Get Complete Project Material File(s) Now! »
Greenhouse gases and climate change
The study of the carbon cycle has recently taken a special relief in the context of the issue of global warming: Two of the greenhouse gases involved: the carbon dioxide (CO2) and methane (CH4), participate in the carbon cycle, as they are the main atmospheric carbon forms. In addition to climate issues, the study of the carbon cycle will allow us to determine the effects on the release of carbon stored in the form of fossil fuels by human activity.
In fact, the global carbon cycle has been greatly altered by human activity in the past decades. Indeed, carbon dioxide resulting from human emissions exceeded natural fluctuations (). The changes in the amount of CO2 in the atmosphere are altering weather patterns and oceanic chemistry. Studies have shown that even though global temperatures can fluctuate without changes in atmospheric CO2, the latter cannot change without affecting the atmospheric temperatures. In addition, CO2 levels are rising higher than ever recorded in the atmosphere (). Therefore it is of high importance to better understand the carbon cycle and its effects on the global climate ().
Global carbon cycle’s carbon sinks
The global carbon cycle is divided into four main carbon sinks connected by pathways of exchange ():
– The lithosphere contains carbon in its carbon and carbonated rocks (30 mGt).
– The hydrosphere contains carbon in its dissolved form (38 000 Gt) and in marine organisms (3 Gt).
– The biosphere contains 2,300 Gt of carbon in the form of biomass and necromass and in soils
– The atmosphere contains 700 Gt of carbon as CO2.
The exchanges of carbon between these fours sinks occur as a result of various chemical, physical, geological, and biological processes. The ocean contains the largest active sink of carbon near the surface of the earth (). In addition, carbon exchange between the different compartments is balanced, which makes the carbon levels stable without human influence ().
The lithosphere contains the largest amounts of carbon in the form of carbonated rocks and fossil fuel (); it does not exchange a lot of carbon naturally with the other compartments. This is due to the fossilization rate of living beings or the sedimentation of carbonated rocks which can take several million years. However, the CO2 emissions in the atmosphere resulting from the use of fossil fuel are the principal flux that concerns this carbon stock.
The hydrosphere and the biosphere are in equilibrium due to the high solubility of the CO2 in water and the important volume of oceans. In fact, oceanic absorption of CO2 is one the most important forms of carbon sequestering. This high absorption rate limits the carbon dioxide in the atmosphere caused by human activities. However, this process may make oceanic waters more acidic due to the increase uptake of carbon, as well as limiting the ocean uptake of CO2 ().
Finally, the biosphere exchanges up to 60 Gt/year of carbon with the atmosphere. This exchange has two sources, while the breathing of animals and plants and fermentation of bacteria releases CO2 into the atmosphere; the photosynthesis (especially of green plants) fixes the carbon in the biomass. The biosphere plays an important role in the carbon cycle, as this compartment is directly influenced by human activity. While it is possible to interact with this compartment, on the one hand, deforestation and land use change can diminish carbon stocks (). On the other hand, tree planting and the protection of existing forests increase carbon stocks ().
Humans and climate change
The concentration of atmospheric carbon during the last 100-200 years increased significantly due to human activities (burning of fossil fuel, natural gas, charcoal, etc.). The burning of fossil fuels, which accumulated during millions of years, released huge amounts of CO2. Another reason for the increase of CO2 in the atmosphere comes from deforestation and forest fires, especially in tropical regions. This also causes fast release of CO2 sinks that were also accumulated during a long time (few years to several centuries based on burnt forest age) ().
By determining the contribution of CO2 to the atmosphere, we can deduce how the carbon cycle influences the global temperature. The rejection of CO2 of anthropogenic origins is responsible for 70% of the global warming, but in return, the atmospheric concentrations of CO2, the global temperature as well as the precipitation affect greatly the carbon cycle.
The carbon cycle feedback loop
Feedback in general is the process in which output from a system are “fed back” as inputs as part of a chain of cause-and-effect that forms a loop. For instance, by determining the contribution of CO2 to the atmosphere, the carbon cycle influences the global temperature. But, in return, the atmospheric concentrations of CO2, the global temperature as well as the precipitations influence several key elements of the carbon cycle.
At the oceanic levels, there is a complex feedback linked to the solubility of CO2. This feedback is negatively correlated to the temperature. In the case of global warming, more CO2 are liberated from oceans into the atmosphere, and therefore contribute to the global warming. This is called a positive feedback. However, the solubility of CO2 depends on its concentration in the atmosphere, thus limiting the effect of the feedback. The dissolution of CO2 in the oceans causes water acidification. Temperature changes are therefore influencing the activity of the plankton, which increases or decreases the oceanic ability to capture CO2 (; ).
In regards to vegetation and thus forests, if the ratio of photosynthesis increases with temperature and CO2, the ratio of the respiration will also increase with temperature. This effect on photosynthesis is generally positive. An increase in terrestrial vegetation has been observed in response to higher temperatures and CO2 levels in the atmosphere (IPCC, 2014 ). However, for certain vegetation types, it has been observed that the respiration increases more as a function of temperature rather than photosynthesis, this makes these ecosystems more as sources and not sinks of carbon in the long term.
Facing these environmental threats, the international community adopted several policies at the national, international and global level. The first United Nations summit concerning the environment took place in 1972 in Stockholm. It was during this summit that the United Nations Environment Program (UNEP) was created in order to debate ecological questions. The countries participating to this summit agreed to meet once each ten years in order to review the state of earth’s environment. Following that year, the most notable summits were as follows:
The Montreal protocol of 1987 which prohibited the chlorofluorocarbons gas use (CFC) as it can lead to the destructions of the atmosphere was successful as it allowed the decrease of atmospheric charges of the CFC (). This first success is still limited because of climate change with the massive injection of greenhouse gases, including firstly CO2, which could destabilize the stratosphere, and amplify the loss of the ozone layer in the atmosphere. The changing climates has socio-economic effects and these effects are already being felt, as they lead to the exodus of some populations worldwide, but also break the balance governing ecosystems and jeopardize the biodiversity of our planet. This led to the creation of the UN Framework Convention on Climate Change (UNFCCC), which came into force in 1994 following the Earth Summit in Rio de Janeiro in 1992. In the Rio Janeiro summit in 1992, the participants agreed on the necessity to stabilize atmospheric concentrations of greenhouse gases. The objective was to limit the abrupt changes to ecosystems, in order to have time to adapt. In 1997, 141 nations signed on the protocol of Kyoto, which engaged the committed nations to reduce by 5.2% their emissions of six greenhouse gases. Recently, the Copenhagen conference which brought together 191 countries, have ratified the UNFCCC. The UNFCCC stressed the importance of forests in regulating climate change and particularly of atmospheric CO2.
Countries in economic development have no commitments in this protocol along with the United States and the main carbon emitters who did not sign. Practically, this agreement allowed the creation of a carbon market. The states which surpass their quota in their carbon emission, can buy carbon credits from other nations that have not surpassed their carbon quotas. These credits allow the nations in need to emit more greenhouse gases. The objective was to motivate the nations to limit their greenhouse gases emissions by giving a monetary value to these emissions. The agreements of Copenhagen, which were signed in 2009, were renegotiations of the agreements of Kyoto. However, no binding commitments were made after the 2012, which marks the end of the Kyoto protocol. However, the 112 participating nations agreed to try and reduce the global temperatures rise by 2oC.
Role of forests in the carbon cycle
In the framework of the international agreements on the limitation of emission of greenhouse gases and temperature emissions, the case of forests and in tropical forest plays a major role. Carbon stocks in forests comprise above- and below-ground carbon in both living and dead organic matter. Globally, forests and soils are estimated to trap around 2.6 GtC/year. However, there are still many uncertainties about the carbon cycle. Indeed, Food and Agriculture Organization of the United Nations (FAO, 2008 ) estimates that the amount of carbon absorbed by the forests can vary between 0.9 and 4.3 GtC/year.
Tropical forests and the carbon stock
Carbon Stocks over land are distributed mostly between forests and northern latitudes (Figure 1.2), but are mostly found in forests, and more precisely in tropical forests. Indeed, studies suggests that tropical forests play a more important role in absorbing carbon with an absorption rate reaching as much as 1 GtC/year or about 40% of the total land based carbon absorption globally. However, tropical forests are principally located in developing countries (Amazon basin, Congo Basin, South-East Asia). These countries which are currently undergoing an economic and demographic growth, and therefore moving from forested to non-forested areas are causing a significant impact on the accumulation of greenhouse gases in the atmosphere, as has forest degradation caused by over-exploitation of forests for timber and wood fuel and intense grazing that is reducing forest regeneration. Therefore, during the 16th conference of the parties to the agreement of climate change of Cancun (2010), the United Nations program for Reducing Emissions from Deforestation and Forest Degradation (UN-REDD) was adopted. This program aims at protecting forests, preserve and increase forest carbon stocks and sustainable forest management. The REDD initiative and its three main supplementary activities are called REDD+. The basic principle of the REDD+ program is that financial compensation be paid by the developed countries to developing countries that manage to reduce their emissions at the national level. The REDD program is based on the fact that when a forest is damaged and destroyed, CO2 is released into the atmosphere. If we manage to reduce the rate of deforestation (complete disappearance of forests) or degradation (damaged forest due to exploitation), then it is possible to reduce the amount of released CO2. However, in order to calculate the magnitude of the reduction in CO2 emissions, it is necessary to create a baseline or reference base against which to compare actual emissions. Therefore it is necessary to be able to quantify the amounts of carbon contained in forests.
Link between carbon and forest biomass
Studies have stated that more than 40% of global vegetation carbon stocks are located in tropical forests (; ). However, forest carbon is not limited to trees and is distributed on average as follows: 45% of carbon is found in the soil, 11% in dead biomass or necromass, and 44% in biomass (both above- and below-ground) (FAO, 2000 ). Moreover, the above-ground biomass (AGB) is generally the most studied, as it is the most accessible. AGB is a biological material derived from living organisms, and it most often refers to plants. Biomass is carbon based and is composed of a mixture of organic molecules containing hydrogen, oxygen, and small quantities of several other atoms. The proportion of carbon in AGB varies depending on the forest type, wood composition, or the environment. However, it ranges between 0.43 and 0.55 (; ; ; ; ).
The importance of quantifying forest biomass
The interest in studying the AGB comes from the fact that the carbon in the AGB is susceptible to be released into the atmosphere by means of deforestation. In addition, land use change in tropical forests is responsible of 15-20% of global greenhouse emissions globally (; ). In contrast, if trees are to be planted, this means more carbon sequestration. However, this natural regeneration of the carbon stock will much likely take several decades (), and a plantation is not, by far, a natural forest. Moreover, even with forest degradation or regeneration, tropical forest can still undergo changes that affect AGB levels. For example, under influence of environmental changes, such as the increase of CO2 levels in the atmosphere. This increase of CO2 might increase the photosynthesis of trees and therefore increase the levels of carbon in trees (; ). Other environmental changes are caused by tree mortality, which can increase the necromass, and therefore affect the release of carbon in the atmosphere ().
As seen earlier, AGB measurement is an important task for better understanding of the carbon cycle. However, accurate measurements of biomass require weighing of the trees after cutting them. This method yields high biomass measurement accuracy however it is destructive and restrictive. Therefore it is necessary to find other methods for biomass estimation in a non-destructive manner.
Biomass estimation with optical and radar data
Currently, existing AGB estimation methods from remote sensing data are either limited in the vertical domain (sensor saturation at certain biomass levels using mainly radar and optical data) or in the horizontal domain (limited horizontal coverage using LiDAR data). Methods using radar and optical data for the estimation of AGB are successful in forests with low to medium levels of AGB (e.g. ; ; ; ; ). Indeed, current techniques based on passive optical sensing have shown limited sensitivity to biomass using medium to high resolution imagery when the biomass reaches intermediate levels (150-200 Mg/ha) (e.g. ; ). This is due to the optical data inability to detect variation in biomass density after complete closure of the canopy top, which can occur from low or intermediate biomass values (depending on forest characteristics). In contrast, the Fourier Transform Textural Ordination (FOTO) using very-high-resolution optical images have been used for non-saturating estimates of tropical forest biomass estimation. As such, this approach may provide higher sensitivity to biomass high levels (>600 t/ha) (e.g. ; ; ).
The synthetic aperture radar (SAR) systems such as PALSAR/ALOS, JERS-1 and SIR-C, as well as airborne SAR such as SETHI and E-SAR were also used as an alternative for biomass estimation. The radar signal saturation threshold with the biomass increases with the increase of the radar wavelength. Indeed, L-band SAR systems (wavelength about 25 cm) are limited to low and intermediate biomass levels, with maximum values reaching 150 t/ha (e.g. ; ; ; ; ; ). This saturation threshold of the radar signal depends on forest characteristics. According to Imhoff et al. ; the saturation levels are closer to 40 t/ha because the saturation thresholds occur before the regression maxima. In boreal forests, saturation levels were observed up to 150 t/ha. Baghdadi et al.  observed saturation levels of the ALOS/PALSAR L-band at biomass levels of 50 t/ha when estimating the biomass for Eucalyptus plantations in Brazil. Luckman et al. (; ) found a saturation point of 60 t/ha in the Central Amazon basin. Le Toan et al. ; Wu et al. ; and Dobson et al.  reported L-band signal saturation levels at 100 t/ha in coniferous forests. In boreal forests, higher saturation levels were observed reaching up to 150 t/ha using PALSAR (Sandberg et al. ). However, with higher radar wavelengths (P-band for example, wavelength about 70 cm) the use of SAR sensors may allow the estimation of biomass at higher biomass levels (). Imhoff et al.  examined AGB levels in broadleaf evergreen forests in Hawaii and coniferous forests in North America and Europe and found saturation levels of 100 Mg/ha for the P-band versus 40 Mg/ha for the L-band. Nizalapur et al.  found that the sensitivity of radar signal to biomass in a tropical dry deciduous forest increases for approximately 150 t/ha for the L-band to 200 t/ha for the P-band.
Table of contents :
1.1 General context
1.1.1 Global Carbon Cycle
1.1.2 Greenhouse gases and climate change
1.1.3 Global carbon cycle’s carbon sinks
1.1.4 Humans and climate change
1.1.5 The carbon cycle feedback loop
1.1.6 Current issues
1.2 Role of forests in the carbon cycle
1.2.1 Tropical forests and the carbon stock
1.2.2 Link between carbon and forest biomass
1.2.3 The importance of quantifying forest biomass
1.3 Biomass estimation
1.3.1 Biomass estimation with optical and radar data
1.3.2 Biomass estimation using allometric relations
1.3.3 Plot aggregate allometry for biomass estimation
1.4 Forest canopy height in relation to forest biomass
1.4.1 Canopy height estimation using radar and optical data
1.4.2 Canopy height estimation using LiDAR data
1.4.3 Spatial extrapolation of LiDAR canopy height estimates
1.5 Forest types in relation to forest biomass
1.6 Organization of the dissertation
1.6.2 Dissertation plan
CHAPTER 2: STUDY AREA AND DATASETS
2.1 Study area
2.2 Datasets description
2.2.1 Spaceborne LiDAR datasets
2.2.2 Airborne LiDAR datasets
126.96.36.199 Small footprint low density LiDAR dataset
188.8.131.52 Small footprint high density LiDAR dataset
2.2.3 Ancillary datasets
184.108.40.206 MODerate-resolution Imaging Spectroradiometer (MODIS) data
220.127.116.11 SRTM digital elevation model data
18.104.22.168 Geological map
22.214.171.124 Forest landscape types map
126.96.36.199 Average rainfall map
CHAPTER 3: CANOPY HEIGHT ESTIMATION IN FRENCH GUIANA WITH LIDAR ICESAT/GLAS DATA USING PRINCIPAL COMPONENT ANALYSIS AND RANDOM FOREST REGRESSIONS
3.2 Materials and methods
3.2.1 Lidar data processing and canopy height estimation
188.8.131.52 Processing the LD dataset
184.108.40.206 Processing the HD dataset
220.127.116.11 Comparison of canopy height estimates from the HD dataset using different estimation methods
18.104.22.168 Comparison of canopy height estimates from the LD and HD datasets
22.214.171.124 Glas data processing
3.2.2 Background on GLAS canopy height estimation
126.96.36.199 Direct method
188.8.131.52 Multiple regression models using GLAS and DEM metrics
184.108.40.206 Proposed techniques for canopy height estimation
220.127.116.11 Random forest regressions using principal components
3.3.1 Direct method
3.3.2 Multiple regression models
18.104.22.168 Using GLAS and DEM metrics
22.214.171.124 Using principal components
3.3.3 Random forest regressions
126.96.36.199 Using GLAS and DEM metrics
188.8.131.52 Using principal components
3.3.4 Model performance in different forest conditions
3.3.5 Error on the estimation of biomass
CHAPTER 4: FOREST CANOPY HEIGHT MAPPING OVER FRENCH GUIANA USING SPACE AND AIRBORNE LIDAR DATA
4.2 Materials and methods
4.2.1 Canopy height mapping using regression-kriging
4.2.2 Canopy height trend mapping using Random Forest regressions
4.2.3 Ordinary krigging of regression residuals
4.2.4 Effects of LiDAR sampling density on precision of the mapped canopy heights.
4.3.1 Canopy height mapping using Random Forest regressions
4.3.2 Canopy height estimation using regression-kriging
4.3.3 Relationship between LiDAR flight lines spacing and the accuracy on the kriged canopy height
CHAPTER 5: COUPLING POTENTIAL OF ICESAT/GLAS AND SRTM FOR THE DISCRIMINATION OF FOREST LANDSCAPE TYPES IN FRENCH GUIANA
5.2 Materials and methods
5.2.2 GLAS waveform processing
5.2.3 Canopy height and roughness index estimations
5.3 Results and discussion
5.3.1 Global analysis of the differences between the GLAS and SRTM elevations
5.3.2 Analysis of the differences between the GLAS and SRTM according to Hc and R
184.108.40.206 Differences between the GLAS and SRTM according to Hc
220.127.116.11 Differences between the GLAS and SRTM according to R
5.3.3 Random Forest classification results
5.3.4 Effect of the GLAS acquisition season
GENERAL CONCLUSIONS AND PERSPECTIVES
6.2.1 Canopy height estimation using GLAS
6.2.2 LiDAR canopy height mapping
18.104.22.168 Non-spatial canopy height mapping
22.214.171.124 Spatial canopy height mapping
126.96.36.199 Canopy height map resolution
188.8.131.52 Canopy height mapping sampling scheme
6.2.3 Above-ground biomass estimation
7.2 Description des jeux de données
7.2.1 Site d’étude
7.2.2 Base de données LiDAR satellitaire
7.2.3 Données du radiomètre spectral à moyenne résolution MODIS
7.2.4 Données issues du Modèle Numérique de Terrain MNT SRTM
7.2.5 Carte géologique
7.2.6 Carte des types de paysage forestier
7.2.7 Carte de précipitation
7.3 Estimation de la hauteur des arbres à partir des données GLAS
7.3.1 Contexte de l’estimation de la hauteur des arbres en utilisant GLAS
7.3.2 Techniques proposées pour l’estimation de la hauteur des arbres
7.4 La spatialisation de la hauteur des arbres LiDAR
7.4.1 Contexte sur la technique régression-krigeage
7.4.2 La cartographie de la hauteur des arbres en utilisant la régression krigeage
7.4.3 Relation entre l’espacement des lignes de vol LiDAR et la précision de la hauteur des arbres krigée
7.5 Le potentiel du couplage GLAS et SRTM pour la discrimination des types de paysage forestier
7.5.1 Classifications des empreintes GLAS
7.5.2 Les effets de la saison sur les acquisitions GLAS
7.6 Conclusions et perspectives
184.108.40.206 La spatialisation de la hauteur des arbres à partir du LiDAR
220.127.116.11 L’estimation de la biomasse