Thesis (M.S.)--Kent State University, 2008. / Title from PDF t.p. (viewed Sept. 28, 2009). Advisor: Donna Witter. Keywords: chlorophyll; Lake Erie; remote sensing; algorithm; atmospheric correction. Includes bibliographical references (p. 58-66).

<p> Savannas are mixed tree-grass systems and as one of the world's largest biomes represent an important component of the Earth system affecting water and energy balances, carbon sequestration and biodiversity as well as supporting large human populations. Savanna vegetation structure and its distribution, however, may change because of major anthropogenic disturbances from climate change, wildfire, agriculture, and livestock production. The overstory and understory may have different water use strategies, different nutrient requirements and have different responses to fire and climate variation. The accurate measurement of the spatial distribution and structure of the overstory and understory are essential for understanding the savanna ecosystem. </p><p> This project developed a workflow for separating the dynamics of the overstory and understory fractional cover in savannas at the continental scale (Australia, South America, and Africa). Previous studies have successfully separated the phenology of Australian savanna vegetation into persistent and seasonal greenness using time series decomposition, and into fractions of photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV) and bare soil (BS) using linear unmixing. This study combined these methods to separate the understory and overstory signal in both the green and senescent phenological stages using remotely sensed imagery from the MODIS (MODerate resolution Imaging Spectroradiometer) sensor. The methods and parameters were adjusted based on the vegetation variation.</p><p> The workflow was first tested at the Australian site. Here the PV estimates for overstory and understory showed best performance, however NPV estimates exhibited spatial variation in validation relationships. At the South American site (Cerrado), an additional method based on frequency unmixing was developed to separate green vegetation components with similar phenology. When the decomposition and frequency methods were compared, the frequency method was better for extracting the green tree phenology, but the original decomposition method was better for retrieval of understory grass phenology. Both methods, however, were less accurate than in the Cerrado than in Australia due to intermingling and intergrading of grass and small woody components. </p><p> Since African savanna trees are predominantly deciduous, the frequency method was combined with the linear unmixing of fractional cover to attempt to separate the relatively similar phenology of deciduous trees and seasonal grasses. The results for Africa revealed limitations associated with both methods. There was spatial and seasonal variation in the spectral indices used to unmix fractional cover resulting in poor validation for NPV in particular. The frequency analysis revealed significant phase variation indicative of different phenology, but these could not be clearly ascribed to separate grass and tree components. </p><p> Overall findings indicate that site-specific variation and vegetation structure and composition, along with MODIS pixel resolution, and the simple vegetation index approach used was not robust across the different savanna biomes. The approach showed generally better performance for estimating PV fraction, and separating green phenology, but there were major inconsistencies, errors and biases in estimation of NPV and BS outside of the Australian savanna environment.</p>

Thèse (M.Sc.)--Université de Sherbrooke (Canada), 2008. / Titre de l'écran-titre (visionné le 1 février 2007). In ProQuest dissertations and theses. Publié aussi en version papier.

<p> In cities, microclimates are created by local mixtures of vegetation, constructed materials, vertical structure, and moisture, with significant consequences for human health, air quality, and resource use. Vegetation can moderate microclimates through evapotranspiration, however this function is dependent on local conditions so its effect may vary over space and time. This dissertation used hyperspectral and thermal remote sensing imagery to derive key observations of urban physical and biophysical properties and model urban microclimates across the megacity of Los Angeles. In Chapter 1, I used Multiple Endmember Spectral Mixture Analysis (MESMA) to map sub-pixel fractions of different vegetation types, as well as other types of urban cover, at 4 m and 18 m resolution over Santa Barbara, California (Wetherley et al., 2017). Fractional estimates correlated with validation fractions at both scales (mean R² = 0.84 at 4 m and R² = 0.76 at 18 m), with accuracy affected by image spatial resolution, endmember spatial resolution, and class spectral (dis)similarity. Accuracy was improved by using endmembers measured at multiple spatial resolutions, likely because they incorporated additional spectral variability that occurred across spatial scales. In Chapter 2, I applied this methodology to derive sub-pixel cover for the greater Los Angeles metropolitan area (4,466 km²) (Wetherley et al., 2018). Further improvement in quantifying sub-pixel vegetation types was achieved by modifying the MESMA shade parameter. Land surface temperature (LST), derived from thermal imagery, was used to model temperature change along vegetation fractional gradients, with slopes of LST change showing significant differences between trees and turfgrass (p &lt; 0.001). Expected per-pixel LST was derived from these gradients based on sub-pixel composition, and when compared to measured LST was found to deviate with a standard deviation of 3.5 &deg;C across the scene. These deviations were negatively related to irrigation and income, while building density was observed to affect tree LST more than it affected turfgrass LST. In Chapter 3, I used the map of Los Angeles landcover, along with data from LiDAR, GIS, and WRF climate variables, to parameterize an urban climate model (Surface Urban Energy and Water Balance Scheme: SUEWS) for 2,123 neighborhoods (each 1 km²) across Los Angeles. Modeled latent fluxes were correlated with remote sensing LST (R² = 0.39) collected over a period of 5 hours, with an overall diurnal pattern modified by irrigation timing. Spatial variability across the study area was related to local landcover, with albedo and vegetation fraction strongly influencing latent and sensible fluxes. A strong regional climatic gradient was observed to affect latent fluxes based on coastal proximity. Overall, this dissertation quantifies the key drivers of urban vegetation function in a large city, and further demonstrates the potential of hyperspectral and thermal imagery for observing city scale surface and microclimate variability.</p><p>

<p> High-spatial resolution measurements of vegetation structure are needed for improving understanding of ecosystem carbon, water and nutrient dynamics, the response of ecosystems to a changing climate, and for biodiversity mapping and conservation, among many research areas. Our ability to make such measurements has been greatly enhanced by continuing developments in remote sensing technology&mdash;allowing researchers the ability to measure numerous forest traits at varying spatial and temporal scales and over large spatial extents with minimal to no field work, which is costly for large spatial areas or logistically difficult in some locations. Despite these advances, there remain several research challenges related to the methods by which three-dimensional (3D) and spectral datasets are joined (remote sensing fusion) and the availability and portability of systems for frequent data collections at small scale sampling locations. Recent advances in the areas of computer vision structure from motion (SFM) and consumer unmanned aerial systems (UAS) offer the potential to address these challenges by enabling repeatable measurements of vegetation structural and spectral traits at the scale of individual trees. However, the potential advances offered by computer vision remote sensing also present unique challenges and questions that need to be addressed before this approach can be used to improve understanding of forest ecosystems. For computer vision remote sensing to be a valuable tool for studying forests, bounding information about the characteristics of the data produced by the system will help researchers understand and interpret results in the context of the forest being studied and of other remote sensing techniques. This research advances understanding of how forest canopy and tree 3D structure and color are accurately measured by a relatively low-cost and portable computer vision personal remote sensing system: 'Ecosynth'. Recommendations are made for optimal conditions under which forest structure measurements should be obtained with UAS-SFM remote sensing. Ultimately remote sensing of vegetation by computer vision offers the potential to provide an 'ecologist's eye view', capturing not only canopy 3D and spectral properties, but also seeing the trees in the forest and the leaves on the trees.</p>

This paper examines the issues that arise in the use of visual interpretation of Landsat data during the analysis, classification and mapping of the natural vegetation of the semi-arid Northern Cape. Initial research involved the classifying and mapping of the vegetation using conventional methods. A vegetation map, accompanying legend and descriptive key were produced. The problems encountered during this process, and the constraints of manpower, time and funds, stimulated the investigation of Landsat imagery as a means of improving the speed and accuracy of vegetation classification and mapping. A study area comprising one Landsat scene and which met certain requirements was selected: a) The area had already been surveyed and mapped at a scale of 1:250 000. b) As many vegetation units as possible were included. c) There was maximum diversity, complexity and variability in terms of soil, geology and terrain morphology. Initially a suitable mapping scale was selected, viz. 1:250 000, as it met the requirements of nature conservation authorities and agricultural planners. The scales of survey and remote sensing were based on this. The basic unit of survey was the 1:50 000 topographical map and satellite imagery at a scale of 1:250 000 was found to meet the requirements of reconnaissance level mapping. The usefulness of Landsat imagery was markedly affected by the quality of image production and enhancement. Optimum image production was vitally important and to this end, interaction between the user and the operations engineer at the Satellite Applications Centre, Hartebeeshoek was essential. All images used, were edge-enhanced and systematically corrected. While these procedures were costly, they proved to be fundamental to the success of the investigation. Precision geometric correction was not required for reconnaissance level investigation. The manual superimposition of the UTM grid, using ground control points from 1:250 000 topographical maps, proved to be accurate and convenient. Pattern recognition on single-band, panchromatic imagery was difficult. The scene lacked crispness and contrast, and it was evident that black and white imagery did not satisfy the objectives of the study. Three-band false colour composite imagery was superior to single-band imagery in terms of clarity and number of cover classes. The addition of colour undoubtedly facilitated visual interpretation. False colour composite imagery was investigated further to establish which year, season and possibly time of season would best suit the objectives of the investigation. It was found that the environmental parameters affecting reflectance are relatively stable over time and it was not necessary to acquire imagery of the same year as field surveys. However, the year of imagery should be chosen so that similar climatic conditions prevail. While, in certain instances, imagery captured during winter had advantages in separating complex mosaics, summer imagery was superior in most respects. Furthermore, given "normal" climatic conditions, the ideal period during which there was maximum contrast between and within ground classes, and thus spectral classes, was narrowed to mid-January to mid-April. Units which were acceptably heterogeneous (relatively homogeneous) in terms of reflectance levels were delineated manually on the image. This delineation was done at three levels of complexity and the units were compared with the vegetation map. A series of field trips aided the interpretation of the images, especially where discrepancies occurred between the map and the image. In general, there was a close degree of correspondence between the prepared vegetation map and the delineated image. Field investigation revealed the image units to be more accurate than those on the vegetation map, and the image served to highlight the inadequacies inherent in classifying and mapping vegetation of extensive areas with limited resources.

This thesis explores techniques which ultimately strive to optimize production systems in rangeland areas of southern Africa. By linking spatially significant, satellite derived data to practical measurements of vegetation structure, valuable insight has been derived on processes of ecosystem function, in the Sabi Sand Game Reserve. A broad ecosystem response mechanism has been established from a conventional Normalized Differentiation Vegetation Index (NDVI). By responding to increases in production, which are driven by disturbance, this index has allowed quantitative systems theory in savanna to be tested and refined. Methods of biomass and production estimation which are specifically designed to reduce the cost and time involved with the more conventional method of destructive harvesting have been tested in the savanna at the Sabi Sand Game Reserve. Results from these estimates relate well with data derived through destructive harvesting in structurally similar savanna. Moreover, by relating the above-ground woody production estimates to remere sensing indices, it was possible to demonstrate that the problem of extrapolation, universal to most biomass and production studies can be overcome. Since remote sensing encompasses an array of tools fundamental to rangeland inventory, monitoring and management, valuable spatially significant information pertaining to ecosystem structure and function has been provided for managers in the Sabi Sand Game Reserve.

The Kafue Flats floodplain wetland system in southern Zambia is under increasing climate and human pressures. Firstly, drought episodes appear more prevalent in recent years in the region and secondly, two dams were built on the lower and upper ends of the wetland in 1972 and 1978, respectively, across the Kafue River which flows through the wetland. The study uses multi-temporal remote sensing to assess change in extent and vigour of green vegetation, and extent of water bodies and dry land cover on the Kafue Flats. The change detection's management value is assessed. Four normalised, co-registered digital Landsat images from 24 September 1984, 3 September 1988, 12 September 1991 and 20 September 1994 were used. The main change detection method used was comparison of classifications, supplemented by Normalised Difference Vegetation Index (NDVI) and Principal Component Analysis (PCA) change detection. Ancillary land use and environmental data were used in interpreting the change in the context of cause and effect. The results indicate inconsistent trends in the changes of most land cover classes, as a result of manipulation of the wetland by man through annual variations in the timing and magnitude of regulated flows into the wetland, as well as burning. However, the results also show spatial reduction in the wetland's dry season dense green reed-grass vegetation in upstream sections which are not affected by the water backing-up above of the lower dam. Sparse green vegetation is replacing the dense green vegetation in these upstream areas. It is inferred that this dry season degradation of the wetland threatens bird species which may use the reeds for dry season nesting. It is proposed that ground surveying and monitoring work at the micro-habitat level is necessary to ascertain the implications of the losses. It is concluded that, in spite of difficulties, multi-temporal remote sensing has a potential role in wetland change assessment on the Kafue Flats at the community level, but that it needs to be supplemented by targeted, micro-habitat level ground surveys.

<p>U&nbsp; radu&nbsp; je&nbsp; analizirano&nbsp; gazdovanje&nbsp; &scaron;umama&nbsp; u&nbsp; za&scaron;tićenim&nbsp; područjima&nbsp; u&nbsp; Srbiji,<br />posmatrajući:&nbsp; 1)&nbsp; ekolo&scaron;ke&nbsp; efekte&nbsp; gazdovanja,&nbsp; i&nbsp; 2)&nbsp; procese&nbsp; prilagođavanja&nbsp; gazdovanja&nbsp; konzervacionim&nbsp; potrebama.&nbsp; Povr&scaron;ina&nbsp; za&scaron;tićenih&nbsp; područja&nbsp; je&nbsp; značajno&nbsp; porasla,&nbsp; ali&nbsp; ima malo&nbsp; informacija o njihovom doprinosu u realizaciji konzervacionih&nbsp; ciljeva. Po&scaron;to se u većini za&scaron;tićenih &scaron;umskih područja u Evropi aktivno gazduje, i dozvoljeno je kori&scaron;ćenje drvne&nbsp; biomase,&nbsp; informacije&nbsp; o&nbsp; efektima&nbsp; gazdovanja&nbsp; su&nbsp; neophodne.&nbsp; Glavna&nbsp; barijera&nbsp; u ekolo&scaron;kim&nbsp; evaluacijama&nbsp; gazdovanja&nbsp; je&nbsp; nepostojanje&nbsp; indikatora&nbsp; kojima&nbsp; bi&nbsp; se&nbsp; mogle izmeriti promene u ekosistemu, a&nbsp; koje&nbsp; nastaju kao rezultat primenjenih mera&nbsp; gazdovanja. Poslednjih&nbsp; godina&nbsp; se&nbsp; sve&nbsp; vi&scaron;e&nbsp; ističe&nbsp; potencijal&nbsp; primene&nbsp; funkcionalnih&nbsp; indikatora,&nbsp; tj. indikatora&nbsp; koji&nbsp; se&nbsp; oslanjaju&nbsp; na&nbsp; informacije&nbsp; o&nbsp; funkcionalnim&nbsp; karakteristikama&nbsp; vrsta.&nbsp; U radu je stoga primenjen funkcionalni pristup za analizu ekolo&scaron;kog efekta gazdovanja. U te&nbsp; svrhe&nbsp; su&nbsp; odabrane&nbsp; karakteristike&nbsp; koje&nbsp; se&nbsp; povezuju&nbsp; sa&nbsp; sposobno&scaron;ću&nbsp; vrsta&nbsp; za&nbsp; brzo<br />usvajanje&nbsp; ili konzervisanje resursa, tj.&nbsp; određuju odgovor vrsta na promene u dostupnim resursima, a koje, na primer,&nbsp; nastaju usled uklanjanja drvne biomase: visina, specifična povr&scaron;ina lista, sadržaj suve materije lista, sadržaj ukupnog azota i ukupnog fosfora. Iste se&nbsp; dovode&nbsp; i&nbsp; u&nbsp; vezu&nbsp; sa&nbsp; primarnom&nbsp; produktivno&scaron;ću&nbsp; i&nbsp; dekompozicijom,&nbsp; procesima&nbsp; od značaja&nbsp; za&nbsp; očuvanje&nbsp; integriteta&nbsp; ekosistema.&nbsp; Po&scaron;to&nbsp; analiza&nbsp; odgovora&nbsp; funkcionalnih indikatora&nbsp; zahteva&nbsp; duži&nbsp; vremenski&nbsp; period&nbsp; posmatranja,&nbsp; u&nbsp; radu&nbsp; je&nbsp; osmi&scaron;ljen&nbsp; drugačiji pristup za izvođenje eksperimenta. U te svrhe su iskori&scaron;ćeni Landsat satelitski snimci, tj. tri snimka sa vremenskim razmakom od 10 godina: 1994.,&nbsp; 2005. i 2015. godina. Izvr&scaron;ena je&nbsp; digitalna&nbsp; klasifikacija&nbsp; snimaka&nbsp; prema&nbsp; sastavu&nbsp; &scaron;uma,&nbsp; nakon&nbsp; čega&nbsp; su&nbsp; detektovane promene&nbsp; u&nbsp; sastavu&nbsp; &scaron;uma&nbsp; za&nbsp; period&nbsp; od&nbsp; ukupno&nbsp; 20&nbsp; godina.&nbsp; Ovi&nbsp; podaci&nbsp; su&nbsp; spojeni&nbsp; sa podacima&nbsp; o&nbsp;&nbsp; funkcionalnim&nbsp; karakteristikama&nbsp; vrsta,&nbsp; kako&nbsp; bi&nbsp; se&nbsp; utvrdile&nbsp; promene&nbsp; u funkcionalnoj kompoziciji. Poslednji korak je bio formiranje modela &scaron;umske krune, kroz koji&nbsp; je&nbsp; određen&nbsp; intenzitet&nbsp; seče.&nbsp; Koristeći&nbsp; podatke&nbsp; o&nbsp; promenama&nbsp; u&nbsp; gustini&nbsp; krune&nbsp; i<br />promenama&nbsp; u&nbsp; funkcionalnoj&nbsp; kompoziciji,&nbsp; sproveden&nbsp; je&nbsp; eksperiment,&nbsp; tj.&nbsp; analiza&nbsp; efekta različitih&nbsp; nivoa&nbsp; intenziteta&nbsp; seče&nbsp; na&nbsp; promene&nbsp; u&nbsp; funkcionisanju&nbsp; &scaron;umskog&nbsp; ekosistema.&nbsp; Za sprovođenje ovog eksperimenta je odabran samo jedan deo Fru&scaron;ke gore, jer je cilj bio da se osmisli&nbsp; pristup za evaluaciju, prilagođen trenutnom konceptu gazdovan ja za&scaron;tićenim &scaron;umskim&nbsp; područjima,&nbsp; i&nbsp; ispita&nbsp; njegova&nbsp; praktičnost.&nbsp; Kao&nbsp; propratni&nbsp; podatak&nbsp; ovoj&nbsp; analizi, sprovedena&nbsp; je&nbsp; i&nbsp; komparativna&nbsp; analiza&nbsp; upravljača&nbsp; za&scaron;tićenih&nbsp; &scaron;umskih&nbsp; područja&nbsp; u&nbsp; Srbiji, kako&nbsp; bi&nbsp; se&nbsp; ispitao&nbsp; nivo&nbsp; ulaganja&nbsp; u&nbsp; konzervacione&nbsp; sposobnosti,&nbsp; kao&nbsp; ključnog&nbsp; procesa&nbsp; u implementaciji&nbsp;&nbsp; konzervacionih&nbsp; ciljeva.&nbsp; U&nbsp; digitalnoj&nbsp; klasifikaciji&nbsp; je&nbsp; postignuta&nbsp; visokapreciznost,&nbsp; sa&nbsp; ukupnom&nbsp; precizno&scaron;ću&nbsp; 94,5%&nbsp; i&nbsp; Kapa&nbsp; koeficijentom&nbsp; 0,93.&nbsp; Potpuno spektralno razdvajanje je postignuto samo za&nbsp; <em>Q. petraea&nbsp;</em> od<em> Tilia tomentosa</em>, i&nbsp; sastojina u<br />kojima se kao dominantne pojavljuju <em>F. silvatica i Tilia tomentosa</em>. Utvđeno je prodiranje<em> F.&nbsp; moesiaca&nbsp;&nbsp;&nbsp; </em>u&nbsp; &scaron;ume&nbsp; <em>Q.&nbsp; petraea,&nbsp;</em> i&nbsp; apsolutno&nbsp; &scaron;irenje&nbsp; vrste&nbsp; <em>Tilia&nbsp; tomentosa,</em>&nbsp; posebno&nbsp; u periodu&nbsp; nakon&nbsp; 2005.&nbsp; godine,&nbsp; a&nbsp; koja&nbsp; prema&nbsp; podacima&nbsp; o&nbsp; karakteristikama&nbsp; vrsta&nbsp; ima najveću&nbsp; kompetetivnu&nbsp; sposobnost&nbsp; u&nbsp; odnosu&nbsp; na&nbsp; druge&nbsp; ispitivane&nbsp; vrste&nbsp; drveća.&nbsp; Rezultati modela gustine krune ukazuju na postepeno proređivanje &scaron;ume od 1994. ka 2015. godini, sa&nbsp; potpunim&nbsp; nestankom&nbsp; &scaron;uma&nbsp; guste&nbsp; krune&nbsp; na&nbsp; prelazu&nbsp; između&nbsp; 1994.&nbsp; i&nbsp; 2005.&nbsp; godine. Utvrđen je značajan efekat proređivanja &scaron;ume na promene u funkcionalnoj kompoziciji. Sa&nbsp; vi&scaron;im&nbsp; intenzitetom&nbsp; proređivanja,&nbsp; CWM&nbsp; indikator&nbsp; se&nbsp; pomerio&nbsp; od&nbsp; konzervativnih karakteristika ka onima koje ukazuju na dominaciju vrsta sklonih brzom sticanju resursa. Prelaz&nbsp; koji&nbsp; je&nbsp; imao&nbsp; značajnog&nbsp; efekta&nbsp; na&nbsp; promene&nbsp; je&nbsp; prelaz&nbsp; iz&nbsp; &scaron;ume&nbsp; sa&nbsp; gustinom&nbsp; krune većom&nbsp; od&nbsp; 65%&nbsp; u&nbsp; &scaron;umu&nbsp; gustine&nbsp; krune&nbsp; između&nbsp; 50-65%,&nbsp; a&nbsp; &scaron;to&nbsp; bi&nbsp; moglo&nbsp; ukazivati&nbsp; i&nbsp; na prelaz u drugi režim funkcionisanja&nbsp; ekosistema.&nbsp; Konzervacioni ciljevi su jo&scaron; uvek slabo integrisani u gazdovanje &scaron;umama u za&scaron;tićenim područjima.</p> / <p>In&nbsp; this&nbsp; work,&nbsp; the&nbsp; forest&nbsp; management&nbsp;&nbsp; in&nbsp; protected&nbsp; areas&nbsp; of&nbsp; Serbia&nbsp; was&nbsp; analysed, following:&nbsp; 1)&nbsp; ecologic&nbsp; effects&nbsp; of&nbsp; management&nbsp; and&nbsp; 2)&nbsp; investments&nbsp; in&nbsp; capabilities essential&nbsp; for&nbsp; integration&nbsp; of&nbsp; conservation&nbsp; objectives.&nbsp; The&nbsp; size&nbsp; of&nbsp; protected&nbsp; areas&nbsp; has significantly&nbsp; grown,&nbsp; but&nbsp; there&nbsp; is&nbsp; a&nbsp; little&nbsp; information&nbsp; on&nbsp; their&nbsp; contribution&nbsp; to&nbsp; real conservation&nbsp; goals.&nbsp; Since&nbsp; in&nbsp; most&nbsp; of&nbsp; European&nbsp; protected&nbsp; forest&nbsp; areas&nbsp; is&nbsp; employed active management, the information on ecological&nbsp; effects of management&nbsp; is necessary. The&nbsp; main&nbsp; barrier&nbsp; in&nbsp; ecologic&nbsp; evaluations&nbsp; of&nbsp; managing&nbsp; is&nbsp; the&nbsp; absence&nbsp; of&nbsp; indicators, which&nbsp; could&nbsp; measure&nbsp; the&nbsp; changes&nbsp; in&nbsp; ecosystem,&nbsp; resulting&nbsp; from&nbsp; applied&nbsp; measures. Recently, the potentiality of&nbsp; functional indicators is more&nbsp; emphasized, i.e. indicators, which lean on information about functional traits of species. So, in this work functional approach&nbsp; was&nbsp; taken&nbsp; to&nbsp; analyse&nbsp; ecologic&nbsp; effects&nbsp; of&nbsp; forest&nbsp; management.&nbsp; For&nbsp; this&nbsp; sake were chosen characteristics, which are connected to capabilities of specie s to acquire or&nbsp; conserve&nbsp; resources,&nbsp; i.e.&nbsp; define&nbsp; the&nbsp; answer&nbsp; of&nbsp; the&nbsp; species&nbsp; on&nbsp; changes&nbsp; in&nbsp; available resources,&nbsp; caused&nbsp; by&nbsp; elimination&nbsp; of&nbsp; wooden&nbsp; mass,&nbsp; i.e.&nbsp; logging:&nbsp; height,&nbsp; specific&nbsp; leaf area, leaf dry matter cont ent, nitrogen content and&nbsp; phosphorus&nbsp; content. The same are connected to the processes of significance for conservation of ecosystem integrity&nbsp; (net primary&nbsp; productivity&nbsp; and&nbsp; decomposition).&nbsp; Since&nbsp; the&nbsp; analysis&nbsp; of&nbsp; functional&nbsp; indicators response&nbsp; demands&nbsp; a&nbsp; longer&nbsp; period&nbsp; of&nbsp; observation,&nbsp; another&nbsp; attitude of&nbsp; performing&nbsp; the experiment was conceptualized. For that purpose Landsat&nbsp; satellite&nbsp; image was&nbsp; used,&nbsp; i.e. three&nbsp; images&nbsp; in&nbsp; interval&nbsp; of&nbsp; ten&nbsp; years:&nbsp; 1994,&nbsp; 2005,&nbsp; and&nbsp; 2015.&nbsp; A&nbsp; digital&nbsp; image classification of&nbsp; forest composition was&nbsp; performed, after which the&nbsp; changes&nbsp; in&nbsp; forest composition&nbsp; were&nbsp; detected&nbsp; over&nbsp; the&nbsp; period&nbsp; of&nbsp; 20&nbsp; years.&nbsp; These&nbsp; data&nbsp; were&nbsp; then connected with the data on functional characteristics of species to determine changes in functional composition.&nbsp;&nbsp; The last step&nbsp; was forming of forest canopy&nbsp; density model, through&nbsp; which&nbsp; was&nbsp; determined&nbsp; the&nbsp; intensity&nbsp; of&nbsp; logging.&nbsp; Using&nbsp; data&nbsp; on&nbsp; changes&nbsp; of forest&nbsp; canopy&nbsp; density&nbsp; model&nbsp; and&nbsp; changes&nbsp; in&nbsp; functional&nbsp; composition,&nbsp; an&nbsp; experiment was&nbsp; performed,&nbsp; i.e.&nbsp; analysis&nbsp; of&nbsp; effects&nbsp; of&nbsp; different&nbsp; levels&nbsp; of&nbsp; logging&nbsp; intensity&nbsp; on changes&nbsp; in&nbsp; forest&nbsp; ecosystem&nbsp; function.&nbsp; For&nbsp; the&nbsp; performing&nbsp; of&nbsp; this&nbsp; experiment&nbsp; was chosen&nbsp; just&nbsp; a&nbsp; part&nbsp; of&nbsp; Fruska&nbsp; gora,&nbsp; as&nbsp; the&nbsp; goal&nbsp; was&nbsp; to&nbsp;&nbsp; try&nbsp; to&nbsp; construct the&nbsp; evaluation approach,&nbsp; adapted to&nbsp; momentary&nbsp; concept of&nbsp; managing&nbsp; in&nbsp; protected&nbsp; forest&nbsp; areas&nbsp; and inspect&nbsp; its&nbsp; feasibility.&nbsp; As&nbsp; an&nbsp;&nbsp; accompanying&nbsp; data&nbsp; with&nbsp; this&nbsp; analysis&nbsp; was&nbsp; performed&nbsp; a comparative&nbsp; analysis&nbsp; of protected forest area managers&nbsp; in Serbia, to&nbsp; examine&nbsp; the level of&nbsp; investments&nbsp; in&nbsp; conservation&nbsp; capabilities,&nbsp; as&nbsp; the&nbsp; key&nbsp; process&nbsp; in&nbsp; implementing conservation goals.</p>

生長於潮間帶的紅樹林是熱帶和亞熱帶地區最具生產力的生態系統之一。香港擁有十個紅樹品種，其覆蓋面積約共三百五十公頃。位於香港西北面的米埔是現時香港最大的紅樹林區。這片紅樹林及其鄰近濕地於一九九五年被列為拉姆薩爾重要的濕地。隨著經濟的迅速發展、污染及一些不可持續的開發，全球紅樹林的面積不斷地萎縮。而香港的紅樹也正面對城市發展及基建的直接威脅。因此，了解及監測紅樹林的生長狀況、覆蓋面積的轉變是紅樹林保育的基礎。遙感是具有成本效益和能提供及時數據的技術，在紅樹林的生態保育及監測上發揮著重要功能。 / 是次研究選擇位於米埔的紅樹林區。通過結合高光譜和雷達數據以及實地磡測，以達到三個目的。第一，利用模式辨認分析找出可提高品種辨識度的光譜帶及雷達數據。第二，把挑選出來的光譜帶及雷達數據組合，利用不同的分類法包括最大概似法、决策樹 C5.0演算法、類神經網路及支持向量機進行紅樹林的品種分類，並籍此測試各分類法的精度。第三，利用植被指數及雷達數據中取得的參數為獨立變量，而在野外點測的葉面積指數 (LAI) 為因變量，通過迴歸分析以估算整片紅樹林的葉面積指數，籍此了解紅樹林現時的生物物理狀況。 / 根據特徵選擇的結果，位於高光譜數據中的綠波段 (570nm, 580nm, 591nm及601nm)、紅波段 (702nm)、紅邊位 (713nm)、近紅外波段 (764nm及774nm)、 短波紅外波段 (1276nm, 1316nm及1629nm) 以及在不同季節取得的過濾後向散射數據是最能辨識品種差異。 / 據品種分類的結果顯示，單用多時後向散射特徵數據存在很大誤差。而在大多的情況下，單用光譜數據比起混合光譜及後向散射數據的分類表現為佳。但對於某些品種來說，後向散射數據能給予比較準確的預測。另外，在同數據組合下，分類法在訓練精度上沒有多大的分別。除了類神經網路分類法以外，其他分類法的測試精度總比其訓練精度低。這說明類神經網路模型比起其他分類法的模型要為穩定，而决策樹模型則被過度訓練。根據生產者及使用者精度分析，因為缺乏足夠的訓練樣本，桐花樹及海桑屬的精度較其他品種為低。 / 據不同植被指數的簡單線性迴歸模型顯示，利用三角植被指數 (TVI)及修正葉綠素吸納比例指數一 (MCARI 1) 對於葉面積指數的估算是最準確。相反地，葉面積指數與從雷達數據中取得的參數關係則比較弱。這表示單用雷達參數不能對葉面積指數進行準確的估算。在結合植被指數及雷達參數的多元逐步迴歸分析下，三角植被指數及在灰度共生矩陣下得出的角二階矩參數能減低葉面積指數估算的誤差。總結以上兩項分析，光譜及雷達數據在紅樹林的品種分類及葉面積指數估算上有互補的作用。 / Mangrove is one of the most productive ecosystems flourished in the intertidal zone of tropical and subtropical regions. Hong Kong has ten true mangrove species covering an approximate area of 350 hectares. Mai Po locating in the northwestern part of Hong Kong nourishes the largest mangrove stand and it was listed as a Wetland of Importance under the Ramsar Convention in 1995. Over the years, areas of mangrove have been shrinking globally due to development, pollution, and other unsustainable exploitation and Hong Kong was no exception. In Hong Kong, mangroves are usually sacrificed for urban development and infrastructure construction. Therefore, it is crucial to monitor their growth conditions, change of extent and possible unsustainable practices threatening their existence. Remote sensing being a cost-effective and timely tool for vegetation conservation is most suitable for such purpose. / Taking Mai Po as study area, this study acquired satellite-borne hyperspectral and radar data supplemented with in situ field survey to achieve three purposes. First, features from the remotely-sensed data that are significant to species discrimination were identified through pattern recognition. Second, selected features grouped into different subsets were used to delineate the boundary of mangrove species through supervised classification. In the meantime, classifiers including maximum likelihood (ML), decision tree C5.0 (DT), artificial neural network (ANN) and support vector machines (SVM) were tested for their accuracy performance. The third purpose is to understand the current biophysical condition of mangrove through leaf area index (LAI) modeling by regressing field-measured LAI against vegetation indices, backscatter and textural measures. / Results from feature selection revealed that hyperspectral narrowbands locating in green at 570nm, 580nm, 591nm, 601nm; red at 702nm; red-edge at 713nm; near infrared at 764nm and 774nm and shortwave infrared at 1276nm, 1316nm and 1629nm as well as the multi-temporal filtered backscatter captured in different seasons have high sensitivity to species difference. / Species-based classification using multi-temporal backscatter features alone do not provide a satisfactory accuracy. Comparatively, results from pure spectral bands have better overall accuracy than that from combining spectral and radar features. However, radar backscatter does improve accuracy of some species. Besides, all classifiers had similar variations of training accuracy under the same feature subset. However, the testing accuracy is much lower with the exception of ANN. Performance of ANN was more stable and robust than other classifiers while serious overtraining occurs for the DT classifier. Moreover, most species were mapped accurately as revealed by the producer’s and user’s accuracy with the exception of A. corniculatum and Sonneratia spp. due to deficiency of training samples. / Simple linear regression model with VIs revealed that triangular vegetation index (TVI) and modified chlorophyll absorption ratio index 1 (MCARI1) had the best relationship with LAI. However, weak relationship was found between field- measured LAI and radar parameters suggesting that radar parameters cannot be used as single predictor for LAI. Results from stepwise multiple regression suggested that TVI combined with GLCM-derived angular second moment (ASM) can reduce the estimation error of LAI. To conclude, the study has demonstrated spectral and radar data are complementarity for accurate species discrimination and LAI mapping. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Wong, Kwan Kit. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2012. / Includes bibliographical references (leaves 434-472). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese. / ACKNOWLEDGEMENTS --- p.II / ABSTRACT --- p.IV / 論文摘要 --- p.VI / TABLE OF CONTENTS --- p.VIII / LIST OF ABBREVIATIONS --- p.XIII / LIST OF TABLES --- p.XV / LIST OF FIGURES --- p.XVIII / Chapter CHAPTER 1 --- INTRODUCTION --- p.1 / Chapter 1.1 --- BACKGROUND TO THE STUDY --- p.1 / Chapter 1.1.1 --- Mangrove Mapping and Monitoring --- p.1 / Chapter 1.1.2 --- Mangrove Mapping and Monitoring --- p.3 / Chapter 1.1.3 --- Role of Remote Sensing in Mangrove Study --- p.4 / Chapter 1.2 --- OBJECTIVES OF THE STUDY --- p.6 / Chapter 1.3 --- SIGNIFICANCE OF THE STUDY --- p.7 / Chapter 1.4 --- ORGANIZATION OF THE THESIS --- p.8 / Chapter CHAPTER 2 --- LITERATURE REVIEW --- p.10 / Chapter 2.1 --- INTRODUCTION --- p.10 / Chapter 2.2 --- FACTORS AFFECTING VEGETATION REFLECTANCE --- p.11 / Chapter 2.2.1 --- Foliar structure and principal constituents --- p.12 / Chapter 2.2.2 --- Foliar optical properties --- p.14 / Chapter 2.2.2.1 --- The visible region (400-700nm) --- p.14 / Chapter 2.2.2.2 --- The red edge (690-740nm) --- p.15 / Chapter 2.2.2.3 --- The near-infrared region (700-1300nm) --- p.16 / Chapter 2.2.2.4 --- The short-wave infrared region (1300-2500nm) --- p.17 / Chapter 2.2.3 --- Canopy architecture --- p.18 / Chapter 2.2.4 --- Background reflectance --- p.19 / Chapter 2.2.5 --- Atmospheric perturbation --- p.20 / Chapter 2.2.6 --- Sun-sensor relationship --- p.22 / Chapter 2.3 --- HYPERSPECTRAL IMAGING AND VEGETATION CLASSIFICATION --- p.23 / Chapter 2.4 --- RADAR IMAGING AND VEGETATION CLASSIFICATION --- p.31 / Chapter 2.5 --- PATTERN RECOGNITION FOR VEGETATION CLASSIFICATION --- p.39 / Chapter 2.5.1 --- The Hughes Phenomenon and Dimensionality Reduction --- p.39 / Chapter 2.5.2 --- Statistical Pattern Recognition and Feature Selection --- p.44 / Chapter 2.5.2.1 --- Search Method --- p.47 / Chapter 2.5.2.1.1 --- Exhaustive search --- p.48 / Chapter 2.5.2.1.2 --- Branch and bound --- p.49 / Chapter 2.5.2.1.3 --- Sequential forward/ backward selection --- p.55 / Chapter 2.5.2.1.4 --- Sequential Floating search --- p.57 / Chapter 2.5.2.1.5 --- Oscillating Search --- p.61 / Chapter 2.5.2.1.6 --- Genetic algorithm --- p.64 / Chapter 2.5.2.2 --- Evaluation criteria --- p.66 / Chapter 2.5.2.2.1 --- Distance measure --- p.67 / Chapter 2.5.2.2.2 --- Information measure --- p.68 / Chapter 2.5.2.2.3 --- Classification error --- p.71 / Chapter 2.5.2.3 --- Feature Selection Stability --- p.72 / Chapter 2.5.3 --- Feature extraction --- p.75 / Chapter 2.6 --- BIOPHYSICAL PARAMETERS MEASUREMENT AND ESTIMATION --- p.77 / Chapter 2.6.1 --- Leaf Area Index (LAI) --- p.78 / Chapter 2.6.2 --- Fraction of Absorbed Photosynthetically Active Radiation (fAPAR) --- p.79 / Chapter 2.6.3 --- In-situ Leaf Area Index Measurement --- p.81 / Chapter 2.6.3.1 --- Direct and Indirect Methods --- p.81 / Chapter 2.6.3.2 --- LAI Estimation through Gap Fraction Inversion --- p.85 / Chapter 2.6.3.3 --- Gap Fraction Ground Measurement --- p.89 / Chapter 2.6.3.3.1 --- LAI-2000 Plant Canopy Analyzer --- p.89 / Chapter 2.6.3.3.2 --- Hemispherical Photography --- p.92 / Chapter 2.6.3.4 --- Correction of Indirect LAI Measurement --- p.99 / Chapter 2.6.3.4.1 --- Clumping --- p.100 / Chapter 2.6.3.4.2 --- Mixture of Green and Non-green Elements --- p.101 / Chapter 2.6.4 --- Empirical Relationship with Spectral Vegetation Indices --- p.102 / Chapter 2.6.4.1 --- Traditional Vegetation Indices --- p.103 / Chapter 2.6.4.2 --- Leaf Area Index Estimation from Hyperspectral and Radar Images --- p.106 / Chapter 2.6.5 --- Physically-based Canopy Reflectance Model Inversion --- p.111 / Chapter 2.6.5.1 --- Canopy Reflectance Model --- p.111 / Chapter 2.6.5.2 --- Model Inversion and Biophysical Parameters Extraction --- p.115 / Chapter 2.7 --- SUMMARY --- p.118 / Chapter CHAPTER 3 --- METHODOLOGY --- p.120 / Chapter 3.1 --- INTRODUCTION --- p.120 / Chapter 3.2 --- STUDY AREA DESCRIPTION --- p.120 / Chapter 3.3 --- METHODOLOGICAL FLOW --- p.124 / Chapter 3.4 --- REMOTE SENSING DATA ACQUISITION AND PROCESSING --- p.127 / Chapter 3.4.1 --- Hyperion - EO-1 --- p.127 / Chapter 3.4.1.1 --- Radiometric correction --- p.127 / Chapter 3.4.1.1.1 --- Vertical strips removal --- p.128 / Chapter 3.4.1.1.2 --- Atmospheric correction --- p.129 / Chapter 3.4.1.1.3 --- Wavelength recalibration --- p.135 / Chapter 3.4.1.1.4 --- SNR enhancement through MNF --- p.137 / Chapter 3.4.1.2 --- Geometric correction --- p.139 / Chapter 3.4.1.3 --- Atmospheric correction algorithms comparison --- p.140 / Chapter 3.4.2 --- ASAR - ENVISAT --- p.141 / Chapter 3.4.2.1 --- Data Acquisition --- p.141 / Chapter 3.4.2.2 --- Data Processing --- p.143 / Chapter 3.4.2.2.1 --- Radiometric and Geometric Correction --- p.145 / Chapter 3.4.2.2.2 --- Speckle Filtering --- p.146 / Chapter 3.5 --- FIELD MEASUREMENTS AND DATA PROCESSING --- p.149 / Chapter 3.5.1 --- Species Distribution --- p.149 / Chapter 3.5.2 --- Leaf Spectra Measurement --- p.151 / Chapter 3.5.2.1 --- Leaf Collection and Handling --- p.152 / Chapter 3.5.2.2 --- ASD FieldSpec 3 Setup --- p.154 / Chapter 3.5.2.3 --- Laboratory setup --- p.156 / Chapter 3.5.2.4 --- Spectra Measurement --- p.158 / Chapter 3.5.2.5 --- Spectral similarity and variability --- p.159 / Chapter 3.5.3 --- In situ Leaf Area Index Measurement --- p.161 / Chapter 3.5.3.1 --- The optical instrument --- p.161 / Chapter 3.5.3.2 --- The LAI survey campaign p163 / Chapter 3.5.3.3 --- Data processing and canopy analysis --- p.166 / Chapter 3.5.3.4 --- Canopy parameter computation gap fraction, LAI, clumping index, mean inclination angle --- p.170 / Chapter 3.5.3.5 --- Field LAI and Their Correlation with Reflectance and Backscattering Coefficient Data Exploration --- p.175 / Chapter 3.6 --- FEATURE SELECTION --- p.175 / Chapter 3.6.1 --- Data Preprocessing and Preparation --- p.178 / Chapter 3.6.2 --- Data Format and Split --- p.183 / Chapter 3.6.3 --- Wrapper-based Approach --- p.185 / Chapter 3.6.4 --- Search Algorithm --- p.187 / Chapter 3.6.5 --- Stability Evaluation --- p.187 / Chapter 3.6.6 --- Feature Frequency analysis --- p.188 / Chapter 3.7 --- MANGROVE SPECIES CLASSIFICATION --- p.189 / Chapter 3.7.1 --- Species Separability --- p.193 / Chapter 3.7.2 --- Gaussian Maximum Likelihood Classifier --- p.193 / Chapter 3.7.3 --- Decision Tree Classifier --- p.194 / Chapter 3.7.4 --- Artificial Neural Network Classifier --- p.197 / Chapter 3.7.5 --- Support Vector Machines Classifier --- p.199 / Chapter 3.7.6 --- Accuracy Assessment --- p.204 / Chapter 3.8 --- LEAF AREA INDEX MODELING --- p.206 / Chapter 3.8.1 --- Preliminary Exploration of Relationship between Hyperspectral bands and LAI --- p.206 / Chapter 3.8.2 --- Vegetation Index Derived from Hyperspectral Data. --- p.206 / Chapter 3.8.3 --- Radar Backscatter and Derived Textural Parameters --- p.208 / Chapter 3.8.4 --- Regression Analysis --- p.211 / Chapter 3.8.5 --- Error Estimation --- p.217 / Chapter 3.9 --- SUMMARY --- p.218 / Chapter CHAPTER 4 --- RESULTS AND DISCUSSION (I) FEATURE SELECTION AND MANGROVE SPECIES CLASSIFICATION --- p.221 / Chapter 4.1 --- INTRODUCTION --- p.221 / Chapter 4.2 --- DATA PROCESSING AND EXPLORATION --- p.221 / Chapter 4.2.1 --- Atmospheric correction algorithms comparison --- p.222 / Chapter 4.2.2 --- Radar Data Speckle Reduction --- p.227 / Chapter 4.2.3 --- Statistical Discrimination of Mangrove Spectral Class --- p.230 / Chapter 4.3 --- FEATURE SELECTION --- p.249 / Chapter 4.3.1 --- Sequential Forward Selection (SFS) --- p.250 / Chapter 4.3.2 --- Sequential Floating Forward Selection (SFFS). --- p.256 / Chapter 4.3.3 --- Oscillating Search (OS) --- p.262 / Chapter 4.3.4 --- Search Algorithms comparison --- p.268 / Chapter 4.3.5 --- Final Subset Selection --- p.270 / Chapter 4.3.6 --- Correlation Analysis --- p.280 / Chapter 4.4 --- IMAGE CLASSIFICATION --- p.283 / Chapter 4.4.1 --- Mangrove Spectral Class Separability --- p.284 / Chapter 4.4.2 --- Gaussian Maximum Likelihood (ML) --- p.288 / Chapter 4.4.3 --- Decision Tree (DT) --- p.297 / Chapter 4.4.4 --- Artificial Neural Network (ANN) --- p.304 / Chapter 4.4.5 --- Support Vector Machines (SVM) --- p.312 / Chapter 4.4.6 --- Algorithm Comparison --- p.321 / Chapter 4.5 --- DISCUSSION AND IMPLICATION --- p.325 / Chapter 4.5.1 --- Feature Selection --- p.325 / Chapter 4.5.2 --- Mangrove Classification --- p.342 / Chapter 4.6 --- SUMMARY --- p.351 / Chapter CHAPTER 5 --- RESULTS AND DISCUSSION (II) - LEAF AREA INDEX MODELING --- p.353 / Chapter 5.1 --- INTRODUCTION --- p.353 / Chapter 5.2 --- DATA EXPLORATION --- p.353 / Chapter 5.2.1 --- Dependent Variable: Field measured LAI --- p.353 / Chapter 5.2.2 --- Independent Variables: Vegetation Index and texture measure --- p.355 / Chapter 5.2.3 --- Hyperspectral bands and LAI --- p.356 / Chapter 5.2.4 --- Normality testing --- p.359 / Chapter 5.2.5 --- Linearity testing --- p.363 / Chapter 5.2.6 --- Outliner detection --- p.365 / Chapter 5.3 --- SIMPLE LINEAR REGRESSION ANALYSIS --- p.366 / Chapter 5.3.1 --- LAI2000 Generalized method --- p.369 / Chapter 5.4 --- STEPWISE MULTIPLE REGRESSION ANALYSIS --- p.381 / Chapter 5.4.1 --- LAI2000 Generalized method --- p.384 / Chapter 5.5 --- DISCUSSION AND IMPLICATION --- p.391 / Chapter 5.5.1 --- LAI model comparison --- p.391 / Chapter 5.5.2 --- Species composition and LAI --- p.393 / Chapter 5.5.3 --- Hyperspectral Bands, Vegetation Indices and LAI --- p.397 / Chapter 5.5.4 --- Backscatter, texture measures and LAI --- p.407 / Chapter 5.5.5 --- Complementarity of Vegetation Index and Radar Parameters --- p.414 / Chapter 5.6 --- SUMMARY --- p.421 / Chapter CHAPTER 6 --- CONCLUSION --- p.423 / Chapter 6.1 --- SUMMARY OF THE STUDY --- p.423 / Chapter 6.2 --- LIMITATION OF THE STUDY --- p.427 / Chapter 6.3 --- RECOMMENDATION --- p.431 / Chapter REFERENCE --- p.434 / Chapter APPENDIX A --- GEOMETRIC CORRECTION OF HYPERSPECTRAL DATA --- p.473 / Chapter APPENDIX B --- SCRIPTS DERIVED FROM FEATURE SELECTION TOOLBOX (FST) FOR FEATURE SELECTION --- p.475 / Chapter APPENDIX C --- PREDICTED LAI(BON) AND LAI(2000) FROM SIMPLE LINEAR REGRESSION MODELS --- p.513 / Chapter APPENDIX D --- PREDICTED LAI(BON) AND LAI(2000) FROM MULTIPLE STEPWISE REGRESSION MODELS --- p.524