A variance-ratio test for supporting a variable mean in kriging

This article discusses the issue of whether to use a variable mean and describes a test that can be used to evaluate whether it is justified to add terms to the drift (deterministic part) of a geostatistical model. The basic model could be the intrinsic one, where the deterministic part is a constant, and the alternate model could be any model that includes a constant term in the expression for the drift. Also, differences between constant- and variable-mean models are discussed.     

The shift of ravi river and the geomorphological features along its course in amritsar and gurdaspur districts of punjab

Satellite remote sensing technique can be effectively utilised in mapping and monitoring the river course changes and associated geomorphological features. Ravi river, flowing along the Indo-Pakistan border, has been in the limelight for its repeated flood havoc during monsoon and abrupt encroachment at some places in the Indian territory, where it was not flowing earlier. This river, meandering in zig-zag fashion along the International boundary in Amritsar and Gurdaspur districts of Punjab, poses perennial threats to the nations’s economy due to extensive destruction happening every year. An attempt has been made to map the shift of this river and the associated geomorphological features along its course using the Indian Remote Sensing Satellite data (IRS-IA and IB LISS-IIFCC) of the period 1991–1993 and the Survey of India topographic sheets of the period 1972–1973. The study shows that there has been drastic changes in the course of Ravi during a span of 20 years due to human activities along its course. The river has shifted its course considerably towards India since its topography is against it. River training structures/bundhs, built by the neighbouring country, across and very near to the earlier river course has been the main reason for this drastic shifting. It is estimated that such massive structures could turn the river course towards India by atleast 1 to 5 km in the border districts of Punjab. This shifting of Ravi along international border poses a serious threat to the Nation’s defence system.     

Transmissivity identification through multi-directional aquifer stimulation

This paper presents a geostatistical approach to multi-directional aquifer stimulation in order to better identify the transmissivity field. Hydraulic head measurements, taken at a few locations but under a number of different steady-state flow conditions, are used to estimate the transmissivity. Well installation is generally the most costly aspect of obtaining hydraulic head measurements. Therefore, it is advantageous to obtain as many informative measurements from each sampling location as possible. This can be achieved by hydraulically stimulating the aquifer through pumping, in order to set-up a variety of flow conditions. We illustrate the method by applying it to a synthetic aquifer. The simulations provide evidence that a few sampling locations may provide enough information to estimate the transmissivity field. Furthermore, the innovation of, or new information provided by, each measurement can be examined by looking at the corresponding spline and sensitivity matrix. Estimates from multi-directional stimulation are found to be clearly superior to estimates using data taken under one flow condition. We describe the geostatistical methodology for using data from multi-directional simulations and address computational issues.     

Sequential sampling to contour an uncertain function

Sampling exploration of uncertain functions to locate critical contour levels is most effective if sampling decisions are made sequentially. A simple sequential exploration strategy, based on pseudo-Bayesian second-moment analysis, is proposed and compared with non-sequential systematic sampling. Repeated application to functions simulated pseudorandomly from stationary random processes on the line and on the plane indicates uniform superiority of the sequential strategy. The method is particularly advantageous when the function of interest,h(X), has an uncertain trend, and in general when the random process that quantifies prior uncertainty onh(X) is highly correlated.     

Applicability of the Dual‐Domain Model to Nonaggregated Porous Media

More theoretical analysis is needed to investigate why a dual‐domain model often works better than the classical advection‐dispersion (AD) model in reproducing observed breakthrough curves for relatively homogeneous porous media, which do not contain distinct dual domains. Pore‐scale numerical experiments presented here reveal that hydrodynamics create preferential flow paths that occupy a small part of the domain but where most of the flow takes place. This creates a flow‐dependent configuration, where the total domain consists of a mobile and an immobile domain. Mass transfer limitations may result in nonequilibrium, or significant differences in concentration, between the apparent mobile and immobile zones. When the advection timescale is smaller than the diffusion timescale, the dual‐domain mass transfer (DDMT) model better captures the tailing in the breakthrough curve. Moreover, the model parameters (mobile porosity, mean solute velocity, dispersivity, and mass transfer coefficient) demonstrate nonlinear dependency on mean fluid velocity. The studied case also shows that when the Peclet number, Pe, is large enough, the mobile porosity approaches a constant, and the mass transfer coefficient can be approximated as proportional to mean fluid velocity. Based on detailed analysis at the pore scale, this paper provides a physical explanation why these model parameters vary in certain ways with Pe. In addition, to improve prediction in practical applications, we recommend conducting experiments for parameterization of the DDMT model at a velocity close to that of the relevant field sites, or over a range of velocities that may allow a better parameterization.     

Parameter estimation in nonlinear environmental problems

Popular parameter estimation methods, including least squares, maximum likelihood, and maximum a posteriori (MAP), solve an optimization problem to obtain a central value (or best estimate) followed by an approximate evaluation of the spread (or covariance matrix). A different approach is the Monte Carlo (MC) method, and particularly Markov chain Monte Carlo (MCMC) methods, which allow sampling from the posterior distribution of the parameters. Though available for years, MC methods have only recently drawn wide attention as practical ways for solving challenging high-dimensional parameter estimation problems. They have a broader scope of applications than conventional methods and can be used to derive the full posterior pdf but can be computationally very intensive. This paper compares a number of different methods and presents improvements using as case study a nonlinear DNAPL source dissolution and solute transport model. This depth-integrated semi-analytical model approximates dissolution from the DNAPL source zone using nonlinear empirical equations with partially known parameters. It then calculates the DNAPL plume concentration in the aquifer by solving the advection-dispersion equation with a flux boundary. The comparison is among the classical MAP and some versions of computer-intensive Monte Carlo methods, including the Metropolis–Hastings (MH) method and the adaptive direction sampling (ADS) method.     

Minimum-variance unbiased quadratic estimation of covariances of regionalized variables

The parameters of covariance functions (or variograms) of regionalized variables must be determined before linear unbiased estimation can be applied. This work examines the problem of minimum-variance unbiased quadratic estimation of the parameters of ordinary or generalized covariance functions of regionalized variables. Attention is limited to covariance functions that are linear in the parameters and the normality assumption is invoked when fourth moments of the data need to be calculated. The main contributions of this work are (1) it shows when and in what sense minimum-variance unbiased quadratic estimation can be achieved, and (2) it yields a well-founded, practicable, and easy-to-automate methodology for the estimation of parameters of covariance functions. Results of simulation studies are very encouraging.     

Analysis of macrodispersion through volume-averaging: Moment equations

Macrodispersion is spreading of a substance induced by spatial variations in local advective velocity at field scales. Consider the case that the steady-state seepage velocity and the local dispersion coefficients in a heterogeneous formation may be modeled as periodic in all directions in an unbounded domain. The equations satisfied by the first two spatial moments of the concentration are derived for the case of a conservative non-reacting solute. It is shown that the moments can be calculated from the solution of well-defined deterministic boundary value problems. Then, it is described how the rate of increase of the first two moments can be calculated at large times using a Taylor-Aris analysis as generalized by Brenner. It is demonstrated that the second-order tensor of macrodispersion (or effective dispersion) can be computed through the solution of steady-state boundary-value problems followed by the determination of volume averages. The analysis is based solely on volume averaging and is not limited by the assumption that the fluctuations are small. The large-time results are valid when the system is in a form of equilibrium in which a tagged particle samples all locations in an appropriately defined phase space with equal probability.