Thesis (MSc)--University of Stellenbosch, 2007. / Bibliography. Also available via the Internet.

Thesis (M.S.)--Oregon State University, 2008. / Printout. Includes bibliographical references. Also available on the World Wide Web.

Thesis (M.S.)--Oregon State University, 1993. / Typescript (photocopy). Includes bibliographical references. Also available on the World Wide Web.

Lidar is able to provide height and cover information which can be used to estimate selected forest attributes precisely. However, for users to evaluate whether the additional cost and complication associated with using Lidar merits adoption requires that the protocol to use lidar be thoroughly described and that a basis for selection of design parameters such as number of field plots and lidar pulse density be described. In our first analysis, we examine these issues by looking at the effects of pulse density and sample size on estimation when wall-to-wall lidar is used with a regression estimator. The effects were explored using resampling simulations. We examine both the effects on precision, and on the validity of inference. Pulse density had almost no effect on precision for the range examined, from 3 to .0625 pulses / m��. The effect of sample size on estimator precision was roughly in accordance with the behavior indicated by the variance estimator, except that for small samples the variance estimator had positive bias (the variance estimates were too small), compromising the validity of inference. In future analyses we plan to provide further context for wall-to-wall lidar-assisted estimation. While there is a lot of literature on modeling, there is limited information on how lidar-assisted approaches compare to existing methods, and what variables can or cannot be acquired, or may be acquired with reduced confidence. We expand our investigation of estimation in our second analysis by examining lidar obtained in a sampling mode in combination with Landsat. In this case we make inference about the feasibility of a lidar-assisted estimation strategy by contrasting its variance estimate with variance estimates from a variety of other sampling designs and estimators. Of key interest was how the precision of a two-stage estimator with lidar strips compared with a plot-only estimator from a simple random sampling design. We found that because the long and narrow lidar strips incorporate much of the landscape variability, if the number of lidar strips was increased from 7 to 15 strips, the precision of estimators with lidar can exceed that of estimators applied to plot-only SRS data for a much larger number of plots. Increasing the number of lidar strips is considered to be highly viable since the costs of field plots can be quite expensive in Alaska, often exceeding the cost of a lidar strip. A Landsat-assisted approach used for either an SRS or a two-stage sample was also found to perform well relative to estimators for plot-only SRS data. This proved beneficial when we combined lidar and Landsat-assisted regression estimators for two-stage designs using a composite estimator. The composite estimator yielded much better results than either estimator used alone. We did not assess the effects of changing the number of lidar strips in combination with using a composite estimator, but this is an important analysis we plan to perform in a future study. In our final analysis we leverage the synergy between lidar and Landsat to improve the explanatory power of auxiliary Landsat using a multilevel modeling strategy. We also incorporate a more sophisticated approach to processing Landsat which reflects temporal trends in individual pixels values. Our approach used lidar as an intermediary step to better match the spatial resolution of Landsat and increase the proportion of area overlapped between measurement units for the different sources of data. We developed two separate approaches for two different resolutions of data (30 m and 90 m) using multiple modeling alternatives including OLS and k nearest neighbors (KNN), and found that both resolution and the modeling approach affected estimates of residual variability, although there was no combination of model types which was a clear winner for all responses. The modeling strategies generally fared better for the 90 m approaches, and future analyses will examine a broader range of resolutions. Fortunately the approaches used are fairly flexible and there is nothing prohibiting a 1000 m implementation. In the future we also plan to look at using a more sophisticated Landsat time-series approach. The current approach essentially dampened the noise in the temporal trend for a pixel, but did not make use of information in the trend such as slope or indications of disturbance ��� which may provide additional explanatory power. In a future study we will also incorporate a multilevel modeling into estimation or mapping strategies and evaluate the contribution of the multilevel modeling strategy relative to alternate approaches. / Graduation date: 2013 / Access restricted to the OSU Community at author's request from June 21, 2012 - Dec. 21, 2012

Thesis (MSc (Forest and Wood Science))—University of Stellenbosch, 2007. / The present study investigated double sampling with regression estimators as a quest for efficiency and effectiveness in forest inventory in Namibian woodlands. Auxiliary data used were obtained from Standard QuickBird satellite scenes (phase 1) for Hans Kanyinga Community Forest from October and November 2004 covering an area of 12,107 hectares, amplified with terrestric data (phase 2) of 2002. The relationships between auxiliary and terrestric variables are described and prediction models were constructed. According to the results of the stepwise procedure with the Mallow’s Cp statistic as the selection criteria, photogrammetric stand density and a combination of the photogrammetric crown area with photogrammetric stand density were the best candidates for predicting the stand volume. The resulting volume model explains 56% of the variation. Photogrammetric stand density was found to be highly correlated to the terrestric stand density with the resulting model explaining 81% of the variation. Photogrammetric crown diameter was found to be correlated with the diameter at breast height measured from the plots which were assessed for spatial tree positions, which enabled the derivation of the diameter distribution. The diameter distribution model explains 43% of the variation. In addition, the actual tree positions were determined using the GPS and surveying techniques (polar positions) involving distance and bearings. GPS tree positions showed a considerable shift of up to 8.67 m. However, only the distance measurements of tress from the plot centre using the infield surveying methods were more reliable. Nevertheless, the influences of the tree positional errors are not of high concern for temporary based sample plots which are normally used in Namibian forest inventories. A reduction in inventory cost was found to be 24% i.e. N$25.79 to N$19.67 per hectare. The results of this study are valid for Kavango region or any other region with similar set of physical and climatic conditions, but caution must be exercised in implementing these results elsewhere under different physical and environmental conditions.

Thesis (MSc)--Stellenbosch University, 2000. / ENGLISH ABSTRACT: The aim of the study was to extract elevation information (such as tree height) from stereo satellite imagery (IRS-I C), to scrutinise the performance of the DTM (Digital Terrain Model) tools as provided by the LH (LeicalHelava) Systems' softcopy system, and subsequently to perform a feasibility study on the application of a practically viable forest inventory design. A softcopy photogrammetry workstation (LH Systems DPW 770), IRS-I C stereo panchromatic satellite imagery, and digital aerial photography at a scale of 1:30000 (scanned at 15 micrometers) was used. The study was conducted over various sites in the Sabie area (province of Mpumalanga) in South Africa, where extensive man made forests with pine and eucalypts are to be found. The extraction of stand parameters such as tree height was performed manually, semi-automatically, and automatically. In addition, the compartment area was determined using a GIS tool. The Digital Surface Models (DSM), representing the canopy structure of the stands, was extracted from the IRS-I C imagery and validated through a comparison of the resulting contours with the corresponding contours generated by aerial photogrammetric methods. Due to the coarse spatial resolution of the IRS-IC imagery (5m) and the suboptimal BIH (BaselHeight) ratio (0.57), only objects featuring a height exceeding 20m could be manually measured with confidence. Furthermore, only the edges of the compartments proved to be suitable for the determination of tree heights (i.e. with a sufficiently large parallax difference and image contrast). The manual determination of tree heights in the IRS-I C imagery yielded accuracies of about 95% compared to the height values of the aerial photographs and the ground data. The application of image enhancement techniques had severe effects on the accuracy of the IRS-IC stereo model, resulting in deviations of about -57m from the 'true' value. It was observed that image matching was only a problem where features changed their appearance (e.g. clearfelled or burnt areas) during the acquisition period of the stereo pair of the satellite imagery. LH Systems' Adaptive Automatic Terrain Extraction (AATE) tool performed very well for the creation of digital terrain and surface models when using digital aerial photography with a high scanning rate. In contrast, the automatic creation of canopy surface models from various forest compartments did not yield any useful results when applied to IRS-l C imagery. AATE could not model the canopy structure properly. The coarse spatial resolution of the satellite imagery in conjunction with the sparse post spacing (20m) and matching errors are most likely to be responsible for this poor performance. Two-phase sampling and the Hugershoff method were chosen for automatically derived height values to be evaluated for possible application in forest inventory. Unfortunately, neither for the determination of the regression estimator for the first method, nor for the calculation of timber volume after application of the Hugershoff method could any useful result be obtained. This is mostly due to the fact that image matching errors and blunders (resulting in tree heights of -885m) were not properly accounted for in the terrain extraction software. However, the outcomes for the manual measurement of tree heights performed on the satellite imagery show that under optimal conditions accuracies can be achieved similar to those for the height determination in small scale aerial photographs, but at lower cost. The obtained height values can then be used for the calculation of timber volume according to Eichhorn's law. Keywords: AATE, blunders, digital photogrammetry, DPW770, forest inventory, Hugershoff IRS-l C, matching error, remote sensing, satellite imagery, two-phase sampling / AFRIKAANSE OPSOMMING: N GANGBAARHEIDSTIIDIE VIR BOSINVENTARIS MET BEHULP VAN STEREO SATELLIETBEELDE MET GEBRUIK VAN SAGTEKOPIE FOTOGRAMMETRIESETEGNOLOGIE: Die doel van hierdie studie was om elevasie inligting (soos boomhoogtes) uit stereo satellietbeelde (IRS-I C) te ontrek, en die DTM (Digitale Terrein Modelle) funksies van die LH Systems se sagtekopie sisteem te evalueer en 'n ondersoek in te stel na praktiese toepassing van die tegnologie in bosvoorraadopname. 'n Sagtekopie fotogrammetriese werkstasie (LH Systems DPW 770), IRS-I C stereo panchromatiese satellietwaarneming en digitale lugfotografie is gebruik. Die studie is uitgevoer oor verskeie areas in die Sabie omgewing (Mpumalanga, Suid-Afrika), waar daar ekstensiewe mensgemaakte woude voorkom met denne en Eucalyptus soorte. Die ekstraksie van opstandparameters soos boomhoogte is uitgevoer met die hand, as ook met semi-outomatiese en outomatiese metodes. Die digitale oppervlakmodelle (wat die kroondakstrukture van die opstande voorstel) was vanaf die IRS-I C beelde onttrek en gevalideer deur vergelyking van die resulterende kontoere met die korresponderende kontoere wat deur lugfotogrammetriese metodes gegenereer is. As gevolg van die growwe ruimtelike resolusie van die IRS-IC waarneming (Sm) en die suboptimale BIH verhouding (0.57) kan slegs voorwerpe met 'n hoogte van meer as 20m met vertroue met die hand gemeet word. Slegs die rande van die vakke is bruikbaar vir die berekening van boomhoogtes (d.w. s. met 'n voldoende paralaksverskil en 'n sterk beeldkontras ). Boomhoogtes wat met die hand bepaal is vanaf IRS-I C beelde is 95% akkuraat in vergelyking met die hoogtewaardes verkry vanaf die lugfoto's en die veldmetings. Die toepassing van beeldverbeteringstegnieke het duidelike invloede op die akkuraatheid van die IRS-IC stereomodel met afwykings van ongeveer -57m vanaf die "werklike" waardes. Daar is ook waargeneem dat beeldooreenstemming slegs 'n probleem is waar terreinvorme se voorkoms verander het (weens afkappings of brande) gedurende die verkrygingsperiode waarin die stereo paar van die satellietbeelde verkry is. LH Systems se Aanpassende Outomatiese Terrein Onttrekkings (Adaptive Automatic Terrain Extraction - AATE) instrument het goed gevaar tydens die gebruik van digitale lugfotografie met Inhoë skanderingstempo. In kontras hiermee het die outomatiese skepping van kroondakoppervlakmodelle van verskeie plantasievakke geen nuttige resultate gelewer wanneer dit op IRS-I C beelde toegepas is nie. Die growwe ruimtelike resolusie van die satellietbeelde tesame met die wye paalspasïering (20m) en passingsfoute is waarskynlik vir hierdie swak resultate verantwoordelik. Twee-fase proefueming en die Hugershoff metode was gebruik vir die bepaling van outomaties afgeleide hoogtewaardes vir evaluering van moonlike toepassing in bosvoorraadopnames. Geen bruikbare resultate kon verkry word vir die vasstelling van die regressieskatter vir die eersgenoemde metode of vir die berekening van die houtvolume volgens die Hugershoff metode nie. Dit is meestal as gevolg van beeld-- ooreenkomsfoute en flaters, (wat tot boomhoogtes van -885m gelei het) wat nie voldoende in ag geneem word in die terreinekstraksie sagteware nie. Die resultate vir die handgemete ('manual') boomhoogtebepaling wat uitgevoer is op die satellietbeelde (op die sagtekopie werkstasie DPW 770), toon dat akkuraathede soortgelyk aan daardie vir hoogte bepaal op klein-skaal lugfotos onder optimale toestande verkry kan word, maar goedkoper. Die hoogtewaardes wat verkry is kan gebruik word vir die berekening van houtvolume volgens die wet van Eichhorn. Sleutelwoorde: AATE, afstandswaarneming, bosvoorraadopnames, digitale fotogrammetrie, DPW770, flaters, Hugershoff, IRS-! C, satellietbeelde, twee-fase proefueming

Improved monitoring of forest biomass is needed to quantify natural and anthropogenic effects on the terrestrial carbon cycle. Landsat's temporal and spatial coverage, fine spatial grain, and long history of earth observations provide a unique opportunity for measuring biophysical properties of vegetation across large areas and long time scales. However, like other multi-spectral data, the relationship between single-date reflectance and forest biomass weakens under certain canopy conditions. Because the structure and composition of a forest stand at any point in time is linked to the stand's disturbance history, one potential means of enhancing Landsat's spectral relationships with biomass is by including information on vegetation trends prior to the date for which estimates are desired. The purpose of this research was to develop and assess a method that links field data, airborne lidar, and Landsat-derived disturbance and recovery history for mapping of forest biomass and biomass change. Our study area is located in eastern Oregon (US), an area dominated by mixed conifer and single species forests. In Chapter 2, we test and demonstrate the utility of Landsat-derived disturbance and recovery metrics to predict current forest structure (live and dead biomass, basal area, and stand height) for 51 field plots, and compare the results with estimates from airborne lidar and single-date Landsat imagery. To characterize the complex nature of long-term (insect, growth) and short-term (fire, harvest) vegetation changes found in this area, we use annual Landsat time series between 1972 and 2010. This required integrating Landsat data from MSS (1972-1992) and TM/ETM+ (1982-present) sensors. In Chapter 2, we describe a method to bridge spectral differences between Landsat sensors, and therefore extent Landsat time-series analyses back to 1972. In Chapter 3, we extend and automate our approach and develop maps of current (2009) and historic (1993-2009) live forest biomass. We use lidar data for model training and evaluate the results with forest inventory data. We further conduct a sensitivity analysis to determine the effects of forest structure, time-series length, terrain and sampling design on model predictions. Our research showed that including disturbance and recovery trends in empirical models significantly improved predictions of forest biomass, and that the approach can be applied across a larger landscape and across time for estimating biomass change. / Graduation date: 2012 / Access restricted to the OSU Community at author's request from Nov. 29, 2011 - Nov. 29, 2012