A busca por elevados índices de produtividade agrícola, bem como a demanda constante pelo uso da terra tem cada vez mais realçado a importância do conhecimento do solo e de suas propriedades, o que se dá, principalmente, através dos levantamentos de solo. No entanto, a obtenção destas informações não é direta, sendo dispendiosa, morosa e pouco atrativa, refletindose na carência de profissionais especializados e, conseqüentemente, na ausência de levantamentos de solos em níveis detalhados. A continuidade na execução de levantamentos, em quaisquer níveis, depende principalmente do emprego de novas técnicas, sobretudo das técnicas de mapeamento digital. Essencialmente por este recurso oferecer agilidade e acuracidade, despertando também o interesse de novos pesquisadores. Inúmeros estudos em levantamentos de solos têm abordado e indicado o uso de dados geoespaciais (modelo digital de elevação, geomorfologia, entre outros) e produtos de sensoriamento remoto espectral (espectrorradiometria aos níveis de laboratório, campo e/ou orbital), porém não há trabalhos unindo estas técnicas de forma concisa, apurada e científica. Neste contexto, este trabalho visa, fundamentalmente, determinar um método que identifique e espacialize classes de solos utilizando múltiplas ferramentas, como o sensoriamento remoto (dados espectrais ao nível laboratorial e orbital), aspectos da paisagem (modelo digital de elevação, declividade, curvatura, rede de drenagem, e outros) e sistemas de informações geográficas (manipulação e cruzamento de informações), obtendo, como produto final, um mapa digital de solos. / The search for high levels of agricultural productivity and the constant demand for the use of land has emphasized the importance of knowledge of soil and its properties, mainly through of the soil mapping. However obtaining this information is not direct, beyond costly, slowly and not very attractive, reflecting on the lack of skilled professionals and in the absence of mappings at levels detailed. The continuity in the implementation of surveys on any level depends mainly on the use of new techniques, especially the digital techniques. Essentially this feature by offering speed and accuracy, also arousing the interest of new researchers. Studies on surveys of land have addressed and indicated the use of geospatial data (digital elevation model, geomorphology, among others) and products of spectral remote sensing (spectroradiometry at the laboratory, field and/or orbital levels), but there is no works uniting these techniques in a concise, accurate and scientific away. In this context, this work is aimed, primarily, determine a method that identifies and spatializing classes of soil using multiple tools, such as remote sensing (spectral data at laboratory and orbital), aspects of the landscape (digital model of elevation, slope, curvature, network of drainage, and others) and geographic information systems (crossing and manipulation of information), obtaining, as the final product, a digital soil map.

Os solos são vitais para todos os ecossistemas terrestres e deles depende a maior parte dos recursos para a manutenção dos seres vivos. Seu uso é importante para a agricultura e, para assim serem usados, é necessário conhecê-los, tanto como são como onde estão na paisagem. Esse conhecimento pode ser adquirido através de levantamentos de solos para os quais existem muitas limitações, como alta demanda financeira, elevado tempo para sua execução e subjetividade associada ao conhecimento tácito dos pedólogos. Por isso, são necessárias novas estratégias que auxiliem a execução de mapas de solos. Uma abordagem promissora é a identificação de unidades naturais do relevo em nível de detalhe, uma vez que é possível predizer a ocorrência de atributos e tipos de solos na paisagem quando associando as feições dos seus perfis com as de sua superfície. Diante disto, este trabalho objetiva testar procedimentos digitais para segmentação detalhada de elementos das encostas e relacioná-los com os atributos e classes taxonômicas de solos. Em uma área de estudo de 2.500 ha, situada na região de Piracicaba (SP), parâmetros geomorfométricos organizados hierarquicamente em regras em uma árvore de decisão foram utilizados para classificar, em escala detalhada (1:10.000), os cinco elementos da encosta (topo, ombro, meia-encosta, sopé coluvial e sopé colúvio-aluvial). Avaliou-se uma estratégia de análise de similaridade visando à identificação de agrupamentos de amostras de solos da mesma classe, a partir de diferentes conjuntos de variáveis. Essa segmentação digital mostrou que é possível explicitar a localização de cada um dos elementos da encosta e que neles dominam perfis de solos que se assemelham. Na maioria dos casos, essa semelhança pode ser comprovada com o uso tanto de análises convencionais como espectrais das amostras de solo coletadas até 1 m de profundidade. Essa classificação digital dos elementos da encosta pode auxiliar no mapeamento de solos detalhados e ultradetalhados (escalas 20.000 ou maiores). / Soils are vital for all terrestrial ecosystems and the majority of resources for maintenance of human beings depend on them. Their use is important for agriculture and, in order to be used in this manner, it is essential to know them, as well as how they are and where they are located in the landscape. This knowledge can be acquired through soil surveys, that have several limitations, such as high financial demand, time-consuming execution and subjectivity associated with the pedologists tacit knowledge. Considering this, new strategies are needed to support the elaboration of soil maps. One promising approach is the identification of detailed natural units of relief, since it is possible to predict the occurrence of attributes and types of soils in the landscape when associating the features of their profiles with those of their surface. Therefore, this research aims to test digital procedures for detailed segmentation of hillslope elements and to relate them to soil attributes and taxonomic classes. In a study area of 2,500 ha located in the Piracicaba (SP) region, geomorphometric parameters hierarchically organized into rules in a decision tree were used in order to classify, in a detailed scale (1:10,000), five hillslope elements (summit, shoulder, backslope, footslope and toeslope). A similarity analysis strategy was used to identify groupings of soil samples from the same class, from different sets of variables. This digital segmentation showed that it is possible to make explicit the location of each one of the hillslope elements, where similar soil profiles are dominant. In most cases, this similarity can be verified with the use of both conventional and spectral analyses of soil samples collected up to a depth of 1 m. This digital classification of hillslope elements can support 1st and 2nd order soil survey (scales 1:36,680 or greater).

As áreas agrícolas vêm se tornando cada vez mais tecnificadas, onde o conhecimento das características físicas, químicas e mineralógicas dos solos se torna imprescindível para maximizar a produtividade. O Brasil possui uma grande extensão territorial, sendo que a maior parte não possui mapas de solos compatíveis com as necessidades agrícolas. Torna-se necessário fornecer subsídios à pesquisa pedológica referente ao aperfeiçoamento de técnicas que venham auxiliar os levantamentos de solos, tornando-os mais ágeis e econômicos. Para tanto, foram traçados os seguintes objetivos: caracterizar o comportamento espectral de solos nos níveis de campo, laboratório e orbital; correlacionar as alterações dos solos ao longo de topossequências com o caráter espectral; verificar a separabilidade das unidades de mapeamento e quantificar atributos dos solos através das respostas radiométricas; avaliar a eficiência prática da técnica. O trabalho foi conduzido em Barra Bonita, SP, onde predominam diferentes unidades de solos tais como LATOSSOLOS, ARGISSOLOS, CAMBISSOLOS e NITOSSOLOS com texturas de arenosas a muito argilosas. Na área foram demarcados pontos para amostragem em forma de grade (100 x 100 m). Todos os pontos de coleta foram georreferenciados, foram realizadas tradagens nas profundidades 0-20 e 80-100 cm. As amostras de terra foram encaminhadas ao laboratório para análises físicas, químicas. Foram obtidos dados espectrais através de espectroradiômetro em laboratório e campo. Foi realizado um mapa de solos detalhado pelo método convencional, incluindo a caracterização de perfis. Através dos dados espectrais obtidos nos níveis orbital e laboratório foram geradas equações discriminantes para os solos e equações lineares de regressão múltipla para vários atributos do solo. Os atributos dos solos foram comparados com valores estimados pelas equações e os valores determinados nas análises de laboratório para verificar a veracidade dos dados espectrais e a variabilidade da metodologia proposta. Na medida em que ocorrem alterações dos solos ao longo de uma topossequência, o comportamento espectral detectado pelos sensores, se altera. A análise descritiva das curvas espectrais descritas em literatura fornece poucos detalhes na discriminação de solos. É possível discriminar solos por sensor terrestre e orbital com 81 e 40 % de acerto respectivamente. A estimativa de teores de ferro por sensores terrestre e orbital auxilia na classificação de solos. É possível quantificar atributos do solo como areia, argila e ferro por sensores em laboratório, e, com menor índice de significância por sensores orbitais. / The agricultural areas are becoming more and more technified, where the knowledge of the physical, chemical and mineralogical characteristics of soils becomes indispensable to maximize crop productivity. Brazil possesses a great territorial extension, and most of the areas do not possess soil maps compatible with the agricultural needs. Modern technological improvements signal the necessity of supplying data to soil research, especially those areas regarding soil survey, with the objective of turning soil survey more agile and economic. The following objectives were traced: characterize the spectral behavior of soils in field, laboratory and orbital levels; correlate the alterations of soils along toposequences with the spectral character; verify the discrimination of mapping units and quantify attributes of the soils through their spectral responses; evaluate the practical efficiency of the technique. The work was done in Barra Bonita, SP, where Oxisol, Alfisol, Inceptisol e Ultisol with textures of sandy to very loamy dominate. In the study area, samples were taken following a 100 x 100 m grid. All the points were georeferenced and soil samples were obtained in the depths 0-20 and 80-100 cm. The soil samples were taken to the laboratory for physical and chemical analyses. After that, the spectral data were obtained using a spectroradiometer in laboratory, field and orbital levels. A detailed soil maps was done by the conventional method, including the characterization of profiles. Through the spectral data obtained in the orbital and laboratory levels, discriminant equations were generated for the soils and linear equations of multiple regression for several attributes of the soil. The soil attributes were compared with values quantified by the equations and values determined in laboratory analyses to verify the accuracy of the spectral data and the variability of the methodology proposal. As the soils modifications occur along a toposequence, the spectral behavior detected by the sensors becomes different. The descriptive analysis of the spectral curves described in literature supplies few details of soil discrimination. It is possible to discriminate soils by terrestrial and orbital sensors with 81 and 40% of success, respectively. The estimation of the iron content by terrestrial and orbital sensors aids in the classification of soils. It is possible to quantify attributes of the soil as sand, clay and iron using a sensor in the laboratory, and, with smaller accuracy by orbital sensors.

Submitted by Jorge Silva (jorgelmsilva@ufrrj.br) on 2016-10-21T17:28:48Z No. of bitstreams: 1 2016 - Alessandro Samuel-Rosa.pdf: 15092171 bytes, checksum: bbe06c922805d4196e0a50c4f2aee7a5 (MD5) / Made available in DSpace on 2016-10-21T17:28:48Z (GMT). No. of bitstreams: 1 2016 - Alessandro Samuel-Rosa.pdf: 15092171 bytes, checksum: bbe06c922805d4196e0a50c4f2aee7a5 (MD5) Previous issue date: 2016-02-24 / CNPq / Modern soil spatial modelling is based on statistical models to explore the empirical relation-ship among environmental conditions and soil properties. These models are a simplification of reality, and their outcome (soil map) will always be in error. What a soil map conveys is what we expect the soil to be, acknowledging that we are uncertain about it. The objective of this thesis is to evaluate important sources of uncertainty in spatial soil modelling, with emphasis on soil and covariate data. Case studies were developed using data from a catchment located in Southern Brazil. The soil spatial distribution in the study area is highly variable, being deter-mined by the geology and geomorphology (coarse spatial scales), and by agricultural practices (fine spatial scales). Four topsoil properties were explored: clay content, organic carbon con-tent, effective cation exchange capacity and bulk density. Five covariates, each with two levels of spatial detail, were used: area-class soil maps, digital elevation models, geologic maps, land use maps, and satellite images. These soil and covariate data constitute the Santa Maria dataset. Two packages for R were created in support to the case studies, the first (pedometrics) con-taining various functions for spatial exploratory data analysis and model calibration, the second (spsann) designed for the optimization of spatial samples using simulated annealing. The case studies illustrated that existing covariates are suitable for calibrating soil spatial models, and that using more detailed covariates results in only a modest increase in the prediction ac-curacy that may not outweigh the extra costs. More efficient means of increasing prediction accuracy should be explored, such as obtaining more soil observations. For this end, one should use objective means for selecting observation locations to minimize the effects of psycholog-ical responses of soil modellers to conceptual and operational factors on the sampling design. This because conceptual and operational difficulties encountered in the field determine how the motivation of soil modellers shifts between learning/verifying soil-landscape relationships and maximizing the number of observations and geographic coverage. For the sole purpose of spa-tial trend estimation, it should suffice to optimize spatial samples aiming only at reproducing the marginal distribution of the covariates. For the joint purpose of optimizing sample configu-rations for spatial trend and variogram estimation, and spatial interpolation, one can formulate a sound multi-objective optimization problem using robust versions of existing sampling algo-rithms. Overall, we have learned that a single, universal recipe for reducing our uncertainty in soil spatial modelling cannot be formulated. Deciding upon efficient ways of reducing our uncertainty requires, first, that we explore the full potential of existing soil and covariate data using sound spatial modelling techniques. / A modelagem espacial do solo moderna usa modelos estat?sticos para explorar a rela??o em-p?rica entre as condi??es ambientais e as propriedades do solo. Esses modelos s?o uma sim-plifica??o da realidade, e seu resultado (mapa do solo) estar? sempre errado. O que um mapa do solo transmite ? o que esperamos que o solo seja, reconhecendo que somos incertos sobre ele. O objetivo dessa tese ? avaliar importantes fontes de incerteza na modelagem espacial do solo, com ?nfase nos dados do solo e covari?veis. Estudos de caso foram desenvolvidos usando dados de uma bacia hidrogr?fica do sul do Brasil. A distribui??o espacial do solo na ?rea de estudo ? vari?vel, sendo determinada pela geologia e geomorfologia (escalas espaciais maiores) e pr?ticas agr?colas (escalas espaciais menores). Quatro propriedades do solo foram explora-das: teor de argila, teor de carbono org?nico, capacidade de troca cati?nica efetiva e densidade. Cinco covari?veis, cada um com dois n?veis de detalhe espacial, foram utilizadas: mapas areais de classes de solo, modelos digitais de eleva??o, mapas geol?gicos, mapas de uso da terra, e imagens de sat?lite. Esses dados constituem o conjunto de dados de Santa Maria. Dois paco-tes para R foram criados, o primeiro (pedometrics) contendo v?rias fun??es para a an?lise explorat?ria espacial de dados e calibra??o de modelos, o segundo (spann) projetado para a optimiza??o de amostras espaciais usando recozimento simulado. Os estudos de caso ilustraram que as covari?veis existentes s?o apropriadas para calibrar modelos espaciais do solo, e que o uso de covari?veis mais detalhadas resulta em modesto aumento na acur?cia de predi??o que pode n?o compensar os custos adicionais. Meios mais eficientes de aumentar a acur?cia de pre-di??o devem ser explorados, como obter mais observa??es do solo. Para esse fim, deve-se usar meios objetivos para a sele??o dos locais de observa??o a fim de minimizar os efeitos das res-postas psicol?gicas dos modeladores do solo a fatores conceituais e operacionais sobre o plano de amostragem. Isso porque as dificuldades conceituais e operacionais encontradas no campo determinam mudan?as na motiva??o dos modeladores do solo entre aprendizagem/verifica??o das rela??es solo-paisagem e maximiza??o do n?mero de observa??es e cobertura geogr?fica. Para estimar a tend?ncia espacial, deve ser suficiente otimizar as amostras espaciais visando so-mente reproduzir a distribui??o marginal das covari?veis. Para otimizar configura??es amostrais para estimar a tend?ncia espacial e o variograma, e interpola??o espacial, pode-se formular um problema de otimiza??o multi-objetivo s?lido usando vers?es robustas de algoritmos de amos-tragem existentes. No geral, aprendemos que uma receita ?nica, universal para a redu??o da incerteza na modelagem espacial do solo n?o pode ser formulada. Decidir sobre formas efi-cazes de redu??o da incerteza requer, em primeiro lugar, que exploremos todo o potencial dos dados existentes usando t?cnicas de modelagem espacial s?lidas.

As fotografias aéreas, bem como as técnicas de estereoscopia, foram amplamente utilizadas para estudos ambientais e da paisagem. Com o avanço do mapeamento digital de solos os parâmetros da rede de drenagem foram sendo substituídos por parâmetros derivados do relevo, nas metodologias de predição de atributos do solo. No entanto, a literatura é ampla na atribuição da rede de drenagem como um fator determinante no mapeamento de solos, havendo a necessidade de desenvolver técnicas para inserir as características dos canais nos diferentes métodos de mapeamento do solo. Dessa forma, objetiva-se desenvolver um mapa de compartimentação da paisagem, através da rede de drenagem e um modelo digital de elevação (MDE), ambos com alta resolução espacial, visando avaliar as suas correlações com os atributos do solo (teor de argila nas profundidades 0-20 e 80-100 cm, gradiente textural, Ferro total (Fe2O3) e a cor do solo) e classes pedológicas. Tais procedimentos poderão auxiliar na produção de métodos base para relacionar a paisagem com a pedologia e o mapeamento. A área de estudo está localizada no município de Rio das Pedras, no estado de São Paulo, Brasil, com 538 km². O levantamento da rede de drenagem foi realizado a partir de fotografias aéreas com a visualização em 3D por estereoscopia digital. O MDE foi criado a partir de curvas altimétricas com equidistância vertical de 5 m. A partir das características da rede de drenagem e do relevo foram calculados os parâmetros morfométricos que representassem os dois elementos ao longo de toda a área de estudo. Com o processamento dos parâmetros foi utilizada a técnica fuzzy k-médias para fazer uma compartimentação da paisagem não supervisionada. Os resultados mostraram que a densidade de drenagem (DD) possui uma correlação negativa com os teores de argila (r = - 0.63), enquanto a correlação com o gradiente textural foi positiva (r = 0.42). O ferro total (Fe2O3) apresentou baixa variabilidade na área e não resultou em correlações significativas. A maior correlação foi com o matiz (r = 0.67), determinando solos mais amarelos nos locais de maior DD. A compartimentação da paisagem separou bem as posições do relevo em relação a DD. Cada compartimento se apresentou como uma unidade de mapeamento, havendo relação direta com classes pedológicas. / Aerial photographs, as well as stereoscopy were widely used for environmental and landscape studies. As digital soil mapping techniques have been developed, drainage network was replaced for relief parameters. However, literature studies have shown vast attribuition between drainage network and soil mapping, bringing the need to develop ways to insert the drainage parameters on different soil mapping strategies. Therefore, this study aims to create a landscape compartment map through drainage network and a digital elevation model (DEM), both with high spatial resolution, in order to evaluate its correlation with five soil attributes (clay content at 0-20 and 80-100 cm, textural gradient, total Iron (Fe2O3) and color) and soil classes. Wih these procedures it will be possible to create base methods to associate landscape with pedology and mapping. The study area has 538 km² and is located at Rio das Pedras municipality in the state of São Paulo, Brazil. The drainage network was created using aerial photographs with digital stereoscopy in 3D and the DEM with altimetric curves. These two geographic basis were used to calculate morphometric parameters that represents the patterns along the study area. The parameters were processed with fuzzy k-means technique to create a non-supervised landscape compartments map. Results showed that drainage density (DD) had a negative correlation with clay content (r = - 0.63), while textural gradient was positive (r = 0.42). The (Fe2O3) had low spatial variability, resulting in non-significant results. The greatest correlation was achieved with soil color (r = 0.67), indicating yellow soils at high DD landscapes. The landscape compartment was able to distinguish the relief positions related to DD. Each compartment was assumed as a soil map unit, presenting straight connections with soil classes.

Graduation date: 1992

Digital soil mapping (DSM) relies on quantitative relationships between easily measured environmental covariates and field and laboratory data. We applied innovative sampling and inference techniques to predict the distribution of soil attributes, taxonomic classes, and dominant vegetation across a 30,000-ha complex Great Basin landscape in southwestern Utah. This arid rangeland was characterized by rugged topography, diverse vegetation, and intricate geology. Environmental covariates calculated from digital elevation models (DEM) and spectral satellite data were used to represent factors controlling soil development and distribution. We investigated optimal sample size and sampled the environmental covariates using conditioned Latin Hypercube Sampling (cLHS). We demonstrated that cLHS, a type of stratified random sampling, closely approximated the full range of variability of environmental covariates in feature and geographic space with small sample sizes. Site and soil data were collected at 300 locations identified by cLHS. Random forests was used to generate spatial predictions and associated probabilities of site and soil characteristics. Balanced random forests and balanced and weighted random forests were investigated for their use in producing an overall soil map. Overall and class errors (referred to as out-of-bag [OOB] error) were within acceptable levels. Quantitative covariate importance was useful in determining what factors were important for soil distribution. Random forest spatial predictions were evaluated based on the conceptual framework developed during field sampling.

The relationships between the spectral properties of Arizona soils and rangelands and their characteristics were studied. The per cent reflectance of soils was determined using a multispectral hand-held radiometer, and the spectral response of Arizona rangeland sites was measured by scanners aboard an orbiting satellite. These spectral properties were related, by means of stepwise multiple regressions, to various soil and site characteristics. This research is presented in three chapters. The first chapter describes the relationships between soil properties and their spectral reflectance as determined in a laboratory environment. The second chapter attempts to correlate spectral properties of soils measured with a radiometer and that measured by scanners aboard an orbiting satellite for a small area near Winkelman, Arizona. The third chapter describes the relationships between the properties of 243 rangeland sites in central and southeastern Arizona and Landsat spectral data values. Determinations of Munsell soil colors and the radiometrically measured reflectance of 163 soils led to the development of charts for converting Munsell color to reflectance. Little difference was found between Munsell color measured in the sun and that measured indoors, and on the average, soil scientists were in agreement 80 per cent of the time. Munsell value, organic carbon, carbonates, and Munsell chroma explained 80 per cent of the variability within the reflectance measurements of these soils. The spectral response of the less than 2 mm soil fraction collected from rangeland surfaces was significantly different from the spectral response of coarser fragments collected from the same surface. In the Winkelman area the radiometrically measured reflectance of the less than 2 mm fraction alone accounted for 46 per cent of the variability and the reflectance of the 13 to 76 mm fraction accounted for 17 per cent of the variability within the satellite measured response. This area had a low vegetative cover and soil-geologic features, particularly soil color, correlated best with the Landsat digital data. Seventy-six per cent of the satellite data were explained by the interaction of the per cent coarse fragments times its reflectance, the average slope of the sites and the per cent soil less than 2 mm fraction times its reflectance. The relationship between the properties of 110 rangeland sites in central Arizona and the sum of the four Landsat spectral bands was determined. The sum of brush and forest crown densities, elevation, soil color,Geology of the site, and the per cent of surface covered with cobbles explained 82 per cent of this variation. An evaluation of field measurements only to explain the variability among mapping units showed the sum of brush and forest crown densities, elevation, clay content, and fragments greater than 2 mm explained 67 per cent of this variation. When satellite data were added to the field measurement site characteristics, the ratio of satellite scanner bands 4+5 to 6+7 becomes the most significant factor in explaining the variation among mapping unit symbols and a greater per cent of the variability could be explained. A similar study conducted on 133 sites in southeastern Arizona gave different results as only 41 per cent of the variability could be explained. It was shown that for central and southern Arizona rangelands, it is possible to define specific relationships between site characteristics and satellite measured spectral response. Less than ten site characteristics and their interactions explain considerable portions of the variability between mapping units for a given survey. These relationships are unique for specific locations, but they could easily be developed for a survey area and effectively used in the mapping process.

Soil is extremely important for providing food, biomass and raw materials, water and nutrient storage; supporting biodiversity and providing foundations for man-made structures. However, its health is threatened by human activities, which can greatly affect the potential of soils to fulfil their functions and, consequently, result in environmental, economic and social damage. These issues require the characterisation of the impact and spatial extent of the problems. This can be achieved through the creation of detailed and comprehensive soil maps that describe both the spatial and vertical variability of key soil properties. Detailed three-dimensional (3D) digital soil maps can be readily used and embedded into environmental models. Three-dimensional soil mapping is not a new concept. However, only with the recent development of more powerful computers has it become feasible to undertake such data processing. Common techniques to estimate soil properties in the three-dimensional space include geostatistical interpolation, or a combination of depth functions and geostatistics. However, these two methods are both partially flawed. Geostatistical interpolation and kriging in particular, estimate soil properties in unsampled locations using a weighted average of the nearby observations. In order to produce the best possible estimate, this form of interpolation minimises the variance of each weighted average, thus decreasing the standard deviation of the estimates, compared to the soil observations. This appears as a smoothing effect on the data and, as a consequence, kriging interpolation is not reliable when the dataset is not sampled with a sampling designs optimised for geostatistics. Depth function approaches, as they are generally applied in literature, implement a spline regression of the soil profile data that aims to better describe the changes of the soil properties with depth. Subsequently, the spline is resampled at determined depths and, for each of these depths, a bi-dimensional (2D) geostatistical interpolation is performed. Consequently, the 3D soil model is a combination of a series of bi-dimensional slices. This approach can effectively decrease or eliminate any smoothing issues, but the way in which the model is created, by combining several 2D horizontal slices, can potentially lead to erroneous estimations. The fact that the geostatistical interpolation is performed in 2D implies that an unsampled location is estimated only by considering values at the same depth, thus excluding the vertical variability from the mapping, and potentially undermining the accuracy of the method. For these reasons, the literature review identified a clear need for developing, a new method for accurately estimating soil properties in 3D – the target of this research, The method studied in this thesis explores the concept of soil specific depth functions, which are simple mathematical equations, chosen for their ability to describe the general profile pattern of a soil dataset. This way, fitting the depth function to a particular sample becomes a diagnostic tool. If the pattern shown in a particular soil profile is dissimilar to the average pattern described by the depth function, it means that in that region there are localised changes in the soil profiles, and these can be identified from the goodness of fit of the function. This way, areas where soil properties have a homogeneous profile pattern can be easily identified and the depth function can be changed accordingly. The application of this new mapping technique is based on the geostatistical interpolation of the depth function coefficients across the study area. Subsequently, the equation is solved for each interpolated location to create a 3D lattice of soil properties estimations. For this way of mapping, this new methodology was denoted as top-down mapping method. The methodology was assessed through three case studies, where the top-down mapping method was developed, tested, and validated. Three datasets of diverse soil properties and at different spatial extents were selected. The results were validated primarily using cross-validation and, when possible, by comparing the estimates with independently sampled datasets (independent validation). In addition, the results were compared with estimates obtained using established literature methods, such as 3D kriging interpolation and the spline approach, in order to define some basic rule of application. The results indicate that the top-down mapping method can be used in circumstances where the soil profiles present a pattern that can be described by a function with maximum three coefficients. If this condition is met, as it was with key soil properties during the research, the top-down mapping method can be used for obtaining reliable estimates at different spatial extents.

Thesis (M.Eng.)--University of Wollongong, 2006. / Typescript. Includes bibliographical references: leaf 174-183.