Named entity recognition goes to old regime France: geographic text analysis for early modern French corpora

ABSTRACT

Geographic text analysis (GTA) research in the digital humanities has focused on projects analyzing modern English-language corpora. These projects depend on temporally specific lexicons and gazetteers that enable place name identification and georesolution. Scholars working on the early modern period (1400–1800) lack temporally appropriate geoparsers and gazetteers and have been reliant on general purpose linked open data services like Geonames. These anachronistic resources introduce significant information retrieval and ethical challenges for early modernists. Using the geography entries of the canonical eighteenth-century Encyclopédie, we evaluate rule-based named entity recognition (NER) systems to pinpoint areas where they would benefit from adjustments for processing historical corpora. As we demonstrate, annotating nested and extended place information is one way to improve early modern GTA. Working with Enlightenment sources also motivates a critique of the landscape of digital geospatial data.     

Can groundwater secure drinking-water supply and supplementary irrigation in new settlements of North-West Cambodia?

Since the end of the Cambodian Civil War in 1998, the population of the Oddar Meanchey province has drastically increased despite the lack of adequate infrastructure, including basic amenities such as drinking-water supply. To improve the access to drinking water, governmental and aid agencies have focussed on drilling shallow boreholes. The use of groundwater for irrigation is also a growing concern to cope with the occasional late arrival of the rainy season or to produce food during the dry season. Since the groundwater resource in the province has not been documented, a 4-year study was undertaken (2011–2014), aiming to estimate the capability of groundwater to supply domestic needs and supplementary irrigation for rice production. Aquifer properties were estimated by combined use of hydrogeological techniques with the geophysical magnetic resonance sounding method. Groundwater storage and recharge were estimated based on new developments in the application of the geophysical method for quantifying specific yield. The median groundwater storage of the targeted sandstone aquifer is 173 mm, the recharge is diffuse and annually ranges from 10 to 70 mm, and the transmissivity is low to medium. Simulations of pumping indicate that the aquifer can easily supply 100 L of drinking water per capita daily, even considering the estimated population in 2030. However, the shallow aquifer can generally not deliver enough water to irrigate paddy fields of several hectares during a 2-month delay in the onset of the monsoon.     

Modelling Transient Ground Surface Temperatures of Past Rockfall Events: Towards a Better Understanding of Failure Mechanisms in Changing Periglacial Environments

Despite the rising interest in mountain permafrost due to climatic changes and a noticed increase of registered rockfall events in the European Alps and other mountain ranges, little is known about transient thermal conditions in the detachment areas of rockfalls. Temperature conditions prior to the rockfall events of 144 past events in the European Alps were modelled with a physically based ground temperature model. To minimise the impact that uncertainty has on interpretations, only relative values were used, that is, percentiles obtained from cumulative distribution functions of the modelled ground surface temperatures from the beginning of the meteorological measurement series up to the event dates. Our results suggest that small and mid‐sized rockfalls (volumes up to 100 000 m³) from high elevation occurred mainly during short‐term periods of unusually high temperatures. This was neither found to be a result of the seasonal distribution (most analysed events in higher elevations occurred from July to September) nor of the longer‐term temporal distribution (most analysed events occurred after 2000) only. Plausible explanations are either a destabilisation related to advective thaw or failure due to stress redistribution caused by large temperature variations. Large deep‐seated rock slope failures (≥100 000 m³) in high elevation occurred all year round.     

Tomography of crustal seismic attenuation in Metropolitan France: implications for seismicity analysis

In this work, we map the absorption properties of the French crust by analyzing the decay properties of coda waves. Estimation of the coda quality factor \(Q_{c}\) in five non-overlapping frequency-bands between 1 and 32 Hz is performed for more than 12,000 high-quality seismograms from about 1700 weak to moderate crustal earthquakes recorded between 1995 and 2013. Based on sensitivity analysis, \(Q_{c}\) is subsequently approximated as an integral of the intrinsic shear wave quality factor \(Q_{i}\) along the ray connecting the source to the station. After discretization of the medium on a 2-D Cartesian grid, this yields a linear inverse problem for the spatial distribution of \(Q_{i}\). The solution is approximated by redistributing \(Q_{c}\) in the pixels connecting the source to the station and averaging over all paths. This simple procedure allows to obtain frequency-dependent maps of apparent absorption that show lateral variations of \(50\%\) at length scales ranging from 50 km to 150 km, in all the frequency bands analyzed. At low frequency, the small-scale geological features of the crust are clearly delineated: the Meso-Cenozoic basins (Aquitaine, Brabant, Southeast) appear as strong absorption regions, while crystalline massifs (Armorican, Central Massif, Alps) appear as low absorption zones. At high frequency, the correlation between the surface geological features and the absorption map disappears, except for the deepest Meso-Cenozoic basins which exhibit a strong absorption signature. Based on the tomographic results, we explore the implications of lateral variations of absorption for the analysis of both instrumental and historical seismicity. The main conclusions are as follows: (1) current local magnitude \(M_{L}\) can be over(resp. under)-estimated when absorption is weaker(resp. stronger) than the nominal value assumed in the amplitude-distance relation; (2) both the forward prediction of the earthquake macroseismic intensity field and the estimation of historical earthquake seismological parameters using macroseismic intensity data are significantly improved by taking into account a realistic 2-D distribution of absorption. In the future, both \(M_{L}\) estimations and macroseismic intensity attenuation models should benefit from high-resolution models of frequency-dependent absorption such as the one produced in this study.     

Which objective function to calibrate rainfall–runoff models for low-flow index simulations?

Much has been written on the subject of objective functions to calibrate rainfall–runoff models. Many studies focus on the best choice for low-flow simulations or different multi-objective purposes. Only a few studies, however, investigate objective functions to optimize the simulations of low-flow indices that are important for water management. Here, we test different objective functions, from single objective functions with different discharge transformations or using low-flow indices, to combinations of single objective functions, and we evaluate their robustness and sensitivity to the rainfall–runoff model. We find that the Kling and Gupta efficiency (KGE) applied to a transformation of discharge is inadequate to fulfil all assessment criteria, whereas the mean of the KGE applied to the discharge and the KGE applied to the inverse of the discharge is sufficient. The robustness depends on the climate variability rather than the objective function and the results are insensitive to the model.

EDITOR A. Castellarin; ASSOCIATE EDITOR C. Perrin     

Sidewall erosion: Insights from in situ-produced 10Be concentrations measured on supraglacial clasts (Mont Blanc massif,France)

Sidewall erosion because of rockfalls is one of the most efficient erosional processes in the highest parts of mountain ranges; it is therefore important to quantify sidewall erosion to understand the long-term evolution of mountainous topography. In this study, we analyse how the ¹⁰Be concentration of supraglacial debris can be used to quantify sidewall erosion in a glaciated catchment. We first analyse, in a glaciated catchment, the cascade of processes that move a rock from a rockwall to a supraglacial location and propose a quantitative estimate of the number of rockfalls statistically mixed in a supraglacial sand sample. This model incorporates the size of the rockwall, a power law distribution of the size of the rockfalls and the mean glacial transport velocity. In the case of the Bossons glacier catchment (Mont Blanc massif), the ¹⁰Be concentrations obtained for supraglacial samples vary from 1.97 ± 0.24 to 23.82 ± 1.68 × 10⁴ atoms g⁻¹. Our analysis suggests that part of the ¹⁰Be concentration dispersion is related to an insufficient number of amalgamated rockfalls that does not erase the stochastic nature of the sidewall erosion. In the latter case, the concentration of several collected samples is averaged to increase the number of statistically amalgamated rockfalls. Variable and robust ¹⁰Be-derived rockwall retreat rates are obtained for three distinct rockfall zones in the Bossons catchment and are 0.19 ± 0.08 mm year⁻¹, 0.54 ± 0.1 mm year⁻¹ and 1.08 ± 0.17 mm year⁻¹. The mean ¹⁰Be retreat rate for the whole catchment (ca. 0.65 mm year⁻¹) is close to the present-day erosion rate derived from other methods. © 2019 John Wiley & Sons, Ltd.     

Noise properties of gravitational lens mass reconstruction

Gravitational lensing is potentially able to observe mass-selected haloes, and to measure the projected cluster mass function. An optimal mass selection requires a quantitative understanding of the noise behaviour in mass maps. This paper is an analysis of the noise properties in mass maps reconstructed from a maximum-likelihood method.
The first part of this work is the derivation of the noise power spectrum and the mass error bars as a straightforward extension of the Kaiser & Squires algorithm for the case of a correlated noise. Very good agreement is found between these calculations and the noise properties measured in the mass reconstructions limited to non-critical clusters of galaxies. It demonstrates that Kaiser & Squires and maximum-likelihood methods have similar noise properties and that the weak lensing approximation is valid for describing these properties .
In a second stage I show that the statistics of peaks in the noise follows accurately the peak statistics of a two-dimensional Gaussian random field (using the BBKS techniques) if the smoothing aperture contains enough galaxies. This analysis provides a full procedure for deriving the significance of any convergence peak as a function of its amplitude and profile.
I demonstrate that a detailed quantitative analysis of the structures in mass maps can be carried out, and that, to a very good approximation, a mass map is the sum of the lensing signal and known two-dimensional Gaussian random noise. A straightforward application is the measurement of the projected mass function in wide-field lensing surveys, down to small mass overdensities that are individually undetectable.     

The oxidation of carbonate green rust into ferric phases:solid-state reaction or transformation via solution

The oxidation of carbonate green rust, GR(CO₃²⁻), in NaHCO₃ solutions at T = 25°C has been investigated through electrochemical techniques, FTIR, XRD, TEM and SEM. The used GR(CO₃²⁻) samples were made of either suspended solid in solution or a thin electrochemically formed layer on the surface of an iron disc. Depending on experimental conditions, oxidation occurs, with or without major modifications of the GR(CO₃²⁻) structure, suggesting the existence of two pathways: solid-state oxidation (SSO) leading to a ferric oxyhydroxycarbonate as the end product, and a dissolution-oxidation-precipitation (DOP) mechanism leading to ferric oxihydroxides such as lepidocrocite, goethite, or ferrihydrite. A formula was proposed for this ferric oxyhydroxycarbonate, Fe₆^IIIO_(2+x)(OH)_(12-2x)(H₂O)_x(CO₃), assuming that the solid-state oxidation reaction is associated to a deprotonation of the water molecules within the interlayers, or of the hydroxyl groups in the Fe(O,H) octahedra layers. The DOP mechanism involves transformation via solution with the occurrence of soluble ferrous-ferric intermediate species. A discussion about factors influencing the oxidation of carbonate green rust is provided hereafter. The ferric oxyhydroxycarbonate can be reduced back to GR(CO₃²⁻) by a reverse solid-state reduction reaction. The potentiality for a solid-state redox cycling of iron to occur may be considered. The stability of the ferric oxyhydroxycarbonate towards thermodynamically stable ferric phases, such as goethite and hematite, was also studied.     

Direct and indirect inversions

A bridge is highlighted between the direct inversion and the indirect inversion. They are based on fundamental different approaches: one is looking after a projection from the data space to the model space while the other one is reducing a misfit between observed data and synthetic data obtained from a given model. However, it is possible to obtain similar structures for model perturbation, and we shall focus on P-wave velocity reconstruction. This bridge is built up through the Born approximation linearizing the forward problem with respect to model perturbation and through asymptotic approximations of the Green functions of the wave propagation equation. We first describe the direct inversion and its ingredients and then we focus on a specific misfit function design leading to a indirect inversion. Finally, we shall compare this indirect inversion with more standard least-squares inversion as the FWI, enabling the focus on small weak velocity perturbations on one side and the speed-up of the velocity perturbation reconstruction on the other side. This bridge has been proposed by the group led by Raul Madariaga in the early nineties, emphasizing his leading role in efficient imaging workflows for seismic velocity reconstruction, a drastic requirement at that time.