic impacts and their interrelations with land cover and land use change in the tropical forests of Chittagong. Field campaigns had been laid out in the periods of autumn/winter 2002/2003 and 2003/2004. Extended data analysis and interpretation provide a bunch of results which are extensively discussed and analysed.

A new quality of information is extracted for the specific study area in terms of quantitative and qualitative estimations of biomass, carbon content and carbon flux based on temporal change detection. These data are a priori of nearly no evidence if regarding the global context of carbon fluxes, but are highly representative for proving for the efficiency and accuracy of calculating carbon content and fluxes (release and sequestration) for specific regions. The sum of regions, all specifically structured and affected by different variations of human inference, makes the whole global aspect. The results of this research work are remarkable and important, because they provide in-depth information on regional level essential for any extrapolation on continental or global level. Methods and strategies can in a medium-term be transferred to other regions of tropical forests worldwide.

This research work will furthermore contribute to the development of a guide of appropriate application of regional earth observation data and methodologies of data analysis for promoting a world-wide implementation of comparable tools in regional initiatives of multi-temporal change detection analysis of forest environments, of estimating biomass and carbon transfers and thus of preparing spatio-temporal information on the specific varieties of regional dynamics of tropical forest cover changes which are correlated to one of the most dangerous impacts on the global environment nowadays.

Conclusions of the research work clearly point out that experts in spatial analysis of forest environments and decision makers have to cooperate more closely in order to develop long-term recommendations for establishing and maintaining networks of regional monitoring programmes for assessment and significant reduction of tropical deforestation worldwide.

Francis Bacon, Francisci de Verulamio Summi Angliae Cancellarii Instauratio Magna (Distributio operis. Eius constituuntur Partes sex. Prima ; Partitiones Scientiarum. Secunda; Novum Organum sive Indicia de Interpretatione Maturae. Tertia; Phaenomena Universi, sive Historia Naturalis & Experimentalis ad condendam Philosophiam. Quarta; Scala Intellectus. Quinta; Prodromi, sive Anticipationes Philosophiae Secundae. Sexta; Philosophia Secunda, sive Scientia Activa). Apud Ioannem Billium, Londini, 1620
John R.McNeill, Something new under the sun – an environmental history of the twentieth century world. Norton, London, 2000

Dresden, November 2007 Prof. Dr. habil. Elmar Csaplovics


Table of Contents

Editorial VII
Preface XI
Abbreviations XVII
Symbols, Names and Units XVIII

1 Introduction 1
1.1 Global Climate System and Recent Climate Change 1
1.2 Carbon Cycle 5
1.3 Forests and Carbon Fluxes 8
1.3.1 Carbon storage in forest ecosystem 8
1.3.2 Carbon flux from forests 11
1.4 Methods for Carbon Flux Estimation 16
1.5 Objective and Research Hypothesis 19

2 Quantifying Forest Attribute and Carbon Flux 23
2.1 Carbon Stock?Estimation using Remote Sensing 23
2.1.1 Carbon stocks measurement in forest ecosystem 23
2.1.2 Remote sensing as a tool for carbon stock estimation 24 Very high spatial resolution data 25 High and intermediate spatial resolution data 25
2.1.3 Spectral reflectance properties of vegetation and forest canopies 27
2.2 Remote Sensing Image Pre-processing 29
2.2.1 Atmospheric effect on radiation 29
2.2.2 Haze computation 30
2.2.3 Dark Subtraction 31
2.2.4 Data normalization 35
2.3 Remote Sensing Image Interpretation 37
2.3.1 Elements of visual interpretation 37
2.3.2 Visual-interpretation keys 38
2.4 Remote Sensing Image Processing 39
2.4.1 Supervised classification 39
2.4.2 Vegetation indices 40
2.4.3 Tasseled Cap transformation 41
2.4.4 Principal Component Analysis 45
2.4.5 Structure and image texture analysis 46
2.4.6 Fusion of higher and lower resolution images 49
2.4.7 Accuracy assessment of classification 50
2.5 Sampling Strategy 51
2.5.1 Double and two-stage sampling 51
2.5.2 First phase sampling 52
2.5.3 Second-phase sampling 53
2.5.4 Determination of sample size 54
2.6 Field Estimation of Forest Biomass / Carbon Pool 55
2.6.1 Estimation by direct measurement 55
2.6.2 Estimation by tree form factor 55
2.6.3 Estimation using functions and tables 57
2.7 Integration of Remote Sensing and Terrestrial Information 58
2.7.1 Incorporation of field measurement on pixel level 58
2.7.2 Double sampling for stratification 59 Background information 59 Strata as a domain of study: 61
2.7.3 Double sampling with regression estimator 62 Background information 62 Residual analysis 65 ‘Dummy’ Variables to Separate Blocks of Data 66 Criteria for comparing candidate models 68
2.7.4 Double sampling with knn estimate 70 Background information 70 Knn estimate procedure 71
2.8 Literature Review: Estimation of Forest Attributes from Remote Sensing Data 72
2.8.1 Boreal forest 73
2.8.2 Temperate and subtropical forests 74
2.8.3 Tropical forests 75

3 Study Area and Stand Information 77
3.1 Location 77
3.2 Climate 77
3.3 Topography 78
3.4 Geology and Soil 78
3.5 Description of Forest Types 79
3.5.1 Tropical wet evergreen forests 80
3.5.2 Tropical semi-evergreen forests 80
3.5.3 Semi-evergreen shrub forest and savannas 81
3.5.4 Moist bamboo brakes 81
3.6 Forest Jurisdiction 82

4 Methodology 83
4.1 Remote Sensing Image Information 83
4.2 Remote Sensing Image Pre-processing 87
4.2.1 Radiometric correction 87 DOS model 90 Improved image-based model (COST model) 91
4.2.2 Geometric corrections 95
4.3 Image Fusion 96
4.4 Preparation of Interpretation Key 96
4.5 Stratification of Landsat ETM+ Image 97
4.6 Field Sampling: 97
4.6.1 Sample plot location 97
4.6.2 Sub-sampling plot size 97
4.6.3 Field measurement of sub-sample plots 100
4.6.4 Estimation of plot biomass 101 Measurement of tree diameter and height 101 Conversion of dbh and height to biomass 101
4.7 Carbon Pool Estimation 102
4.7.1 Stratification 102
4.7.2 Regression estimator 102
4.7.3 Knn estimation 103
4.7.4 Best method selection 104 Prediction error estimation 104 Bias calculation 104
4.8 Carbon Release Estimation 105

5 Results 107
5.1 Atmospheric Correction 107
5.2 Vegetation Class Interpretation 108
5.2.1 Optimal band selection 108
5.2.2 Tropical vegetation interpretation 109
5.2.3 Individual class separation 111
5.2.4 Selective interpretation key 120
5.3 Carbon Pool Estimation for Recent Time 122
5.3.1 Using forest stratification 122 Stratification using terrestrial information 122 Stratification without terrestrial information 124 Estimates of carbon 128
5.3.2 Using regression technique 130 Simple regression 131 Multiple regression 137 Addition of image texture 138 Addition of dummy variables 140
5.3.3 Using knn method 145
5.4 Comparison of Methods 147
5.5 Carbon Assessment for Historical Time 149
5.6 Carbon Release estimation 153
5.6.1 Deforestation assessment by change matrix 153
5.6.2 Carbon flux study 157
5.7 Classification Accuracy 159

6 Discussion 167
6.1 Forest Carbon Estimation from Satellite Sensor Data 167
6.2 Carbon Flux Estimation 172

7 Conclusion 175
7.1 Important Research Findings 175
7.2 Limitations of the Study 175
7.2.1 Remote sensing datasets 175
7.2.2 Topography 176
7.2.3 Location uncertainty 177
7.2.4 Mixed pixel and classification 177
7.2.5 Sampling error 177
7.2.6 Measurement error 178
7.2.7 Estimation error 178
7.2.8 Lack of below-ground carbon information 179
7.3 Recommendation and Future Outlooks 179

References 183
Appendix 211


Gewicht 0.4750
Lieferzeit 2 - 6 Werktage