Variogram Matrix Functions for Vector Random Fields with Second-Order Increments

The variogram matrix function is an important measure for the dependence of a vector random field with second-order increments, and is a useful tool for linear predication or cokriging. This paper proposes an efficient approach to construct variogram matrix functions, based on three ingredients: a univariate variogram, a conditionally negative definite matrix, and a Bernstein function, and derives three classes of variogram matrix functions for vector elliptically contoured random fields. Moreover, various dependence structures among components can be derived through appropriate mixture procedures demonstrated in this paper. We also obtain covariance matrix functions for second-order vector random fields through the Schoenberg–Lévy kernels.     

Support vector machine applied to settlement of shallow foundations on cohesionless soils

The determination of settlement of shallow foundations on cohesionless soil is an important task in geotechnical engineering. Available methods for the determination of settlement are not reliable. In this study, the support vector machine (SVM), a novel type of learning algorithm based on statistical theory, has been used to predict the settlement of shallow foundations on cohesionless soil. SVM uses a regression technique by introducing an ε – insensitive loss function. A thorough sensitive analysis has been made to ascertain which parameters are having maximum influence on settlement. The study shows that SVM has the potential to be a useful and practical tool for prediction of settlement of shallow foundation on cohesionless soil.     

Analysis of soil erosion characteristics in small watersheds with particle swarm optimization, support vector machine, and artificial neuronal networks

Sand production by soil erosion in small watershed is a complex physical process. There are few physical models suitable to describe the characteristics of the intense erosion in domestic loess plateau. Introducing support vector machine (SVM) oriented to small sample data and possessing good extension property can be an effective approach to predict soil erosion because SVM has been applied in hydrological prediction to some extent. But there are no effective methods to select the rational parameters for SVM, which seriously limited the practical application of SVM. This paper explored the application of intelligence-based particle swarm optimization (PSO) algorithm in automatic selection of parameters for SVM, and proposed a prediction model by linking PSO and SVM for small sample data analysis. This method utilized the high efficiency optimization property and swarm paralleling property of PSO algorithm and the relatively strong learning and extending capacity of SVM. For an example of Huangfuchuan small watershed, its intensive fragmentation and intense erosion earn itself the name of “worst erosion in the world”. Using four characteristics selection algorithms of correlation feature selection, the primary affecting factors for soil erosion in this small watershed were determined to be the channel density, ravine area, sand rock proportion, and the total vegetation coverage. Based on the proposed PSO–SVM algorithm, the soil erosion modulus in the small watershed was predicted. The accuracy of the simulation and prediction was good, and the average error was 3.85%. The SVM predicting model was based on the monitoring data of sand production. The construction of the SVM erosion modulus prediction model for the small watershed comprehensively reflected the complex mechanism of soil erosion and sand production. It had certain advantage and relatively high practical value in small sample prediction in the discipline of soil erosion.     

Multi-model multi-analysis ensemble weather forecasting on the grid for the South Eastern Mediterranean Region

Weather forecasting is based on the use of numerical weather prediction (NWP) models that are able to perform the necessary calculations that describe/predict the major atmospheric processes. One common problem in weather forecasting derives from the uncertainty related to the chaotic behaviour of the atmosphere. A solution to that problem is to perform in addition to “deterministic” forecasts, “stochastic” forecasts that provide an estimate of the prediction skill. A computationally feasible approach towards this aim is to perform “ensemble forecasts”. Indeed, in the frame of SEE-GRID-SCI EU funded project a Regional scale Multi-model, Multi-analysis ensemble forecasting system (REFS) was built and ported on the Grid infrastructure. REFS is based on the use of four limited area models (namely BOLAM, MM5, ETA, and NMM) that are run using a multitude of initial and boundary conditions over the Mediterranean. This paper presents the tools and procedures followed for developing this application at a production level.     

Predicting Lahar-Inundation Zones: Case Study in West Mount Pinatubo, Philippines

This paper demonstrates techniques for pre-eruption prediction of lahar-inundation zones in areas where a volcano has not erupted within living memory and/or where baseline geological information about past lahars could be scarce or investigations to delimit past lahars might be incomplete. A lahar source (or proximal lahar-inundation) zone is predicted based on ratio of vertical descent to horizontal run-out of eruptive deposits that spawn lahars. Immediate post-eruption distal lahar-inundation zones are predicted based on “pre-eruption” distal lahar-inundation zones and on spatial factors derived from a digital elevation model. Susceptibility to distal lahar-inundation is estimated by weights-of-evidence, by logistic regression and by evidential belief functions. Predictive techniques are applied using a geographic information system and are tested in western part of Pinatubo volcano (Philippines). Predictive maps are compared with a forecast volcanic-hazard map through validation against a field-based volcanic-hazard map. The predictive model of proximal lahar-inundation zone has “true positive” prediction accuracy, “true negative” prediction accuracy, “false positive” prediction error and “false negative” prediction error that are similar to those of the forecast volcanic-hazard map. The predictive models of distal lahar inundation zones have higher “true positive” prediction accuracy and lower “false negative” prediction error than the forecast volcanic-hazard map, although the latter has higher “true negative” prediction accuracy and lower “false positive” prediction error than the former. The results illustrate utility of proposed predictive techniques in providing geo-information could be used, howbeit with caution, for planning to mitigate potential lahar hazards well ahead of an eruption that could generate substantial source materials for lahar formation.     

Development of a framework of automated water quality parameter optimization and its application

This paper presents a methodology and framework for the development of an automated least-squares optimization tool for calibrating water quality parameters in QUAL2E. The method has been applied to estimate the optimal water quality parameters in simulation of stream water quality for the Anyang stream in Korea. The Monte Carlo analysis is used to assess the relative importance of model parameters for water quality constituents. It is found that μ_max and ρ are the most influential parameters for Chlorophyll-a modeling and K ₁ and K ₃ are critical parameters for variation of DO and BOD in the Anyang stream. A computer program for automated parameter calibration has been developed using a nonlinear GRG optimization algorithm. The application framework provides an intuitive and easy-to-use interface and allows visual evaluation of results. According to the simulation results, the automated approach is computationally efficient for evaluation of model parameters and converges on a best fit more rapidly and reliably than a trial and error method. The methodology proposed herein can be extended to other models to obtain the best possible parameter values.     

Artificial Neural Networks: A Solution to the Ambiguity in Prediction of Engineering Properties of Fine-Grained Soils

Determination of soaked california bearing ratio (CBR) and compaction characteristics of soils in the laboratory require considerable time and effort. To make a preliminary assessment of the suitability of soils required for a project, prediction models for these engineering properties on the basis of laboratory tests—which are quick to perform, less time consuming and cheap—such as the tests for index properties of soils, are preferable. Nevertheless researchers hold divergent views regarding the most influential parameters to be taken into account for prediction of soaked CBR and compaction characteristics of fine-grained soils. This could be due to the complex behaviour of soils—which, by their very nature, exhibit extreme variability. However this disagreement is a matter of concern as it affects the dependability of prediction models. This study therefore analyses the ability of artificial neural networks and multiple regression to handle different influential parameters simultaneously so as to make accurate predictions on soaked CBR and compaction characteristics of fine-grained soils. The results of simple regression analyses included in this study indicate that optimum moisture content (OMC) and maximum dry density (MDD) of fine-grained soils bear better correlation with soaked CBR of fine-grained soils than plastic limit and liquid limit. Simple regression analyses also indicate that plastic limit has stronger correlation with compaction characteristics of fine-grained soils than liquid limit. On the basis of these correlations obtained using simple regression analyses, neural network prediction models and multiple regression prediction models—with varying number of input parameters are developed. The results reveal that neural network models have more ability to utilize relatively less influential parameters than multiple regression models. The study establishes that in the case of neural network models, the relatively less powerful parameters—liquid limit and plastic limit can also be used effectively along with MDD and OMC for better prediction of soaked CBR of fine-grained soils. Also with the inclusion of less significant parameter—liquid limit along with plastic limit the predictions on compaction characteristics of fine-grained soils using neural network analysis improves considerably. Thus in the case of neural network analysis, the use of relatively less influential input parameters along with stronger parameters is definitely beneficial, unlike conventional statistical methods—for which, the consequence of this approach is unpredictable—giving sometimes not so favourable results. Very weak input parameters alone need to be avoided for neural network analysis. Consequently, when there is ambiguity regarding the most influential input parameters, neural network analysis is quite useful as all such influential parameters can be taken to consideration simultaneously, which will only improve the performance of neural network models. As soils by their very nature, exhibit extreme complexity, it is necessary to include maximum number of influential parameters—as can be determined easily using simple laboratory tests—in the prediction models for soil properties, so as to improve the reliability of these models—for which, use of neural networks is more desirable.     

Palaeoclimate estimates for the Middle Miocene Schrotzburg flora (S Germany): a multi-method approach

We present a detailed palaeoclimate analysis of the Middle Miocene (uppermost Badenian–lowermost Sarmatian) Schrotzburg locality in S Germany, based on the fossil macro- and micro-flora, using four different methods for the estimation of palaeoclimate parameters: the coexistence approach (CA), leaf margin analysis (LMA), the Climate-Leaf Analysis Multivariate Program (CLAMP), as well as a recently developed multivariate leaf physiognomic approach based on an European calibration dataset (ELPA). Considering results of all methods used, the following palaeoclimate estimates seem to be most likely: mean annual temperature ∼15–16°C (MAT), coldest month mean temperature ∼7°C (CMMT), warmest month mean temperature between 25 and 26°C, and mean annual precipiation ∼1,300 mm, although CMMT values may have been colder as indicated by the disappearance of the crocodile Diplocynodon and the temperature thresholds derived from modern alligators. For most palaeoclimatic parameters, estimates derived by CLAMP significantly differ from those derived by most other methods. With respect to the consistency of the results obtained by CA, LMA and ELPA, it is suggested that for the Schrotzburg locality CLAMP is probably less reliable than most other methods. A possible explanation may be attributed to the correlation between leaf physiognomy and climate as represented by the CLAMP calibration data set which is largely based on extant floras from N America and E Asia and which may be not suitable for application to the European Neogene. All physiognomic methods used here were affected by taphonomic biasses. Especially the number of taxa had a great influence on the reliability of the palaeoclimate estimates. Both multivariate leaf physiognomic approaches are less influenced by such biasses than the univariate LMA. In combination with previously published results from the European and Asian Neogene, our data suggest that during the Neogene in Eurasia CLAMP may produce temperature estimates, which are systematically too cold as compared to other evidence. This pattern, however, has to be further investigated using additional palaeofloras.Dedicated to Prof. Dr. Harald Walther, Dresden, on the occasion of his 75th birthday.     

Procedure for the selection of runout model parameters from landslide back-analyses: application to the Metropolitan Area of San Salvador, El Salvador

The existing procedures for the selection of runout model parameters from back-analyses do not allow integrating different types of runout criteria and generally lack a systematic approach. A new method based on receiver operating characteristic (ROC) analyses and aimed at overcoming these limitations is herein proposed. The method consists of estimating discrete classifiers for every runout simulation associated with a set of model parameters. The set of parameters that yields the best prediction is selected using ROC metrics and space. The procedure is illustrated with the back-analyses of a rainfall-triggered debris flow that killed 300–500 people in the Metropolitan Area of San Salvador in 1982. The selected model parameters are used to estimate forward predictions for scenarios that correspond to different return periods. The proposed procedure may be useful in the assessment of areas potentially affected by landslides. In turn, this information can be used in the production or updating of land use plans and zonations, similar to that currently being carried out by the Office for Urban Planning of the Metropolitan Area of San Salvador in El Salvador.     

A new approach for crude oil price prediction based on stream learning

Crude oil is the world's leading fuel, and its prices have a big impact on the global environment, economy as well as oil exploration and exploitation activities. Oil price forecasts are very useful to industries, governments and individuals. Although many methods have been developed for predicting oil prices, it remains one of the most challenging forecasting problems due to the high volatility of oil prices. In this paper, we propose a novel approach for crude oil price prediction based on a new machine learning paradigm called stream learning. The main advantage of our stream learning approach is that the prediction model can capture the changing pattern of oil prices since the model is continuously updated whenever new oil price data are available, with very small constant overhead. To evaluate the forecasting ability of our stream learning model, we compare it with three other popular oil price prediction models. The experiment results show that our stream learning model achieves the highest accuracy in terms of both mean squared prediction error and directional accuracy ratio over a variety of forecast time horizons.     

Application of a Generalised Influence Function Method for Subsidence Prediction in Multi-seam Longwall Extraction

The purpose of mining subsidence prediction is to produce a reliable assessment of ground movement arising from underground mineral extraction. The results of the prediction are used to assess the likelihood of the associated effects on surface structures. In most countries, the assessment of mining subsidence has become an essential part of mining plans, which must be approved by relevant government bodies and mining regulators. It is therefore important to develop a subsidence prediction method that is suitable for a particular country or mine field. Further to the recent development of a Generalised Influence Function Method (GIFM) for subsidence prediction at RMIT University, a case study in Hunter coalfield of in New South Wales, Australia is presented to illustrate the applicability of the GIFM approach for subsidence prediction in multi-seam longwall mining. A computer program is used to calculate subsidence, horizontal displacement and principle strains arising from the extraction of longwall panels. The observed subsidence across the longwall panels and the corresponding ground movements are compared to the model’s output and the results analysed. A discussion of the discrepancies between the GIFM models and the behaviour of complex geological strata is presented. The GIFM method is found to be a powerful tool when applied to complex extraction configurations and can produce useful output for mining subsidence assessments. Of particular importance is its ability to provide both tensile and compressive strain information over the whole affected areas which would otherwise not have been available for the assessment of damage potential to surface structures.     

Application of a fuzzy operator to susceptibility estimations of coal mine subsidence in Taebaek City,Korea

Quantitative determination of locations vulnerable to ground subsidence at mining regions is necessary for effective prevention. In this paper, a method of constructing subsidence susceptibility maps based on fuzzy relations is proposed and tested at an abandoned underground coal mine in Korea. An advantage of fuzzy combination operators over other methods is that the operation is mathematically and logically easy to understand and its implementation to GIS software is simple and straightforward. A certainty factor analysis was used for estimating the relative weight of eight major factors influencing ground subsidence. The relative weight of each factor was then converted into a fuzzy membership value and integrated as a subsidence hazard index using fuzzy combination operators, which produced coal mine subsidence susceptibility maps. The susceptibility maps were compared with the reported ground subsidence areas, and the results showed high accuracy between our prediction and the actual subsidence. Based on the root mean square error and accuracy in terms of success rates, fuzzy γ-operator with a low γ value and fuzzy algebraic product operator, specifically, are useful for ground subsidence prediction. Comparing the results of a fuzzy γ-operator and a conventional logistic regression model, the performance of the fuzzy approach is comparative to that of a logistic regression model with improved computational. A field survey done in the area supported the method’s reliability. A combination of certainty factor analysis and fuzzy relations with a GIS is an effective method to determine locations vulnerable to coal mine subsidence.     

Topic 3: State of the art of mathematical modelling of agricultural impact on groundwater pollution

The state of the art of mathematical modelling of groundwater pollution by agricultural activities is well advanced at the present time and a large number of models are available for a wide variety of water pollution problems. Also a variety of computational methods are available from very simple ones to the most advanced. In general, there are two ways of obtaining a water quality model: “from below,” when water quality data based on actual observation or measurement are explicitly analyzed by the calculus; and “from above,” when the calculus is available in advance as a theory and the user tries to find its interpretation for a given case and from rather limited data. Classifications may be based on the type of modelled processes, the mathematical approach, or on the purpose of the models. Eleven papers were presented in the Symposium on this topic, with most of the models being of the dynamic type.     

Comparison of numerical models of two debris flows in the Cortina d’ Ampezzo area,Dolomites, Italy

The accurate prediction of runout distances, velocities and the knowledge of flow rheology can reduce the casualties and property damage produced by debris flows, providing a means to delineate hazard areas, to estimate hazard intensities for input into risk studies and to provide parameters for the design of protective measures. The application of most of models that describe the propagation and deposition of debris flow requires detailed topography, rheological and hydrological data that are not always available for the debris-flow hazard delineation and estimation. In the Cortina d’Ampezzo area, Eastern Dolomites, Italy, most of the slope instabilities are represented by debris flows; 325 debris-flow prone watersheds have been mapped in the geomorphological hazard map of this area. We compared the results of simulations of two well-documented debris flows in the Cortina d’Ampezzo area, carried on with two different single-phase, non-Newtonian models, the one-dimensional DAN-W and the two-dimensional FLO-2D, to test the possibility to simulate the dynamic behaviour of a debris flow with a model using a limited range of input parameters. FLO-2D model creates a more accurate representation of the hazard area in terms of flooded area, but the results in terms of runout distances and deposits thickness are similar to DAN-W results. Using DAN-W, the most appropriate rheology to describe the debris-flow behaviour is the Voellmy model. When detailed topographical, rheological and hydrological data are not available, DAN-W, which requires less detailed data, is a valuable tool to predict debris-flow hazard. Parameters obtained through back-analysis with both models can be applied to predict hazard in other areas characterized by similar geology, morphology and climate.     

Inverse analysis for geomaterial parameter identification using Pareto multiobjective optimization

An inverse analysis method that combines the back propagation neural network (BPNN) and vector evaluated genetic algorithm (VEGA) was proposed to identify mechanical geomaterial parameters for a more accurate prediction of deformation. The BPNN is used to replace the time‐consuming numerical calculations, thus enhancing the efficiency of the inverse analysis. The VEGA is used to find the Pareto‐optimal solutions to multiobjective functions. Unlike traditional back‐analysis methods which are based on only 1 type of field measurement and a single objective function, this proposed method can consider multiple field observations simultaneously. The proposed method was applied to the Shapingba foundation pit excavation located in Chongqing city, China. Two types of measurements are considered in the method simultaneously: the displacements in the x‐direction (north orientation) and those in the y‐direction (east orientation). Five deformation modulus parameters for artificial backfill soil, silty clay, siltstone, sandstone, and mudstone were selected as the inversion parameters. Compared with the weighted sum approach, the proposed method was demonstrated as an efficient multi‐objective optimization tool for back calculating undetermined parameters. After performing a forward‐calculation using the optimized parameters obtained by the inverse analysis, the predicted results were well consistent with the practical deformation in magnitude and trend.     

Modeling tunneling-induced ground surface settlement development using a wavelet smooth relevance vector machine

Accurate prediction of ground surface settlement is necessary for effectively controlling the settlement that develops during tunneling. Many models have been established for this purpose by extracting the relationship between the settlement and the factors that influence it. However, most of the models focused on the maximum ground surface settlement and do not involve dynamic and real-time predictions. This paper investigated how tunneling-induced ground surface settlement developed using a smooth relevance vector machine with a wavelet kernel (wsRVM). Various factors that affect this settlement, including geometrical, geological and shield operational parameters were considered. The model was applied to earth pressure balance (EPB) shield-driven tunnels. The results indicate that the prediction model performs well and that the distribution of the predictions can provide a measure of the prediction uncertainty. Unlike conventional methods that requireadditional efforts to determine relevant model parameters, the proposed method can optimize the parameters in the training process. The results of the parametric study conducted show that the model performance can be improved by the optimization and that the method can serve as a simple tool for practitioners to use in estimating ground surface settlement development during tunneling.     

Effects of Rock Properties on Specific Cutting Energy in Linear Cutting of Sandstones by Picks

Summary. Specific cutting energy (SE) has been widely used to assess the rock cuttability for mechanical excavation purposes. Some prediction models were developed for SE through correlating rock properties with SE values. However, some of the textural and compositional rock parameters i.e. texture coefficient and feldspar, mafic, and felsic mineral contents were not considered. The present study is to investigate the effects of previously ignored rock parameters along with engineering rock properties on SE. Mineralogical and petrographic analyses, rock mechanics, and linear rock cutting tests were performed on sandstone samples taken from sites around Ankara, Turkey. Relationships between SE and rock properties were evaluated using bivariate correlation and linear regression analyses. The tests and subsequent analyses revealed that the texture coefficient and feldspar content of sandstones affected rock cuttability, evidenced by significant correlations between these parameters and SE at a 90% confidence level. Felsic and mafic mineral contents of sandstones did not exhibit any statistically significant correlation against SE. Cementation coefficient, effective porosity, and pore volume had good correlations against SE. Poisson’s ratio, Brazilian tensile strength, Shore scleroscope hardness, Schmidt hammer hardness, dry density, and point load strength index showed very strong linear correlations against SE at confidence levels of 95% and above, all of which were also found suitable to be used in predicting SE individually, depending on the results of regression analysis, ANOVA, Student’s t-tests, and R² values. Poisson’s ratio exhibited the highest correlation with SE and seemed to be the most reliable SE prediction tool in sandstones.     

An extreme learning machine approach for slope stability evaluation and prediction

This paper presents slope stability evaluation and prediction with the approach of a fast robust neural network named the extreme learning machine (ELM). The circular failure mechanism of a slope is formulated based on its material, geometrical and environmental parameters such as the unit weight, the cohesion, the internal friction angle, the slope inclination, slope height and the pore water ratio. The ELM is proposed to evaluate the stability of slopes subjected to potential circular failures by means of prediction of the factor of safety (FS). Substantial slope cases collected worldwide are utilized to illustrate and assess the capability and predictability of the ELM on slope stability analysis. Based on the mean absolute percentage errors and the correlation coefficients between the original and predicted FS values, comparisons are demonstrated between the ELM and the generalized regression neural network (GRNN) as well as the prediction models generated from the genetic algorithms. Moreover, sensitivity analysis of the slope parameters and the ELM model parameters is carried out based on the two utilized evaluation functions. The time expense of the ELM on slope stability analysis is also investigated. The results prove that the ELM is advantageous to the GRNN and the genetic algorithm based models in the analysis of slope stability. Hence, the ELM can be a promising technique for approaching the problems in geotechnical engineering.     

A Comparative Analysis of Cognitive Systems for the Prediction of Drillability of Rocks and Wear Factor

This paper demonstrates the applicability of cognitive systems or neural networks in predicting the drillibality of rocks and wear factor using engineering properties of rocks. Drillability of rocks is a useful guide for evaluating the suitability of drills for different ground operations. The wear factor of different materials subsequently helps in the selection of proper drills for different drilling operations. Different rocks were tested for Protodyakonov index, impact strength index, shore hardness number, Schmidt hammer number, drillability and micro bit chisels for wear factor. The data obtained from the tests were used to train and test the neural network. Results from the analysis demonstrate that cognitive systems are an effective tool in the prediction and suitability of drilling operations. Application of these predictive models can be a useful tool to obtain the value of these important parameters, they can save time and help to avoid the tedious process of instrumentation.