Estimation or simulation? That is the question

The issue of smoothing in kriging has been addressed either by estimation or simulation. The solution via estimation calls for postprocessing kriging estimates in order to correct the smoothing effect. Stochastic simulation provides equiprobable images presenting no smoothing and reproducing the covariance model. Consequently, these images reproduce both the sample histogram and the sample semivariogram. However, there is still a problem, which is the lack of local accuracy of simulated images. In this paper, a postprocessing algorithm for correcting the smoothing effect of ordinary kriging estimates is compared with sequential Gaussian simulation realizations. Based on samples drawn from exhaustive data sets, the postprocessing algorithm is shown to be superior to any individual simulation realization yet, at the expense of providing one deterministic estimate of the random function.

Reliability assessment of energy limited systems using sequential montecarlo simulation

This paper explains why the reliability assessment of energy limited systems requires more detailed models for primary generating resources availability, internal and external generating dispatch and customer demand than the ones commonly used for large power systems and presents a methodology based on the full sequential Montecarlo simulation technique with AC power flow for their long term reliability assessment which can properly include these detailed models. By means of a real example, it is shown how the simplified modeling traditionally used for large power systems leads to pessimistic predictions if it is applied to an energy limited system and also that it cannot predict all the load point adequacy problems. © 2006 IEEE.

Post-processing of sequential indicator simulation realizations for modeling geologic bodies

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Projected sequential Gaussian processes: flexible interpolation for large data sets

Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential framework for inference in such projected processes is presented, where the observations are considered one at a time. We introduce a C++ library for carrying out such projected, sequential estimation which adds several novel features. In particular we have incorporated the ability to use a generic observation operator, or sensor model, to permit data fusion. We can also cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the variogram parameters is based on maximum likelihood estimation. We illustrate the projected sequential method in application to synthetic and real data sets. We discuss the software implementation and suggest possible future extensions.

Projected sequential Gaussian processes:a C++ tool for interpolation of large datasets with heterogeneous noise

Heterogeneous datasets arise naturally in most applications due to the use of a variety of sensors and measuring platforms. Such datasets can be heterogeneous in terms of the error characteristics and sensor models. Treating such data is most naturally accomplished using a Bayesian or model-based geostatistical approach; however, such methods generally scale rather badly with the size of dataset, and require computationally expensive Monte Carlo based inference. Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential Bayesian framework for inference in such projected processes is presented. The observations are considered one at a time which avoids the need for high dimensional integrals typically required in a Bayesian approach. A C++ library, gptk, which is part of the INTAMAP web service, is introduced which implements projected, sequential estimation and adds several novel features. In particular the library includes the ability to use a generic observation operator, or sensor model, to permit data fusion. It is also possible to cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the covariance parameters is explored, including the impact of the projected process approximation on likelihood profiles. We illustrate the projected sequential method in application to synthetic and real datasets. Limitations and extensions are discussed. © 2010 Elsevier Ltd.

Soil CO(2) emission estimated by different interpolation techniques

Soil CO(2) emissions are highly variable, both spatially and across time, with significant changes even during a one-day period. The objective of this study was to compare predictions of the diurnal soil CO(2) emissions in an agricultural field when estimated by ordinary kriging and sequential Gaussian simulation. The dataset consisted of 64 measurements taken in the morning and in the afternoon on bare soil in southern Brazil. The mean soil CO(2) emissions were significantly different between the morning (4.54 mu mol m(-2) s(-1)) and afternoon (6.24 mu mol m(-2) s(-1)) measurements. However, the spatial variability structures were similar, as the models were spherical and had close range values of 40.1 and 40.0 m for the morning and afternoon semivariograms. In both periods, the sequential Gaussian simulation maps were more efficient for the estimations of emission than ordinary kriging. We believe that sequential Gaussian simulation can improve estimations of soil CO(2) emissions in the field, as this property is usually highly non-Gaussian distributed.

Uncertainties in the prediction of spatial variability of soil CO2 emissions and related properties

The soil CO2 emission has high spatial variability because it depends strongly on soil properties. The purpose of this study was to (i) characterize the spatial variability of soil respiration and related properties, (ii) evaluate the accuracy of results of the ordinary kriging method and sequential Gaussian simulation, and (iii) evaluate the uncertainty in predicting the spatial variability of soil CO2 emission and other properties using sequential Gaussian simulations. The study was conducted in a sugarcane area, using a regular sampling grid with 141 points, where soil CO2 emission, soil temperature, air-filled pore space, soil organic matter and soil bulk density were evaluated. All variables showed spatial dependence structure. The soil CO2 emission was positively correlated with organic matter (r = 0.25, p < 0.05) and air-filled pore space (r = 0.27, p < 0.01) and negatively with soil bulk density (r = -0.41, p < 0.01). However, when the estimated spatial values were considered, the air-filled pore space was the variable mainly responsible for the spatial characteristics of soil respiration, with a correlation of 0.26 (p < 0.01). For all variables, individual simulations represented the cumulative distribution functions and variograms better than ordinary kriging and E-type estimates. The greatest uncertainties in predicting soil CO2 emission were associated with areas with the highest estimated values, which produced estimates from 0.18 to 1.85 t CO2 ha-1, according to the different scenarios considered. The knowledge of the uncertainties generated by the different scenarios can be used in inventories of greenhouse gases, to provide conservative estimates of the potential emission of these gases.

SPATIAL UNCERTAINTY OF NUTRIENT LOSS BY EROSION IN SUGARCANE HARVESTING SCENARIOS

The assessment of spatial uncertainty in the prediction of nutrient losses by erosion associated with landscape models is an important tool for soil conservation planning. The purpose of this study was to evaluate the spatial and local uncertainty in predicting depletion rates of soil nutrients (P, K, Ca, and Mg) by soil erosion from green and burnt sugarcane harvesting scenarios, using sequential Gaussian simulation (SGS). A regular grid with equidistant intervals of 50 m (626 points) was established in the 200-ha study area, in Tabapuã, São Paulo, Brazil. The rate of soil depletion (SD) was calculated from the relation between the nutrient concentration in the sediments and the chemical properties in the original soil for all grid points. The data were subjected to descriptive statistical and geostatistical analysis. The mean SD rate for all nutrients was higher in the slash-and-burn than the green cane harvest scenario (Student’s t-test, p<0.05). In both scenarios, nutrient loss followed the order: Ca>Mg>K>P. The SD rate was highest in areas with greater slope. Lower uncertainties were associated to the areas with higher SD and steeper slopes. Spatial uncertainties were highest for areas of transition between concave and convex landforms.

Analysis and spatial modeling of the indoor radon Swiss data

The present research deals with an important public health threat, which is the pollution created by radon gas accumulation inside dwellings. The spatial modeling of indoor radon in Switzerland is particularly complex and challenging because of many influencing factors that should be taken into account. Indoor radon data analysis must be addressed from both a statistical and a spatial point of view. As a multivariate process, it was important at first to define the influence of each factor. In particular, it was important to define the influence of geology as being closely associated to indoor radon. This association was indeed observed for the Swiss data but not probed to be the sole determinant for the spatial modeling. The statistical analysis of data, both at univariate and multivariate level, was followed by an exploratory spatial analysis. Many tools proposed in the literature were tested and adapted, including fractality, declustering and moving windows methods. The use of Quan-tité Morisita Index (QMI) as a procedure to evaluate data clustering in function of the radon level was proposed. The existing methods of declustering were revised and applied in an attempt to approach the global histogram parameters. The exploratory phase comes along with the definition of multiple scales of interest for indoor radon mapping in Switzerland. The analysis was done with a top-to-down resolution approach, from regional to local lev¬els in order to find the appropriate scales for modeling. In this sense, data partition was optimized in order to cope with stationary conditions of geostatistical models. Common methods of spatial modeling such as Κ Nearest Neighbors (KNN), variography and General Regression Neural Networks (GRNN) were proposed as exploratory tools. In the following section, different spatial interpolation methods were applied for a par-ticular dataset. A bottom to top method complexity approach was adopted and the results were analyzed together in order to find common definitions of continuity and neighborhood parameters. Additionally, a data filter based on cross-validation was tested with the purpose of reducing noise at local scale (the CVMF). At the end of the chapter, a series of test for data consistency and methods robustness were performed. This lead to conclude about the importance of data splitting and the limitation of generalization methods for reproducing statistical distributions. The last section was dedicated to modeling methods with probabilistic interpretations. Data transformation and simulations thus allowed the use of multigaussian models and helped take the indoor radon pollution data uncertainty into consideration. The catego-rization transform was presented as a solution for extreme values modeling through clas-sification. Simulation scenarios were proposed, including an alternative proposal for the reproduction of the global histogram based on the sampling domain. The sequential Gaussian simulation (SGS) was presented as the method giving the most complete information, while classification performed in a more robust way. An error measure was defined in relation to the decision function for data classification hardening. Within the classification methods, probabilistic neural networks (PNN) show to be better adapted for modeling of high threshold categorization and for automation. Support vector machines (SVM) on the contrary performed well under balanced category conditions. In general, it was concluded that a particular prediction or estimation method is not better under all conditions of scale and neighborhood definitions. Simulations should be the basis, while other methods can provide complementary information to accomplish an efficient indoor radon decision making.

Soil CO2 emission estimated by different interpolation techniques

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Uncertainties in the prediction of spatial variability of soil CO2 emissions and related properties

A emissão de CO2 do solo apresenta alta variabilidade espacial, devido à grande dependência espacial observada nas propriedades do solo que a influenciam. Neste estudo, objetivou-se: caracterizar e relacionar a variabilidade espacial da respiração do solo e propriedades relacionadas; avaliar a acurácia dos resultados fornecidos pelo método da krigagem ordinária e simulação sequencial gaussiana; e avaliar a incerteza na predição da variabilidade espacial da emissão de CO2 do solo e demais propriedades utilizando a simulação sequencial gaussiana. O estudo foi conduzido em uma malha amostral irregular com 141 pontos, instalada sobre a cultura de cana-de-açúcar. Nesses pontos foram avaliados a emissão de CO2 do solo, a temperatura do solo, a porosidade livre de água, o teor de matéria orgânica e a densidade do solo. Todas as variáveis apresentaram estrutura de dependência espacial. A emissão de CO2 do solo mostrou correlações positivas com a matéria orgânica (r = 0,25, p < 0,05) e a porosidade livre de água (r = 0,27, p <0,01) e negativa com a densidade do solo (r = -0,41, p < 0,01). No entanto, quando os valores estimados espacialmente (N=8833) são considerados, a porosidade livre de água passa a ser a principal variável responsável pelas características espaciais da respiração do solo, apresentando correlação de 0,26 (p < 0,01). As simulações individuais propiciaram, para todas as variáveis analisadas, melhor reprodução das funções de distribuição acumuladas e dos variogramas, em comparação à krigagem e estimativa E-type. As maiores incertezas na predição da emissão de CO2 estiveram associadas às regiões da área estudada com maiores valores observados e estimados, produzindo estimativas, ao longo do período estudado, de 0,18 a 1,85 t CO2 ha-1, dependendo dos diferentes cenários simulados. O conhecimento das incertezas gerado por meio dos diferentes cenários de estimativa pode ser incluído em inventários de gases do efeito estufa, resultando em estimativas mais conservadoras do potencial de emissão desses gases.

Quantification of uncertainties associated with space-time estimates of short-term soil CO2 emissions in a sugar cane area

The characterization of soil CO2 emissions (FCO2) is important for the study of the global carbon cycle. This phenomenon presents great variability in space and time, a characteristic that makes attempts at modeling and forecasting FCO2 challenging. Although spatial estimates have been performed in several studies, the association of these estimates with the uncertainties inherent in the estimation procedures is not considered. This study aimed to evaluate the local, spatial, local-temporal and spatial-temporal uncertainties of short-term FCO2 after harvest period in a sugar cane area. The FCO2 was featured in a sampling grid of 60m×60m containing 127 points with minimum separation distances from 0.5 to 10m between points. The FCO2 was evaluated 7 times within a total period of 10 days. The variability of FCO2 was described by descriptive statistics and variogram modeling. To calculate the uncertainties, 300 realizations made by sequential Gaussian simulation were considered. Local uncertainties were evaluated using the probability values exceeding certain critical thresholds, while the spatial uncertainties considering the probability of regions with high probability values together exceed the adopted limits. Using the daily uncertainties, the local-spatial and spatial-temporal uncertainty (Ftemp) was obtained. The daily and mean emissions showed a variability structure that was described by spherical and Gaussian models. The differences between the daily maps were related to variations in the magnitude of FCO2, covering mean values ranging from 1.28±0.11μmolm-2s-1 (F197) to 1.82±0.07μmolm-2s-1 (F195). The Ftemp showed low spatial uncertainty coupled with high local uncertainty estimates. The average emission showed great spatial uncertainty of the simulated values. The evaluation of uncertainties associated with the knowledge of temporal and spatial variability is an important tool for understanding many phenomena over time, such as the quantification of greenhouse gases or the identification of areas with high crop productivity. © 2013 Elsevier B.V.

Métodos geoestatísticos na caracterização espacial de óxidos de ferro em diferentes pedoformas

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)

Incerteza espacial da emissão de 'CO IND.2' do solo em dois sistemas de manejo da palhada da cana-de-açúcar

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Incertezas na estimativa da variabilidade espacial da emissão de CO2 do solo e propriedades edáficas em área de cana crua

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES)