Sensitivity analysis of 2D steady-state shallow water flow. Application to free surface flow model calibration

This paper presents the analytical properties of the solutions of the sensitivity equations for steady-state, two-dimensional shallow water flow. These analytical properties are used to provide guidelines for model calibration and validation. The sensitivity of the water depth/level and that of the longitudinal unit discharge are shown to contain redundant information. Under subcritical conditions, the sensitivities of the flow variables are shown to obey an anisotropic elliptic equation. The main directions of the contour lines for water depth and the longitudinal unit discharge sensitivity are parallel and perpendicular to the flow, while they are diagonal to the flow for the transverse unit discharge sensitivity. Moreover, the sensitivity for all three variables extends farther in the transverse direction than in the longitudinal direction, the anisotropy ratio being a function of the sole Froude number. For supercritical flow, the sensitivity obeys an anisotropic hyperbolic equation. These findings are confirmed by application examples on idealized and real-world simulations. The sensitivities to the geometry, friction coefficient or model boundary conditions are shown to behave in different ways, thus providing different types of information for model calibration and validation.     

Feed forward neural network and interpolation function models to predict the soil and subsurface sediments distribution in Bam,Iran

An application of the artificial neural network (ANN) approach for predicting mean grain size using electric resistivity data from Bam city is presented. A feed forward back propagation network was developed employing 45 sets of input data. The input variables in the ANN model are the electrical resistivity, water table as a Boolean value and depth; the output is the mean grain size. To demonstrate the authenticity of this approach, the network predictions are compared with those from interpolation methods and the same data. This comparison shows that the ANN approach performs better results. The predicted and observed mean grain size values were compared and show high correlation coefficients. The ANN approach maps show a high degree of correlation with well data based grain size maps and can therefore be used conservatively to better understand the influence of input parameters on sedimentological predictions.     

Constraining groundwater modeling with magnetic resonance soundings

Magnetic resonance sounding (MRS) is a noninvasive geophysical method that allows estimating the free water content and transmissivity of aquifers. In this article, the ability of MRS to improve the reliability of a numerical groundwater model is assessed. Thirty-five sites were investigated by MRS over a ～5000 km(2) domain of the sedimentary Continental Terminal aquifer in SW Niger. Time domain electromagnetic soundings were jointly carried out to estimate the aquifer thickness. A groundwater model was previously built for this section of the aquifer and forced by the outputs from a distributed surface hydrology model, to simulate the observed long-term (1992 to 2003) rise in the water table. Uncertainty analysis had shown that independent estimates of the free water content and transmissivity values of the aquifer would facilitate cross-evaluation of the surface-water and groundwater models. MRS results indicate ranges for permeability (K = 1 × 10(-5) to 3 × 10(-4) m/s) and for free water content (w = 5% to 23% m(3) /m(3) ) narrowed by two orders of magnitude (K) and by ～50% (w), respectively, compared to the ranges of permeability and specific yield values previously considered. These shorter parameter ranges result in a reduction in the model's equifinality (whereby multiple combinations of model's parameters are able to represent the same observed piezometric levels), allowing a better constrained estimate to be derived for net aquifer recharge (～22 mm/year).     

Scapolite stoichiometry

Over the range Ca/(Ca + Na) from zero to 0.75, scapolites vary linearly between Na₄Al₃Si₉O₂₄Cl and NaCa₃Al₅Si₇O₂₄CO₃, by means of a coupled replacement of Na₃Si₂Cl by Ca₃Al₂CO₃. Ca includes minor Sr, Na includes K, and CO₃ may include substantial SO₄. From Ca/(Ca + Na) = 0.75 to 1.00, variation is between NaCa₃Al₅Si₇O₂₄CO₃ and Ca₄Al₆Si₆O₂₄CO₃, by means of a coupled replacement of NaSi by CaAl, as in plagioclase. Anion substitution is complete at the 0.75 point. Recalculation of scapolite analyses on the basis of Al + Si atoms = 12.0, as suggested by computed unit cell contents, offers a reasonable basis for normalization of atomic proportions. On this basis, Ca + Sr + Na + K + Fe atoms total very close to 4.0. New limits on the minor element content of scapolite are proposed. In view of the inflection at 0.75, subdivision of the scapolites at Ca/(Ca + Na) ratios of 0.25, 0.5, and 0.75 seems more logical than the 0.2, 0.5, and 0.8 limits used hitherto for marialite, dipyre, mizzonite, and meionite.     

Predicting road system speeds using spatial structure variables and network characteristics

Spatial regression is applied to GPS floating car measurements to build a predictive model of road system speed as a function of link type, time period, and spatial structure. The models correct for correlated spatial errors and autocorrelation of speeds. Correlation neighborhoods are based on either Euclidean or network distance. Econometric and statistical methods are used to choose the best model form and statistical neighborhood. Models of different types have different coefficient estimates and fit quality, which might affect inferences. Speed predictions are validated against a holdout sample to illustrate the usefulness of spatial regression in road system speed monitoring.      

Spatially distributed long‐term hydrologic simulation using a continuous SCS CN method‐based hybrid hydrologic model

A continuous Soil Conservation Service (SCS) curve number (CN) method that considers time‐varied SCS CN values was developed based on the original SCS CN method with a revised soil moisture accounting approach to estimate run‐off depth for long‐term discontinuous storm events. The method was applied to spatially distributed long‐term hydrologic simulation of rainfall‐run‐off flow with an underlying assumption for its spatial variability using a geographic information systems‐based spatially distributed Clark's unit hydrograph method (Distributed‐Clark; hybrid hydrologic model), which is a simple few parameter run‐off routing method for input of spatiotemporally varied run‐off depth, incorporating conditional unit hydrograph adoption for different run‐off precipitation depth‐based direct run‐off flow convolution. Case studies of spatially distributed long‐term (total of 6 years) hydrologic simulation for four river basins using daily NEXRAD quantitative precipitation estimations demonstrate overall performances of Nash–Sutcliffe efficiency (E_NS) 0.62, coefficient of determination (R²) 0.64, and percent bias 0.33% in direct run‐off and E_NS 0.71, R² 0.72, and percent bias 0.15% in total streamflow for model result comparison against observed streamflow. These results show better fit (improvement in E_NS of 42.0% and R² of 33.3% for total streamflow) than the same model using spatially averaged gauged rainfall. Incorporation of logic for conditional initial abstraction in a continuous SCS CN method, which can accommodate initial run‐off loss amounts based on previous rainfall, slightly enhances model simulation performance; both E_NS and R² increased by 1.4% for total streamflow in a 4‐year calibration period. A continuous SCS CN method‐based hybrid hydrologic model presented in this study is, therefore, potentially significant to improved implementation of long‐term hydrologic applications for spatially distributed rainfall‐run‐off generation and routing, as a relatively simple hydrologic modelling approach for the use of more reliable gridded types of quantitative precipitation estimations.     

Knowledge of the three French underground laboratory sites

From March 1994 to August 1996, ANDRA conducted a large reconnaissance work on three sites for possible underground research laboratories. These surveys highlighted for each site the presence of a geological formation, the properties of which meet the requirements of the French Fundamental Safety Rules. On the East site, the selected formation is a 130 m thick argillite layer. On the Gard site the layer is composed of close-packed siltites which are 200–400 m thick. On the Vienne site a hercynian granodiorite batholith was found under a 160 m thick jurassic overdurden.