CHAPTER can act as both source and sink of

CHAPTER 1. INTRODUCTION

1.1.        
Background

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

Forests sequester a large
amount of carbon and plays a crucial role in the global agenda of climate
change.  Forest can act as both source
and sink of carbon. When the forest is healthy and growing, carbon is sequestrated
from atmosphere; but when the forests are destroyed, overharvested, or burned,
they no longer contribute in sequestration and become a source of CO2
which increase climate change (Hussin et al., 2014). Hence, quantification of forest
biomass is of vital importance to assess productivity – a critical information
for carbon budget accounting, carbon flux monitoring and for understanding the
forest ecosystem response to climate change (Watham et al., 2016; Nandy et al., 2017). Meanwhile, reforestation,
afforestation and avoiding deforestation are mechanisms of tackling climate
change (Hunt, 2009; Luong et al., 2015). 
In addition, estimation of the forest carbon stocks not only contributes
in reducing emissions from deforestation and forest degradation (REDD); but
also in sustainable management of the forest (Hussin et al., 2014).

The quantification of
biomass and carbon sequestration in tropical forests is particularly relevant
within the United Nations Framework Convention on Climate Change (UNFCC). The
UNFCC adopted Kyoto Protocol which sets binding targets to industrialized
countries for reducing greenhouse gases emissions (Breidenich et al., 1998; Protocol, 2011; Hussin et al., 2014). The Bali Action Plan Conference of
the Parties (COP-13) in 2007 opened an avenue for developing countries to
participate in forest carbon financing through the mechanism of reducing
emissions by reducing emissions from deforestation and forest degradation (REDD)
(Hussin et al., 2014;
Luong et al., 2015). Under the REDD mechanism, countries
will need to measure and monitor the emissions of CO2 resulting from
deforestation and degradation within their borders (Luong et al., 2015). Emissions are converted to carbon credits in the carbon
trade. All the greenhouse gas inventories and emissions reduction programs
require scientifically robust methods to quantify forest carbon storage over
time across extensive landscapes (Gonzalez et al., 2010). Vietnam has been participating in UN-REDD as a potential
member of carbon trade, which requires estimation of biomass/carbon stock in
the country to be prepared for REDD implementation.

Remotely sensed data integrated with
forest inventories has been becoming an effective approach used to estimate above
ground biomass (AGB) and hence ultimately carbon stocks. Remote sensing-based
studies relate reflectance recorded at the sensor with ground-based
measurements to estimate biomass (Tucker et al., 1985; Sader et al., 1989; Gibbs et al., 2007; Kumar et al., 2015). Recently, many studies in different
regions have found strong correlations between biomass and reflectance at
different wavelengths (Kumar et al., 2015). Kumar et al. (2015) also concluded that for regional level
where field data are scarce or difficult to collect, remote sensing is the superlative
method to estimate biomass since its enhanced spatial, spectral, and
radiometric characteristics (Delegido
et al., 2011; Irons et al., 2012; Chrysafis et al.,
2017) can further
contribute to accurate, spatially explicit estimations of forest inventory
parameters, and
improved update frequency with a lower cost for monitoring forests and
measuring variables (Andersson et al., 2009; Dube & Mutanga, 2015; Yadav & Nandy, 2015). Therefore, this method to become a
popular method and widely used for biomass estimation.

Optical remote sensing data
which provide a wide range of spatial and temporal resolutions, have been widely
utilized for forest biomass assessment applying different forms of techniques (Foody et al.,
2003;
Lu, 2005; Rahman et al.,
2005;
Hyde
et al., 2006; Li et al., 2008; Kumar et al., 2015). Relating to using optical
data for biomass assessment, the approaches namely multiple regression
analysis, k-nearest neighbor, and machine learning (artificial neural network (ANN),
random forest, ect.) either have been commonly applied (Phua & Saito, 2003; Kumar et al.,
2015) or especially appropriate for medium-spatial resolution data
(Franco-Lopez et
al., 2001; Soenen et al., 2010). By using these techniques, various types of both vegetation
indices (VIs) and band ratios obtained from optical data are also utilized to estimate
biomass by correlating vegetation index values or band ratio values with field measurement
(Dong et al.,
2003;
Kumar et al., 2015).

In recent year,
machine-learning algorithms were trialed for capability to perform flexible
input-output nonlinear mappings between remotely sensed data and biomass (Montes et al.,
2011;
Gleason & Im, 2012; Prasad et al.,
2012;
Wang
et al., 2016). Typically, ANNs and
support vector regressions were employed to couple with VIs to build monitoring
models with improved prediction accuracy of remote estimation of biomass (Wang et al.,
2016). Among a variety of machine learning techniques, the
emerging Random Forest (RF) algorithm proposed by Leo Breiman and Cutler Adele
in 2001 has been regarded as one of the most precise prediction methods for
classification and regression, as it can model complex interactions among input
variables and is relatively robust in regard to outliers (Wang et al.,
2016).

Forests cover
nearly 40% of the total land area of Nepal (Oli & Shrestha, 2009) which
signifies the amount of carbon in the forests of Nepal. But national forest
inventory data on changes in forest cover, biomass stocks, carbon emissions and
carbon removals on a periodic basis are limited (Acharya, et al., 2009).
In order to capture the benefits accruing from climate change scenario, there
is an urgent need of obtaining reliable baseline statistics on carbon stocks
and fluxes in forest which requires advanced remote sensing technologies (Oli
& Shrestha, 2009). In addition, carbon credit buyers will expect the use of
a robust methodology of carbon accounting and monitoring (Acharya, et al., 2009)
while commencing carbon trade. Hence, it becomes crucial to produce a credible
estimate of national forest carbon stocks and sources of carbon emissions, to
determine a national reference scenario and develop a national REDD strategy in
Nepal (MOFSC, 2009).

1.2.        
Statement
of problem

The quantification, mapping
and monitoring of biomass are now key issues due to the importance of forest biomass
in ecosystem and biomass role as a renewable energy source in many countries around
the world. However, detailed ground-based information of total biomass are scarce
(Sierra et al.,
2007;
Hussin et al., 2014). AGB estiation for the
tropical and sub-tropical area is still a challenging task and requires
accurate and consistent measurement methods because these forest areas are
characteristed with complex stands and varying environmental conditions (Lu, 2005; Kumar et al., 2015). A shortage of information of global biomass due to uncertainties
in accuracy and cost has been existed as a considerable issue for which is needed
further investigation (Nguyen, 2010; Hussin et al., 2014). Moreover, according to Lu (2006), it is essential to integrate
remotely sensed data and forest inventory data, so as to develop appropriate approach
for AGB estimation. According to Zianis and Mencuccini
(2004), it is essential to
develop and impliment effective methods to estimate AGB for carbon quantification
which can be vital sources to monitor changes in carbon stocks (Ketterings et
al., 2001; Hussin et al., 2014).

 

1.3.        
Objectives
and research questions

1.3.1. Research Questions

   The
study is proposing to investigate following questions:


How to extract spatial, spectral and topographic (elevation, slope,
aspect) variables from remotely sensed data?


Which spectral, spatial and topographic variables are
relevant to biomass estimation?


How good is the machine learning technique (random forest regression
model) for estimating biomass?

}  What
is the relationship and potential of Sentinel-2 MSI and Landsat-8 OLI original
and synthetic bands for AGB prediction?


What is the amount of above ground biomass in Yok Don
National park ?

1.3.2. Research Objectives

The main objective of this research is
to develop a method to accurately estimate AGB using remote sensing approach in
which the potential of Sentinel-2 with spectral information in AGB prediction
and the potential enhancement over previously available Landsat-8 OLI imagery
will be examined. To obtain this key aim, four specified objectives structure
this study:

}  To
extract the spatial, spectral and topographic
variables from satellite imagery;

}  To
identify the correlation and potential of Sentinel-2 MSI and Landsat-8 OLI
original and synthetic bands for AGB prediction;

}  To
determine optimized spatial, spectral and topographic
variables for AGB assessment;

}  To
estimate and map the AGB in Yok Don National park.

1.4.        
Thesis
structure

This thesis is divided into six chapters
as follows. Chapter 1 will provide an introduction on the importance of forest biomass
estimation, as well as the necessity of the application of remote sensing techniques
in biomass estimation. The research objectives and questions thesis structure
is also presented in the later part of the chapter. In Chapter 2, the
literature review is presented which provides a brief overview on information
on the application of remote sensing technology in forest biomass studies, such
as biomass concepts and definitions, overview of tools and techniques for
biomass estimation, learning machine techniques for biomass estimation, current
findings and knowledge gaps, particularly in the study area. The research
methods applied for addressing the objectives and a description of the study
area and data used are presented in Chapter 3. Findings of the project are revealed
in Chapter 4 whereas a discussion of findings achieved from applying the
methods is presented in Chapter 5. Chapter 6 provides conclusions and
recommendations of the research.