Optimization of the generalization capability for rainfall–runoff modeling by neural networks: the case of the <Emphasis Type="Italic">Lez</Emphasis> aquifer (southern France)     

Line Kong A Siou  Anne Johannet  Borrell Estupina Valérie  Séverin Pistre 《Environmental Earth Sciences》2012,65(8):2365-2375

Neural networks are increasingly used in the field of hydrology due to their properties of parsimony and universal approximation with regard to nonlinear systems. Nevertheless, as a result of the existence of noise and approximations in hydrological data, which are very significant in some cases, such systems are particularly sensitive to increased model complexity. This dilemma is known in machine learning as bias–variance and can be avoided by suitable regularization methods. Following a presentation of the bias–variance dilemma along with regularization methods such as cross-validation, early stopping and weight decay, an application is provided for simulating and forecasting karst aquifer outflows at the Lez site. The efficiency of this regularization process is thus demonstrated on a nonlinear, partially unknown basin. As a last step, results are presented over the most intense rainfall event found in the database, which allows assessing the capability of neural networks to generalize with rare or extreme events.  相似文献   

Class I Methanol Emission Around DR 21 (OH)     

S. V. Polushkin  I. E. Val’tts 《Astronomy Reports》2010,54(6):496-508

Results of new observations of the vicinity of DR 21 (OH) conducted on the 20-m Onsala radio telescope are presented. The goal was to search for associations between molecular hydrogen emission tracing shock waves and class I methanol maser emission. Observations at 44 and 36 GHz have shown that an extensive region of faint methanol maser emission elongated North-South is probably present in the vicinity of DR 21(OH). The linear size of this structure may be a factor of ten larger than the central region in DR 21(OH) that emits at 44 GHz. Three maser emission peaks are clearly visible in the northern (DR 21N), central (DR 21(OH)), and southern (vicinity of DR 21 West) parts of this structure. Many other structures are also embedded in this region, including the protostellar disk ERO 3 previously detected at 6.7 GHz. Maser components of these objects are formed with velocities from -5 to-2 km/s, with a velocity gradient from -5 in the North to -2 km/s in the South. The spatial resolution of the map is not high enough to distinguish fine structures at 44 GHz associated with spots and jets emitting in molecular hydrogen.  相似文献   

A Physically-Based Scheme For The Urban Energy Budget In Atmospheric Models   总被引：28，自引：1，他引：27  

Valéry Masson 《Boundary-Layer Meteorology》2000,94(3):357-397

An urban surface scheme for atmospheric mesoscale models ispresented. A generalization of local canyon geometry isdefined instead of the usual bare soil formulation currently usedto represent cities in atmospheric models. This allows refinement ofthe radiative budgets as well as momentum, turbulent heat and ground fluxes.The scheme is aimed to be as general as possible, in order to representany city in the world, for any time or weather condition(heat island cooling by night, urban wake, water evaporation after rainfalland snow effects).Two main parts of the scheme are validated against published data.Firstly, it is shown that the evolution of the model-predictedfluxes during a night with calm winds is satisfactory, considering both the longwave budget and the surface temperatures. Secondly, the original shortwave scheme is tested off-line and compared to the effective albedoof a canyon scale model. These two validations show that the radiative energy input to the urban surface model is realistic.Sensitivity tests of the model are performed for one-yearsimulation periods, for both oceanic and continental climates. The scheme has the ability to retrieve, without ad hoc assumptions, the diurnal hysteresis between the turbulent heat flux and ground heat flux. It reproduces the damping of the daytime turbulent heat flux by the heat storage flux observed in city centres. The latent heat flux is negligible on average,but can be large when short time scales are considered (especially afterrainfall). It also suggests that in densely built areas, domesticheating can overwhelm the net radiation, and supply a continuous turbulentheat flux towards the atmosphere. This becomes very important inwinter for continental climates. Finally, a comparison with a vegetation scheme shows that the suburban environment can be represented with a bare soil formulation for large temporal or spatial averages (typical of globalclimatic studies), but that a surface scheme dedicated to the urban surface is necessary when smaller scales are considered: town meteorological forecasts, mesoscale or local studies.  相似文献   

Model-based optimization of source locations for 3D acoustic seismic full-waveform inversion     

Valérie Winner  Pascal Edme  Hansruedi Maurer 《Geophysical Prospecting》2023,71(1):3-16

Besides classical imaging techniques, full-waveform inversion is an increasingly popular method to derive elastic subsurface properties from seismic data. High-resolution velocity models can be obtained, and spatial sampling criteria are less strict than for imaging methods, because the entire information content of the seismic waveforms is used. As high operational costs arise from seismic surveys, the acquirable data volume is often limited by economic criteria. By selecting optimal locations for seismic sources, the information content of the data can be maximized, and the number of sources and thus the acquisition costs can be reduced compared with standard acquisition designs. The computation of such optimized designs for large-size 3D inverse problems at affordable computational cost is challenging. By using a sequential receiver-wise optimization strategy, we substantially reduce the computational requirements of the optimization process. We prove the applicability of this method by means of numerical 3D acoustic examples. Optimized source designs for different receiver patterns are computed for a realistic subsurface model, and the value of the designs is evaluated by comparing checkerboard inversion tests with different acquisition designs. Our examples show that inversion results with higher accuracy can be obtained with the optimized designs, regardless of the number of sources, the number of receivers, or the receiver distribution. Larger benefits of the optimized designs are visible when a sparse receiver geometry is used.  相似文献   

Acoustic Ray Tracing Comparisons in the Context of Geodetic Precise off-shore Positioning Experiments     

Pierre Sakic  Valérie Ballu  Wayne C. Crawford  Guy Wöppelmann 《Marine Geodesy》2018,41(4):315-330

The GNSS-Acoustics (GNSS-A) method couples acoustics with GNSS to allow the precise localization of a seafloor reference in a global frame. This method can extend on-shore GNSS networks and allows the monitoring of hazardous oceanic tectonic phenomena. The goal of this study is to test the influence of both acoustics ray tracing techniques and spatial heterogeneities of acoustic wave speed on positioning accuracy. We test three different ray tracing methods: the eikonal method (3D sound speed field), the Snell-Descartes method (2D sound speed profile), and an equivalent sound speed method. We also compare the processing execution time. The eikonal method is compatible with the Snell-Descartes method (by up to 10 ppm in term of propagation time difference) but takes approximately a thousand times longer to run. We used the 3D eikonal ray tracing to characterize the influence of a lateral sound speed gradient on acoustic ray propagation and positioning accuracy. For a deep water (? 3,000 m) situation, frequent in subduction zones such as the Lesser Antilles, not accounting for lateral sound speed gradients can induce an error of up to 5 cm in the horizontal positioning of a seafloor transponder, even when the GNSS-A measurements are made over the barycenter of a seafloor transponder array.  相似文献   

A kinematic GPS methodology for sea surface mapping,Vanuatu   总被引：1，自引：0，他引：1  

Marie-Noelle Bouin  Valérie Ballu  Stéphane Calmant  Jean-Michel Boré  Eric Folcher  Jérôme Ammann 《Journal of Geodesy》2009,83(12):1203-1217

During the past few decades, satellite altimetry has brought tremendous new knowledge about the spatial and temporal variations of sea surface heights over the Earth’s oceans. However, the precision is limited over short wavelengths and in coastal areas, and other methods such as kinematic GPS may be needed to fill in this information. We present kinematic GPS work aimed at mapping the sea surface height, with special attention to the precision one can expect. Active marine subduction zones, like the Vanuatu archipelago, may present short wavelength, high amplitude undulations of the sea surface height that are difficult to map with satellite altimetry. This paper presents the methodology used around Santo Island, in Vanuatu, to obtain a well-resolved local sea surface map with a precision of 5–15 cm limited by the sea conditions and the distance from the coastal reference station. We present the results of three campaigns in 2004, 2006 and 2007. Careful observation of the ship behaviour along the surveys as well as simultaneous recording of the ship attitude variations is mandatory to obtain reliable results. We show that the ship GPS antenna height varies with the ship’s velocity and we suggest a method to correct this effect. The final precision is estimated using the crossover differences method.  相似文献