Search USGS⌕ Search
Home/Archive/USGS

USGS · 70016924

Sampling design for spatially distributed hydrogeologic and environmental processes

G. Christakos·Ricardo A. Olea
View original source ↗Export metadata ↓

Abstract

A methodology for the design of sampling networks over space is proposed. The methodology is based on spatial random field representations of nonhomogeneous natural processes, and on optimal spatial estimation techniques. One of the most important results of random field theory for physical sciences is its rationalization of correlations in spatial variability of natural processes. This correlation is extremely important both for interpreting spatially distributed observations and for predictive performance. The extent of site sampling and the types of data to be collected will depend on the relationship of subsurface variability to predictive uncertainty. While hypothesis formulation and initial identification of spatial variability characteristics are based on scientific understanding (such as knowledge of the physics of the underlying phenomena, geological interpretations, intuition and experience), the support offered by field data is statistically modelled. This model is not limited by the geometric nature of sampling and covers a wide range in subsurface uncertainties. A factorization scheme of the sampling error variance is derived, which possesses certain atttactive properties allowing significant savings in computations. By means of this scheme, a practical sampling design procedure providing suitable indices of the sampling error variance is established. These indices can be used by way of multiobjective decision criteria to obtain the best sampling strategy. Neither the actual implementation of the in-situ sampling nor the solution of the large spatial estimation systems of equations are necessary. The required values of the accuracy parameters involved in the network design are derived using reference charts (readily available for various combinations of data configurations and spatial variability parameters) and certain simple yet accurate analytical formulas. Insight is gained by applying the proposed sampling procedure to realistic examples related to sampling problems in two dimensions.

Explore related subjects

Advances in Water ResourcesArticle

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

G. Christakos, Ricardo A. Olea. 1992. Sampling design for spatially distributed hydrogeologic and environmental processes. https://doi.org/10.1016/0309-1708(92)90008-p

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related USGS reports

USGS2025

Karhunen–Loève deep learning method for surrogate modeling and approximate Bayesian parameter estimation

We evaluate the performance of the Karhunen–Loève Deep Neural Network (KL-DNN) framework for surrogate modeling and approximate Bayesian parameter estimation in partial differential equation models. In the surrogate model, the Karhunen–Loève (KL) expansions are used for the dimensionality reduction of the number of unknown parameters and variables, and a deep neural network is employed to relate the reduced space of parameters to that of the state variables. The KL-DNN surrogate model is used to formulate a maximum-a-posteriori-like least-squares problem, which is randomized to draw samples of the posterior distribution of the parameters. We test the proposed framework for a hypothetical unconfined aquifer via comparison with the forward MODFLOW and inverse PEST++ iterative ensemble smoother (IES) solutions as well as the state-of-the-art Fourier neural operator (FNO) and deep operator networks (DeepONets) operator learning surrogate models. Our results show that the KL-DNN surrogate model outperforms FNO and DeepONet for forward predictions. For solving inverse problems, the randomized algorithm provides the same or more accurate Bayesian predictions of the parameters than IES as evidenced by the higher log predictive probability of both the estimated parameter field and the forecast hydraulic head. The posterior mean obtained from the randomized algorithm is closer to the reference parameter field than that obtained with FNO as the maximum a posteriori estimate.

Advances in Water Resources↗
USGS2025

The acoustic-Doppler current profiler (ADCP): A comprehensive tool for river-reach hydromorphodynamics

This paper introduces the use of acoustic Doppler current profiler (ADCP) measurements as input for the Acoustic Mapping Velocimetry (AMV) method, a technique for characterizing the dynamics of riverine bedforms. The performance of this new approach, ADCP-AMV, is compared with input from a multibeam echosounder through a field study conducted on the Mississippi River (USA). A virtual ADCP tool has been created to support the ADCP-AMV measurements with optimal data density predictions. To the authors’ knowledge, this is the first time ADCP measurements have been used in conjunction with the AMV dune-tracking method. Subsequently, the paper discusses the coupling of ADCP-AMV measurements with ancillary data extracted from the ADCP. These ancillary data are processed using previously developed protocols to characterize hydrodynamics and the suspended sediment distribution in the water column. This paper emphasizes the capability of ADCPs to characterize open-channel river hydromorphodynamic parameters with high spatiotemporal resolution. Recommendations to accurately and efficiently acquire these multi-variable measurements and derived datasets are discussed.

Tennessee↗
USGS2024

Precipitation uncertainty estimation and rainfall-runoff model calibration using iterative ensemble smoothers

The introduction of iterative ensemble smoothers (IES) for parameter calibration opens avenues for expanding parameter space in surface water hydrologic modeling. Here, we have introduced independent parameters into a model calibration experiment to estimate errors in rainfall forcing data. This approach has the potential to estimate rainfall errors using other hydrological observations and to improve model calibration. Using high-resolution rain gauge data, we estimated “real” rainfall errors across the Turkey River watershed at storm and daily scales. Tests on synthetic and real-world scenarios successfully estimated errors correlated with observed values – even at daily scales. However, a bias remained from model parameter compensation, and identifying errors was challenging for low precipitation and snowfall. Despite synthetic results showing good error correlation, the biases in parameter identification masked potential improvements in hydrological calibration. This study highlights the potential of IES to provide additional information on rainfall errors, even only using streamflow observations.

Advances in Water Resources↗