Robust Data Worth Analysis with Surrogate Models

Highly detailed physically based groundwater models are often applied to make predictions of system states under unknown forcing. The required analysis of uncertainty is often unfeasible due to the high computational demand. We combine two possible solution strategies: (1) the use of faster surrogate models; and (2) a robust data worth analysis combining quick first-order second-moment uncertainty quantification with null-space Monte Carlo techniques to account for parametric uncertainty. A structurally and parametrically simplified model and a proper orthogonal decomposition (POD) surrogate are investigated. Data worth estimations by both surrogates are compared against estimates by a complex MODFLOW benchmark model of an aquifer in New Zealand. Data worth is defined as the change in post-calibration predictive uncertainty of groundwater head, river-groundwater exchange flux, and drain flux data, compared to the calibrated model. It incorporates existing observations, potential new measurements of system states (“additional” data) as well as knowledge of model parameters (“parametric” data). The data worth analysis is extended to account for non-uniqueness of model parameters by null-space Monte Carlo sampling. Data worth estimates of the surrogates and the benchmark suggest good agreement for both surrogates in estimating worth of existing data. The structural simplification surrogate only partially reproduces the worth of “additional” data and is unable to estimate “parametric” data, while the POD model is in agreement with the complex benchmark for both “additional” and “parametric” data. The variance of the POD data worth estimates suggests the need to account for parameter non-uniqueness, like presented here, for robust results.     

Probabilistic seismic hazard assessment of the Pyrenean region

A unified probabilistic seismic hazard assessment (PSHA) for the Pyrenean region has been performed by an international team composed of experts from Spain and France during the Interreg IIIA ISARD project. It is motivated by incoherencies between the seismic hazard zonations of the design codes of France and Spain and by the need for input data to be used to define earthquake scenarios. A great effort was invested in the homogenisation of the input data. All existing seismic data are collected in a database and lead to a unified catalogue using a local magnitude scale. PSHA has been performed using logic trees combined with Monte Carlo simulations to account for both epistemic and aleatory uncertainties. As an alternative to hazard calculation based on seismic sources zone models, a zoneless method is also used to produce a hazard map less dependant on zone boundaries. Two seismogenic source models were defined to take into account the different interpretations existing among specialists. A new regional ground-motion prediction equation based on regional data has been proposed. It was used in combination with published ground-motion prediction equations derived using European and Mediterranean data. The application of this methodology leads to the definition of seismic hazard maps for 475- and 1,975-year return periods for spectral accelerations at periods of 0 (corresponding to peak ground acceleration), 0.1, 0.3, 0.6, 1 and 2 s. Median and percentiles 15% and 85% acceleration contour lines are represented. Finally, the seismic catalogue is used to produce a map of the maximum acceleration expected for comparison with the probabilistic hazard maps. The hazard maps are produced using a grid of 0.1°. The results obtained may be useful for civil protection and risk prevention purposes in France, Spain and Andorra.     

碳酸盐岩孔隙结构参数构建与储层参数反演(英文)

碳酸盐岩储层孔隙结构相对碎屑岩更复杂,常用的岩石物理模型不能较好的描述其孔隙结构的变化规律,且岩石孔隙结构的差异较大程度上会影响岩石的弹性性质。本文首先利用岩石薄片分析了碳酸盐岩的微观孔隙结构。然后基于Gassmann方程和Eshelby-Walsh椭球包体裂缝理论,在合理的假设前提下给出了一种新的岩石物理建模方法,并且从中提取了一个参数来表征孔隙结构的变化规律。最后,基于全波列测井数据,我们利用该方法计算了单井的孔隙度,并与用常规方法预测的结果进行了比较,同时进行了地震储层参数反演。研究结果表明,孔隙结构对岩石的弹性性质的影响较大,且新的建模方法预测的孔隙度误差仅为0.74%。因此,该方法可有效的减小孔隙结构对计算各岩石弹性参数的影响并提高孔隙度的预测精度。     

On the modified groundwater flow equation: analytical solution via iteration method

To take into account the variability of the medium through which the groundwater flow takes place, we presented the groundwater flow equation within a confined aquifer with prolate coordinates. The new equation is a perturbed singular equation. The perturbed parameters is introduced and can be used as accurately replicate the variability of the aquifer from one point to another. When the perturbed parameter tends to zero, we recover the Theis equation. We solved analytically and iteratively the new equation. We compared the obtained solution with experimental observed data together with existing solutions. The comparison shows that the modified equation predicts more accurately the physical problem than the existing model. Copyright © 2015 John Wiley & Sons, Ltd.     

Analysis and modeling of the seasonal South China Sea temperature cycle using remote sensing

The present paper describes the analysis and modeling of the South China Sea (SCS) temperature cycle on a seasonal scale. It investigates the possibility to model this cycle in a consistent way while not taking into account tidal forcing and associated tidal mixing and exchange. This is motivated by the possibility to significantly increase the model’s computational efficiency when neglecting tides. The goal is to develop a flexible and efficient tool for seasonal scenario analysis and to generate transport boundary forcing for local models. Given the significant spatial extent of the SCS basin and the focus on seasonal time scales, synoptic remote sensing is an ideal tool in this analysis. Remote sensing is used to assess the seasonal temperature cycle to identify the relevant driving forces and is a valuable source of input data for modeling. Model simulations are performed using a three-dimensional baroclinic-reduced depth model, driven by monthly mean sea surface anomaly boundary forcing, monthly mean lateral temperature, and salinity forcing obtained from the World Ocean Atlas 2001 climatology, six hourly meteorological forcing from the European Center for Medium range Weather Forecasting ERA-40 dataset, and remotely sensed sea surface temperature (SST) data. A sensitivity analysis of model forcing and coefficients is performed. The model results are quantitatively assessed against climatological temperature profiles using a goodness-of-fit norm. In the deep regions, the model results are in good agreement with this validation data. In the shallow regions, discrepancies are found. To improve the agreement there, we apply a SST nudging method at the free water surface. This considerably improves the model’s vertical temperature representation in the shallow regions. Based on the model validation against climatological in situ and SST data, we conclude that the seasonal temperature cycle for the deep SCS basin can be represented to a good degree. For shallow regions, the absence of tidal mixing and exchange has a clear impact on the model’s temperature representation. This effect on the large-scale temperature cycle can be compensated to a good degree by SST nudging for diagnostic applications.     

NUMERICAL SIMULATION OF INUNDATION IN SHENZHEN CITY BY FINTE DIFFERENCE MODEL MIXED WITH 1D AND 2D UNSTEADY FLOW

ＮＵＭＥＲＩＣＡＬＳＩＭＵＬＡＴＩＯＮＯＦＩＮＵＮＤＡＴＩＯＮＩＮＳＨＥＮＺＨＥＮＣＩＴＹＢＹＦＩＮＴＥＤＩＦＦＥＲＥＮＣＥＭＯＤＥＬＭＩＸＥＤＷＩＴＨ１ＤＡＮＤ２ＤＵＮＳＴＥＡＤＹＦＬＯＷ￥ＣＨＥＮＧＸｉａｏｔａｏ；ａｎｄＱＩＵＪｉｎｗｅｉ（Ｓｅｎ...     

A two-layer model for studying 2D dissolved pollutant runoff over impermeable surfaces

Dissolved pollutants in stormwater are a main contributor to water pollution in urban environments. However, many existing transport models are semi-empirical and only consider one-dimensional flows, which limit their predictive capacity. Combining the shallow water and the advection–diffusion equations, a two-dimensional physically based model is developed for dissolved pollutant transport by adopting the concept of a ‘control layer’. A series of laboratory experiments has been conducted to validate the proposed model, taking into account the effects of buildings and intermittent rainfalls. The predictions are found to be in good agreement with experimental observations, which supports the assumption that the depth of the control layer is constant. Based on the validated model, a parametric study is conducted, focusing on the characteristics of the pollutant distribution and transport rate over the depth. The hyetograph, including the intensity, duration and intermittency, of rainfall event has a significant influence on the pollutant transport rates. The depth of the control layer, rainfall intensity, surface roughness and area length are dominant factors that affect the dissolved pollutant transport. Finally, several perspectives of the new pollutant transport model are discussed. This study contributes to an in-depth understanding of the dissolved pollutant transport processes on impermeable surfaces and urban stormwater management.     

Hindered settling with an apparent particle diameter concept

In this work a new expression has been developed to predict the settling velocity of a sediment particle which is dispersed in a sediment-fluid mixture during a turbulent flow. A concept of apparent particle diameter has been introduced and is defined by the diameter of the spherical volume in which the particle can move randomly after collision with other particles in suspension. The effect of suspension concentration is studied on the mass density of the sediment-fluid mixture. It has been shown that the settling velocity of sediment particle in a sediment-fluid mixture is a function of different characteristics of the sediment particle such as settling velocity in clear fluid, suspension concentration, relative mass density and Reynolds number. The model has shown good agreement when compared with previously published experimental data and it’s prediction accuracy is superior than the other existing models.     

V S Profiles from Noise Cross Correlation at Local and Small Scale

Ambient noise measurements have been performed at local and small scales in the Neapolitan and surrounding areas (Campania, southern Italy) by employing two broad-band Kinemetrics Q330 stations, equipped with Episensor ES-T three component accelerometers. In both experiments frequency time analysis (FTAN method) has been performed on the vertical and radial components of noise cross correlations to retrieve the Rayleigh wave dispersion (Green??s function). At local scale, over an interstation distance of about 26?km, the group velocity dispersion values have been compared with those obtained from FTAN analysis on recordings of two earthquakes with similar path. At small scale, measurements have been carried out over an interstation distance of about 440?m in the public gardens of Scampia, the northernmost quarter of Naples. The Rayleigh wave group velocity dispersion data obtained from noise cross correlation, have been combined with those from active seismic experiment along the same alignment, but shorter (120?m offset). The non linear inversion of such a dispersion curve has allowed the definition of V _S models to depths of 100?m, in agreement with nearby stratigraphy. Moreover, a good agreement has resulted for the resonance frequency among the H/V ratio, the ellipticity of the fundamental mode computed for the chosen V _S model, and the average two-dimensional (2D) spectral amplification computed along a cross section representative of the Scampia quarter.     

Approaches to modelling coupled flow and reaction in a 2D cementation experiment

Porosity evolution at reactive interfaces is a key process that governs the evolution and performances of many engineered systems that have important applications in earth and environmental sciences. This is the case, for example, at the interface between cement structures and clays in deep geological nuclear waste disposals. Although in a different transport regime, similar questions arise for permeable reactive barriers used for biogeochemical remediation in surface environments.The COMEDIE project aims at investigating the coupling between transport, hydrodynamics and chemistry when significant variations of porosity occur. The present work focuses on a numerical benchmark used as a design exercise for the future COMEDIE-2D experiment. The use of reactive transport simulation tools like Hytec and Crunch provides predictions of the physico-chemical evolutions that are expected during the future experiments in laboratory. Focus is given in this paper on the evolution during the simulated experiment of precipitate, permeability and porosity fields.A first case is considered in which the porosity is constant. Results obtained with Crunch and Hytec are in relatively good agreement. Differences are attributable to the models of reactive surface area taken into account for dissolution/precipitation processes. Crunch and Hytec simulations taking into account porosity variations are then presented and compared. Results given by the two codes are in qualitative agreement, with differences attributable in part to the models of reactive surface area for dissolution/precipitation processes. As a consequence, the localization of secondary precipitates predicted by Crunch leads to lower local porosities than for predictions obtained by Hytec and thus to a stronger coupling between flow and chemistry. This benchmark highlights the importance of the surface area model employed to describe systems in which strong porosity variations occur as a result of dissolution/precipitation. The simulation of highly non-linear reactive transport systems is also shown to be partly dependent on specific numerical approaches.     

Penetration of electric field from the near ground atmospheric layer to the ionosphere

A mathematical model has been proposed for describing quasi-stationary atmospheric electric fields with approximate, but fairly accurate allowance for ionospheric conductivity. It is shown that some well-known models of electric field penetration from the Earth into the ionosphere have been deemed inadequate, though they work well in the atmosphere below 50 km. In these models, the arbitrarily specified boundary condition in the upper boundary of the atmosphere omits the existing good conductor or adds not existent conductor. The maximum possible field in our model is far less than in models where ionospheric conductivity is not taken into account, but vastly larger than in models based on the approximation with infinite Pedersen conductivity in the upper ionosphere.     

Estimating the atmospheric ozone transmission for solar radiation models

Accurate calculation of the solar radiation at the earth’s surface requires evaluation of the atmospheric ozone transmission. To use the existing methods one needs to know the atmospheric ozone amount. The serious lack of ozone data makes almost impossible the application of these methods in most of the world. In this paper a new method of estimating the overall ozone attenuation in the Northern Hemisphere is presented. Multiyear monthly latitudinal average ozone data are used. A daily ozone transmission function is proposed, and values of it are calculated for each latitudinal circle from 0° to 70°N at 10 longitude-degree intervals. By regression techniques parameterized expressions of the daily ozone transmission in the Northern Hemisphere are obtained as a function of the latitude. The model is applied to Athens, and the results are in excellent agreement with those obtained by extremely detailed calculations of the ozone transmission. The reliability of the model, under special regional ozone conditions, is investigated. Calculations for 67 locations in the Northern Hemisphere prove that the deviations do not exceed 1 percent. The influence of the year-by-year ozone variability is also examined, and the maximum deviation is found to be near 1.6 percent. The proposed model can be easily incoporated into solar radiation models in order to provide the ozone depletion at any place in the Northern Hemisphere where atmospheric ozone data are not available.     

A semi-implicit three-dimensional numerical model for non-hydrostatic pressure free-surface flows on an unstructured,sigma grid

A semi-implicit 3-D numerical formulation for solving non-hydrostatic pressure free-surface flows on an unstructured,sigma grid is proposed.Pressure-splitting and 9 semi-implicit methods are inherited and reformed from Casulli’s z-coordinate formulation.The non-orthogonal sigma-coordinate transformation leads to additional terms. The resulting linear system for the non-hydrostatic correction is diagonally dominant but unsymmetric,and it is solved by the BiCGstab method.In contrast with z-coordinate non-hydrostatic models,the new model fits vertical boundaries much better,which is important for the long-time simulation of sediment transport and riverbed deformation.A lock-exchange density flow is computed to determine whether the new scheme is able to simulate non-hydrostatic free-surface flows.The new model is further verified using the field data of a natural river bend of the lower Yangtze River.Good agreement between simulations and earlier research results,field data is obtained, indicating that the new model is applicable to hydraulic projects in real rivers.     

Collapse of a nonductile concrete frame: Evaluation of analytical models

The current paper presents nonlinear dynamic analyses that simulate shaking table tests performed on a four‐column reinforced concrete frame described in a companion paper. The frame consists of two ductile and two nonductile columns interconnected by a stiff beam. In order to validate existing analytical models for nonductile concrete columns, a blind comparison of the test data and results of the analysis is performed. The analysis adequately captures the drift response and correctly detects collapse of the structure; however, strength degradation due to cover spalling is exaggerated in the analytical model. Refinement of the analysis by changing the concrete cover model results in an excellent agreement between the test data and analysis results at the initiation of shear failure and collapse of the frame. The experimental data are further compared with lumped‐plasticity nonlinear models used in engineering practice. The results suggest that the sudden strength degradation used in ASCE/SEI 41‐06 results in an exaggerated estimate of the displacement demands. It is also observed that ignoring the strength degradation, using an elastic‐perfectly‐plastic model, provides a good estimate of the displacement demands when strength degradation is not severe. Copyright © 2008 John Wiley & Sons, Ltd.     

Modeling the water resources of the Black and Mediterranean Sea river basins and their impact on regional mass changes

For the first time, a dedicated release of the hydrology and water use model WaterGAP3, has been developed to spatially explicit calculate hydrological fluxes within river basins draining into the Mediterranean and Black Sea. The main differences between the new regional version of the global WaterGAP3 model and the previously applied global version WaterGAP2 can be found in the spatial resolution, snow modeling, and water use modeling. Comparison with observations shows that WaterGAP3 features a more realistic representation of modeled river runoff and inflow into both seas. WaterGAP3 generates more inflow to both seas than WaterGAP2. In the WaterGAP3 simulation, contributions to the total runoff into the Black Sea from individual discharge regions show in general a good agreement to climatology derived runoff, but lesser importance of Georgian rivers for the basin's water. After the successful model validation WaterGAP3 has been applied to correct estimates of seawater mass derived from the GRACE gravity mission and to account for freshwater inflow into both basins. The performance of the WaterGAP3 regional solution has been evaluated by comparing the seawater mass derived from GRACE corrected for the leakage of continental hydrology, to an independent estimate derived from steric-corrected satellite altimetry with steric correction from regional oceanographic models. The agreement is higher in the Mediterranean Sea than in the Black Sea. Results using WaterGAP3 and WaterGAP2 are not significantly different. However the agreement with the altimetry-derived results is higher using WaterGAP2, due to the smaller annual amplitude of the continental hydrology leakage from WaterGAP3. We conclude that the regional model WaterGAP3 is capable of realistically quantifying water mass variation in the region, further developments have been identified.     

Experimental study of near-bed concentration and sediment vertical mixing parameter for vertical concentration distribution in the surf zone

Two formulae for the near-bed concentration(C_a) and the sediment vertical mixing parameter(m) are established based on a large scale wave flume experiment.The advantage of the new formulae is that the turbulent kinetic energy induced by wave breaking can be taken into account;the formula for C_a is in terms of the near-bed,time-averaged turbulent kinetic energy,and the formula for m is in terms of depth-and time-averaged turbulent kinetic energy.A new expression for suspended sediment load also is established by depth integration of the vertical distribution of the suspended sediment concentration obtained on basis of the new formulae.Equation validation is done by comparing the predicted C_a and m to measurements for different types of waves(regular wave,wave group,and irregular wave),and good agreement is found.The advantages of the proposed formulae over previous formulae also are discussed.     

Comparison of eigensensitivity and ANN based methods in model updating of an eight-story building

Analytical models prepared from field drawings do not generally provide results that match with experimental results.The error may be due to uncertainties in the property of materials,size of members and errors in the modelling process.It is important to improve analytical models using experimentally obtained data.For the past several years,data obtained from ambient vibration testing have been successfully used in many cases to update and match dynamic behaviors of analytical models with real structures.This paper presents a comparison between artificial neural network(ANN) and eigensensitivity based model updating of an existing multi-story building.A simple spring-mass analytical model,developed from the structural drawings of the building,is considered and the corresponding spring stiffness and lumped mass of all floors are chosen as updating parameters.The advantages and disadvantages of these updating methods are discussed.The advantage is that both methods ensure a physically meaningful model which canbe further employed in determining structural response and health monitoring.     

Nonlinear FE model updating and reconstruction of the response of an instrumented seismic isolated bridge to the 2010 Maule Chile earthquake

Nonlinear finite element (FE) modeling has been widely used to investigate the effects of seismic isolation on the response of bridges to earthquakes. However, most FE models of seismic isolated bridges (SIB) have used seismic isolator models calibrated from component test data, while the prediction accuracy of nonlinear FE models of SIB is rarely addressed by using data recorded from instrumented bridges. In this paper, the accuracy of a state‐of‐the‐art FE model is studied through nonlinear FE model updating (FEMU) of an existing instrumented SIB, the Marga‐Marga Bridge located in Viña del Mar, Chile. The seismic isolator models are updated in 2 phases: component‐wise and system‐wise FEMU. The isolator model parameters obtained from 23 isolator component tests show large scatter, and poor goodness of fit of the FE‐predicted bridge response to the 2010 Mw 8.8 Maule, Chile Earthquake is obtained when most of those parameter sets are used for the isolator elements of the bridge model. In contrast, good agreement is obtained between the FE‐predicted and measured bridge response when the isolator model parameters are calibrated using the bridge response data recorded during the mega‐earthquake. Nonlinear FEMU is conducted by solving single‐ and multiobjective optimization problems using high‐throughput cloud computing. The updated FE model is then used to reconstruct response quantities not recorded during the earthquake, gaining more insight into the effects of seismic isolation on the response of the bridge during the strong earthquake.     

A synthetic Earth gravity model based on a topographic-isostatic model

The Earth??s gravity field is related to the topographic potential in medium and higher degrees, which is isostatically compensated. Hence, the topographic-isostatic (TI) data are indispensable for extending an available Earth Gravitational Model (EGM) to higher degrees. Here we use TI harmonic coefficients to construct a Synthetic Earth Gravitational Model (SEGM) to extend the EGMs to higher degrees. To achieve a high-quality SEGM, a global geopotential model (EGM96) is used to describe the low degrees, whereas the medium and high degrees are obtained from the TI or topographic potential. This study differes from others in that it uses a new gravimetric-isostatic model for determining the TI potential. We test different alternatives based on TI or only topographic data to determine the SEGM. Although the topography is isostatically compensated only to about degree 40?C60, our study shows that using a compensation model improves the SEGM in comparison with using only topographic data for higher degree harmonics. This is because the TI data better adjust the applied Butterworth filter, which bridges the known EGM and the new high-degree potential field than the topographic data alone.     

Spatial modeling of environmental vulnerability of marine finfish aquaculture using GIS-based neuro-fuzzy techniques

Combining GIS with neuro-fuzzy modeling has the advantage that expert scientific knowledge in coastal aquaculture activities can be incorporated into a geospatial model to classify areas particularly vulnerable to pollutants. Data on the physical environment and its suitability for aquaculture in an Irish fjard, which is host to a number of different aquaculture activities, were derived from a three-dimensional hydrodynamic and GIS models. Subsequent incorporation into environmental vulnerability models, based on neuro-fuzzy techniques, highlighted localities particularly vulnerable to aquaculture development.The models produced an overall classification accuracy of 85.71%, with a Kappa coefficient of agreement of 81%, and were sensitive to different input parameters. A statistical comparison between vulnerability scores and nitrogen concentrations in sediment associated with salmon cages showed good correlation.Neuro-fuzzy techniques within GIS modeling classify vulnerability of coastal regions appropriately and have a role in policy decisions for aquaculture site selection.