Compositional Bayesian indicator estimation

Indicator kriging is widely used for mapping spatial binary variables and for estimating the global and local spatial distributions of variables in geosciences. For continuous random variables, indicator kriging gives an estimate of the cumulative distribution function, for a given threshold, which is then the estimate of a probability. Like any other kriging procedure, indicator kriging provides an estimation variance that, although not often used in applications, should be taken into account as it assesses the uncertainty of the estimate. An alternative approach to indicator estimation is proposed in this paper. In this alternative approach the complete probability density function of the indicator estimate is evaluated. The procedure is described in a Bayesian framework, using a multivariate Gaussian likelihood and an a priori distribution which are both combined according to Bayes theorem in order to obtain a posterior distribution for the indicator estimate. From this posterior distribution, point estimates, interval estimates and uncertainty measures can be obtained. Among the point estimates, the median of the posterior distribution is the maximum entropy estimate because there is a fifty-fifty chance of the unknown value of the estimate being larger or smaller than the median; that is, there is maximum uncertainty in the choice between two alternatives. Thus in some sense, the latter is an indicator estimator, alternative to the kriging estimator, that includes its own uncertainty. On the other hand, the mode of the posterior distribution estimator, assuming a uniform prior, is coincidental with the simple kriging estimator. Additionally, because the indicator estimate can be considered as a two-part composition which domain of definition is the simplex, the method is extended to compositional Bayesian indicator estimation. Bayesian indicator estimation and compositional Bayesian indicator estimation are illustrated with an environmental case study in which the probability of the content of a geochemical element in soil being over a particular threshold is of interest. The computer codes and its user guides are public domain and freely available.     

A kernel spatial density estimation allowing for the analysis of spatial clustering. Application to Monsoon Asia Drought Atlas data

A nonparametric density estimate that incorporates spatial dependency has not been studied in the literature. In this article, we propose a new spatial density estimator that depends on two kernels: one controls the distance between observations while the other controls the spatial dependence structure. The uniform almost sure convergence of the density estimate is established with the rate of convergence. The consistency of the mode of this kernel density is also studied. Then a spatial hierarchical unsupervised clustering algorithm based on the mode estimate is presented. Some simulations as well as an application to the Monsoon Asia Drought Atlas data illustrate the efficiency of our algorithm, and a comparison of the spatial structures of these data detected by the density estimate and clustering algorithm are done.     

Nonparametric bias-corrected variogram estimation under non-constant trend

In geostatistics, the approximation of the spatial dependence structure of a process, through the estimation of the variogram or the covariogram of the variable under consideration, is an important issue. In this work, under a general spatial model, including a mean or trend function, and without assuming any parametric model for this function and for the dependence structure of the process, a general nonparametric estimator of the variogram is proposed. The new approach consists in applying an iterative algorithm, using the residuals obtained from a nonparametric local linear estimation of the trend function, jointly with a correction of the bias due to the use of these residuals. A simulation study checks the validity of the presented approaches in practice. The broad applicability of the procedures is demonstrated on a real data set.     

On the computation of area probabilities based on a spatial stochastic model for precipitation cells and precipitation amounts

A main task of weather services is the issuing of warnings for potentially harmful weather events. Automated warning guidances can be derived, e.g., from statistical post-processing of numerical weather prediction using meteorological observations. These statistical methods commonly estimate the probability of an event (e.g. precipitation) occurring at a fixed location (a point probability). However, there are no operationally applicable techniques for estimating the probability of precipitation occurring anywhere in a geographical region (an area probability). We present an approach to the estimation of area probabilities for the occurrence of precipitation exceeding given thresholds. This approach is based on a spatial stochastic model for precipitation cells and precipitation amounts. The basic modeling component is a non-stationary germ-grain model with circular grains for the representation of precipitation cells. Then, we assign a randomly scaled response function to each precipitation cell and sum these functions up to obtain precipitation amounts. We derive formulas for expectations and variances of point precipitation amounts and use these formulas to compute further model characteristics based on available sequences of point probabilities. Area probabilities for arbitrary areas and thresholds can be estimated by repeated Monte Carlo simulation of the fitted precipitation model. Finally, we verify the proposed model by comparing the generated area probabilities with independent rain gauge adjusted radar data. The novelty of the presented approach is that, for the first time, a widely applicable estimation of area probabilities is possible, which is based solely on predicted point probabilities (i.e., neither precipitation observations nor further input of the forecaster are necessary). Therefore, this method can be applied for operational weather predictions.     

Estimating threshold-exceeding probability maps of environmental variables with Markov chain random fields

Estimating and mapping spatial uncertainty of environmental variables is crucial for environmental evaluation and decision making. For a continuous spatial variable, estimation of spatial uncertainty may be conducted in the form of estimating the probability of (not) exceeding a threshold value. In this paper, we introduced a Markov chain geostatistical approach for estimating threshold-exceeding probabilities. The differences of this approach compared to the conventional indicator approach lie with its nonlinear estimators—Markov chain random field models and its incorporation of interclass dependencies through transiograms. We estimated threshold-exceeding probability maps of clay layer thickness through simulation (i.e., using a number of realizations simulated by Markov chain sequential simulation) and interpolation (i.e., direct conditional probability estimation using only the indicator values of sample data), respectively. To evaluate the approach, we also estimated those probability maps using sequential indicator simulation and indicator kriging interpolation. Our results show that (i) the Markov chain approach provides an effective alternative for spatial uncertainty assessment of environmental spatial variables and the probability maps from this approach are more reasonable than those from conventional indicator geostatistics, and (ii) the probability maps estimated through sequential simulation are more realistic than those through interpolation because the latter display some uneven transitions caused by spatial structures of the sample data.     

Estimation of rainfall elasticity of streamflow in Australia

Abstract

Estimates of rainfall elasticity of streamflow in 219 catchments across Australia are presented. The rainfall elasticity of streamflow is defined here as the proportional change in mean annual streamflow divided by the proportional change in mean annual rainfall. The elasticity is therefore a simple estimate of the sensitivity of long-term streamflow to changes in long-term rainfall, and is particularly useful as an initial estimate of climate change impact in land and water resources projects. The rainfall elasticity of streamflow is estimated here using a hydrological modelling approach and a nonparametric estimator. The results indicate that the rainfall elasticity of streamflow (? _P ) in Australia is about 2.0–3.5 (observed in about 70% of the catchments), that is, a 1% change in mean annual rainfall results in a 2.0–3.5% change in mean annual streamflow. The rainfall elasticity of streamflow is strongly correlated to runoff coefficient and mean annual rainfall and streamflow, where streamflow is more sensitive to rainfall in drier catchments, and those with low runoff coefficients. There is a clear relation-ship between the ? _P values estimated using the hydrological modelling approach and those estimated using the nonparametric estimator for the 219 catchments, although the values estimated by the hydrological modelling approach are, on average, slightly higher. The modelling approach is useful where a detailed study is required and where there are sufficient data to reliably develop and calibrate a hydrological model. The nonparametric estimator is useful where consistent estimates of the sensitivity of long-term streamflow to climate are required, because it is simple to use and estimates the elasticity directly from the historical data. The nonparametric method, being model independent, can also be easily applied in comparative studies to data sets from many catchments across large regions.     

Regional flood frequency analysis using residual kriging in physiographical space

In this article, an approach using residual kriging (RK) in physiographical space is proposed for regional flood frequency analysis. The physiographical space is constructed using physiographical/climatic characteristics of gauging basins by means of canonical correlation analysis (CCA). This approach is a modified version of the original method, based on ordinary kriging (OK). It is intended to handle effectively any possible spatial trends within the hydrological variables over the physiographical space. In this approach, the trend is first quantified and removed from the hydrological variable by a quadratic spatial regression. OK is therefore applied to the regression residual values. The final estimated value of a specific quantile at an ungauged station is the sum of the spatial regression estimate and the kriged residual. To evaluate the performance of the proposed method, a cross‐validation procedure is applied. Results of the proposed method indicate that RK in CCA physiographical space leads to more efficient estimates of regional flood quantiles when compared to the original approach and to a straightforward regression‐based estimator. Copyright © 2010 John Wiley & Sons, Ltd.     

Comparison of two approaches for generation of daily rainfall data

There has been extensive research on the problem of stochastically generating daily rainfall sequences for use in water management applications. Srikanthan and McMahon [Australia Water Resources Council, Canberra, 1985] proposed a transition probability matrix (TPM) model that performs better for Australian rainfall than many alternative models, particularly where long records (say 100 years) are available. Boughton [Report 99/9, CRC for Catchment Hydrology, Monash University, Melbourne, 21pp, 1999] incorporated an empirical adjustment into the TPM model that allows the model to reproduce the observed variability in the annual rainfall. More recently, Harrold et al. [Water Resour Res 39(10, 12):1300, 1343, 2003a,b] proposed nonparametric models for the generation of daily rainfall occurrences and rainfall amounts on wet days. By conditioning on short, medium and long-term characteristics, this approach is also able to preserve the variability in annual rainfall. In this study, the above two approaches were used to generate daily rainfall data for Sydney and Melbourne, and the results evaluated. Both approaches preserved most of the daily, monthly and annual characteristics that were compared, with the nonparametric approach providing marginally better performance at the cost of greater model complexity. The nonparametric approach was also able to preserve the variability and persistence in the annual number of wet days.     

A note on point process analysis of Chinese earthquake data

In the analysis of Chinese earthquake history byLee andBrillinger (1979), the autointensity function of the earthquake sequence is estimated, based on a parametric model for the catalog inclusion probabilities. We show that an estimate of the autointensity function with essentially the same features can be obtained by smoothing the data to yield a nonparametric estimate of the inclusion probabilities. In addition, however, we illustrate that a somewhat less smooth estimate would lead to an autointensity function with little structure. The shape of the estimated autointensity function is hence dependent on the assumed smoothness of the inclusion probabilities. This is an issue which cannot be resolved statistically.     

Unbiased estimation of flood risk with the GEV distribution

Conventional flood frequency analysis is concerned with providing an unbiased estimate of the magnitude of the design flow exceeded with the probabilityp, but sampling uncertainties imply that such estimates will, on average, be exceeded more frequently. An alternative approach is therefore, to derive an estimator which gives an unbiased estimate of flow risk: the difference between the two magnitudes reflects uncertainties in parameter estimation. An empirical procedure has been developed to estimate the mean true exceedance probabilities of conventional estimates made using a GEV distribution fitted by probability weighted moments, and adjustment factors have been determined to enable the estimation of flood magnitudes exceeded with, on average, the desired probability.     

Evaluation of a surface energy balance method based on optical and thermal satellite imagery to estimate root‐zone soil moisture

Various remote‐sensing methods are available to estimate soil moisture, but few address the fine spatial resolutions (e.g. 30‐m grid cells) and root‐zone depth requirements of agricultural and other similar applications. One approach that has been previously proposed to estimate fine‐resolution soil moisture is to first estimate the evaporative fraction from an energy balance that is inferred from optical and thermal remote‐sensing images [e.g. using the Remote Sensing of Evapotranspiration (ReSET) algorithm] and then estimate soil moisture through an empirical relationship to evaporative fraction. A similar approach has also been proposed to estimate the degree of saturation. The primary objective of this study is to evaluate these methods for estimating soil moisture and degree of saturation, particularly for a semi‐arid grassland with relatively dry conditions. Soil moisture was monitored at 28 field locations in south‐eastern Colorado with herbaceous vegetation during the summer months of 3 years. In situ soil moisture and degree of saturation observations are compared with estimates calculated from Landsat imagery using the ReSET algorithm. The in situ observations suggest that the empirical relationships with evaporative fraction that have been proposed in previous studies typically provide overestimates of soil moisture and degree of saturation in this region. However, calibrated functions produce estimates with an accuracy that may be adequate for various applications. The estimates produced by this approach are more reliable for degree of saturation than for soil moisture, and the method is more successful at identifying temporal variability than spatial variability in degree of saturation for this region. Copyright © 2015 John Wiley & Sons, Ltd.     

Sequential kriging and cokriging: Two powerful geostatistical approaches

A sequential linear estimator is developed in this study to progressively incorporate new or different spatial data sets into the estimation. It begins with a classical linear estimator (i.e., kriging or cokriging) to estimate means conditioned to a given observed data set. When an additional data set becomes available, the sequential estimator improves the previous estimate by using linearly weighted sums of differences between the new data set and previous estimates at sample locations. Like the classical linear estimator, the weights used in the sequential linear estimator are derived from a system of equations that contains covariances and cross-covariances between sample locations and the location where the estimate is to be made. However, the covariances and cross-covariances are conditioned upon the previous data sets. The sequential estimator is shown to produce the best, unbiased linear estimate, and to provide the same estimates and variances as classic simple kriging or cokriging with the simultaneous use of the entire data set. However, by using data sets sequentially, this new algorithm alleviates numerical difficulties associated with the classical kriging or cokriging techniques when a large amount of data are used. It also provides a new way to incorporate additional information into a previous estimation.     

A fully non-stationary linear coregionalization model for multivariate random fields

This paper introduces an extension of the traditional stationary linear coregionalization model to handle the lack of stationarity. Under the proposed model, coregionalization matrices are spatially dependent, and basic univariate spatial dependence structures are non-stationary. A parameter estimation procedure of the proposed non-stationary linear coregionalization model is developed under the local stationarity framework. The proposed estimation procedure is based on the method of moments and involves a matrix-valued local stationary variogram kernel estimator, a weighted local least squares method in combination with a kernel smoothing technique. Local parameter estimates are knitted together for prediction and simulation purposes. The proposed non-stationary multivariate spatial modeling approach is illustrated using two real bivariate data examples. Prediction performance comparison is carried out with the classical stationary multivariate spatial modeling approach. According to several criteria, the prediction performance of the proposed non-stationary multivariate spatial modeling approach appears to be significantly better.     

Probabilistic data integration for characterization of spatial distribution of residual LNAPL

The spatial distribution of residual light non-aqueous phase liquid (LNAPL) is an important factor in reactive solute transport modeling studies. There is great uncertainty associated with both the areal limits of LNAPL source zones and smaller scale variability within the areal limits. A statistical approach is proposed to construct a probabilistic model for the spatial distribution of residual NAPL and it is applied to a site characterized by ultra-violet-induced-cone-penetration testing (CPT–UVIF). The uncertainty in areal limits is explicitly addressed by a novel distance function (DF) approach. In modeling the small-scale variability within the areal limits, the CPT–UVIF data are used as primary source of information, while soil texture and distance to water table are treated as secondary data. Two widely used geostatistical techniques are applied for the data integration, namely sequential indicator simulation with locally varying means (SIS–LVM) and Bayesian updating (BU). A close match between the calibrated uncertainty band (UB) and the target probabilities shows the performance of the proposed DF technique in characterization of uncertainty in the areal limits. A cross-validation study also shows that the integration of the secondary data sources substantially improves the prediction of contaminated and uncontaminated locations and that the SIS–LVM algorithm gives a more accurate prediction of residual NAPL contamination. The proposed DF approach is useful in modeling the areal limits of the non-stationary continuous or categorical random variables, and in providing a prior probability map for source zone sizes to be used in Monte Carlo simulations of contaminant transport or Monte Carlo type inverse modeling studies.     

Risk assessment of an interacting structure–soil system due to liquefaction

The present study deals with the non‐linear stochastic dynamic analysis of a soil–structure interacting system. The ultimate objective is to determine the risk of damage to the system due to liquefaction under a wide range of earthquake intensities. A Monte Carlo simulation approach is followed in conjunction with advanced finite element procedures. The stochastic spatial variability of soil properties and the randomness of the seismic excitation are taken into account in order to estimate the statistics of the response, measured in terms of uniform foundation settlement and tilting. Specifically, soil properties are modelled as non‐Gaussian random fields and seismic excitations as non‐stationary random processes. The probabilistic characteristics of the stochastic field modelling soil properties are established from in situ tests. The risk of damage to the soil–structure system due to liquefaction is assessed by establishing fragility curves, which are of paramount importance for risk assessment and management studies of such systems. Fragility curves express the probability of exceeding various thresholds in the response. The relative effect of the variability of various soil parameters on the variability of the response is also examined. Copyright © 2002 John Wiley & Sons, Ltd.     

A distribution function based bandwidth selection method for kernel quantile estimation

The key problem in nonparametric frequency analysis of flood and droughts is the estimation of the bandwidth parameter which defines the degree of smoothing. Most of the proposed bandwidth estimators have been based on the density function rather than the cumulative distribution function or the quantile that are the primary interest in frequency analysis. We propose a new bandwidth estimator derived from properties of quantile estimators. The estimator builds on work by Altman and Léger (1995). The estimator is compared to the well-known method of least squares cross-validation (LSCV) using synthetic data generated from various parametric distributions used in hydrologic frequency analysis. Simulations suggest that our estimator performs at least as well as, and in many cases better than, the method of LSCV. In particular, the use of the proposed plug-in estimator reduces bias in the estimation as compared to LSCV. When applied to data sets containing observations with identical values, typically the result of rounding or truncation, the LSCV and most other techniques generally underestimates the bandwidth. The proposed technique performs very well in such situations.     

Estimation and prediction of the spatial occurrence of fish species using Bayesian latent Gaussian models

A methodological approach for modelling the occurrence patterns of species for the purpose of fisheries management is proposed here. The presence/absence of the species is modelled with a hierarchical Bayesian spatial model using the geographical and environmental characteristics of each fishing location. Maps of predicted probabilities of presence are generated using Bayesian kriging. Bayesian inference on the parameters and prediction of presence/absence in new locations (Bayesian kriging) are made by considering the model as a latent Gaussian model, which allows the use of the integrated nested Laplace approximation ( INLA ) software (which has been seen to be quite a bit faster than the well-known MCMC methods). In particular, the spatial effect has been implemented with the stochastic partial differential equation (SPDE) approach. The methodology is evaluated on Mediterranean horse mackerel (Trachurus mediterraneus) in the Western Mediterranean. The analysis shows that environmental and geographical factors can play an important role in directing local distribution and variability in the occurrence of species. Although this approach is used to recognize the habitat of mackerel, it could also be for other different species and life stages in order to improve knowledge of fish populations and communities.     

Geostatistical simulation of spatial variability of convective storms in Mexico City Valley

Precipitation is one of the main components of the hydrological cycle and knowledge of its spatial distribution is fundamental for the prediction of other closely related environmental variables, for example, runoff, flooding and aquifer recharge. Most of the precipitation in Mexico City is due to convective storms characterized by a high spatial variability, implying that modeling its behavior is very complex. In this work stochastic simulation techniques with a geostatistical approach were applied to model the spatial variability of the rainfall of three convective storms. The analysis of the results shows that using the proposed methodology spatial distributions of rain are obtained that reproduce the statistical characteristics present in the available information.     

Probabilistic site-dependent non-linear spectra

This paper presents a probabilistic approach to the estimation of lateral strengths required to provide an adequate control of inelastic deformations in structures during severe earthquake ground motions. In contrast to a deterministic approach, the approach presented herein accounts explicitly for the variability of the response of non-linear systems due to the inherent uncertainties in the intensity and characteristics of the input excitation by considering the probability distribution of maximum inelastic strength demands. This study is based on the computation of non-linear strength demands of single-degree-of-freedom (SDOF) systems experiencing different levels of inelastic deformation when subjected to 124 recorded earthquake ground motions. Using empirical cumulative distribution functions site-dependent probabilistic non-linear spectra were computed for six probabilities of exceedance of different levels of inelastic deformation. It is concluded that the lateral strength required to control displacement ductility demands is significantly affected by the maximum tolerable inelastic deformation, the system's period of vibration, the local site conditions and the level of risk in exceeding the maximum tolerable deformations.