A new structural model is presented for the Permo-Carboniferous Lorraine Basin (NE France), a major intramountain basin that developed during the latest stages of the Variscan orogeny (ca 315–270 Ma). This basin is buried in NE France below the Paris Basin but outcrops in southern Germany (Saar-Nahe Basin). Digitalized well logs and reprocessed seismic data were used to decipher the kinematic evolution of this basin located along the Rhenohercynian orogenic suture zone. The basin initiated during the late collision stage (Late Namurian-Westphalian) in a wedge- top position upon the Saxothüringian retrowedge. The syn-orogenic sequences are delimited to the north by the major SE-directed Metz Thrust, which is part of the backthrust system that propagated during Late Westphalian times. Seismic data provide evidence of negative tectonic inversion, allowing the formation of syn-rift depocenters (Stephanian-Early Permian) above the former anticlines. Erosion of these anticlines results in a major unconformity (base of Stephanian) marking the onset of post-orogenic collapse stage. The late Early Permian shortening (Saalian phase) reactivated former thrusts and normal faults, thus generating uplift of the basin. The post- orogenic phase is complex and diachronous at basin scale, and both compression and extension can be recorded in the same area over a short period (<10 Ma). The Late Carboniferous negative tectonic inversion along the Rhenohercynian suture zone is proposed to result from the lithospheric delamination of the Variscan orogenic roots. The associated upwelling of asthenospheric material is recorded by intense magmatic activity, and can be, in turn, considered as the main trigger for the subsequent thermal subsidence of the Paris basin.

In the general context of the development of renewable energy in the Hauts-de-France region (N France), some growing interest has been focused recently on the potential of deep geothermy. This area displays favourable conditions due to the burial of a regionally well-defined reservoir, i.e. the Dinantian karstic and brecciated limestones (Lower Carboniferous, 360-330 Ma), below the Nord-Pas-de-Calais coal-bearing Upper Carboniferous basin, developed by flexural subsidence in the foreland of the Northern Variscan frontal thrust system. The predominance of shales within the molassic basin as well as within the basal units at the floor of the thrust wedge (the Lower Devonian clastic units) are furthermore likely to form a large-scale permeability barrier potentially favouring the localization of hot waters within the underlying carbonate reservoir. The occurrence of a Dinantian regional geothermal resource has already been proven in Southern Belgium in the Hainaut coal basin area (the eastern prolongation of the Northern France coal basin) where the temperature in three geothermal wells reaches about 70°C.To provide further constraints on such potential deep geothermal field in a structurally complex setting (a laterally segmented thrust front), the geometry of the Dinantian reservoir in northern France has been investigated through the integration and interpolation in a 3-D model of a large database including 1 128 boreholes and 532 km of reprocessed, interpreted and depth-converted seismic reflection profiles. The results of the 3-D modelling indicate that the Dinantian reservoir is present at depth over a large area covering approximately 7675 km² in northern France-southern Belgium. It extends at least 30 to 40 km south of the coal mining district area, underneath the Ardennes Allochthonous Unit of the Northern Variscan Front. The Dinantian reservoir is less than 200 m deep in the Lille metropolitan area and strongly deepens southward through a sharp flexure. It reaches 1000-3000 m depth underneath the coal basin and a maximum of 6944 m depth at the southern end of the study area. Overall, the Dinantian reservoir is structured along two main directions oriented N70-80° and N110-130°, related respectively to deep frontal Variscan thrusts and lateral-oblique ramps. The Dinantian reservoir ends west of Douai against a major complex lateral ramp system forming a first-order transfer zone within the Northern France Variscan thrust front. The latter localizes a set of strongly dipping N110-N130 faults (the Artois faults) representing second-order structures produced during subsequent deformation periods i.e. the Late Carboniferous-Permian rifting event and the Tertiary inversion related to the far-field accommodation of the Alpine-Pyrenean shortening.

An increasing number of cities are interested in deep geothermal energy in order to increase the share of renewable energies in their district heating networks. To reduce the risks related to deep geothermal energy operations, reliable digital models are needed: they make it possible to predict the depth of aquifers away from borehole locations, and their thermal and hydrological evolution by supporting detailed water and heat flow simulations. This paper presents a 3D geomodel developed for this purpose in the southern Paris Basin of France in the Orléans area. The 3D geomodel integrates various data such as reprocessed and interpreted seismic lines, well data, and a pre-existing larger-scale and lower-resolution 3D geological model. The resulting 3D geomodel gives a new and reliable representation of the main aquifers underlying the study area. Within the framework of the project, hydrological and thermal simulations were then performed based on this 3D geomodel. Other environmental investigations (e.g. CO2 storage) and teaching/communication activities could also benefit from the dataset.

In Ile-de-France region, in the center of Paris Basin, geothermal energy contributes to a large extent to the supply of heating networks with about 50 of the 70 deep geothermal installations dedicated to district heating in France. Those installations mainly exploit the Dogger limestones between 1500-2000 m deep, which are present throughout the Paris Basin. In the case of Centre Val-de Loire region, south of Paris Basin, deep geothermal energy is very little developed, only one geothermal well is currently in operation and targeting the Triassic aquifer at Chateauroux on the southern edge of the basin. A former doublet had also targeted the Trias at Melleray (Orléans metropolis) in the 1980’s but was shut down after one year due to reinjection problem. In 2019, Orléans metropolis, in collaboration with BRGM, has launched a program in order to investigate its deep geothermal resources like the Dogger and Trias aquifers between 900 m and 1500 m deep. This action is in line with Orléans métropolis Territorial Climate Air Energy Plan (PCAET) and master plan for the heating networks adopted which foresee 65 000 additional dwellings to be connected using geothermal energy based heating networks. In order to reduce the risks of failure of deep geothermal drilling, one of the prerequisites is a better knowledge of the subsurface. This requires the development of an accurate 3D subsurface geomodel as well as the most reliable possible hydrodynamic and thermal parameters to assess the geothermal potential. The purpose of this work was to produce a 3D geological model of the Dogger and Triassic units, on the scale of Orléans Metropolis, based on hydrocarbon and geothermal well data as well as interpretation of 2D seismic data. Seismic data acquired in the 1960s and the 1980s were processed and interpreted. A particular attention was paid to the Sennely fault and its geometry. It crosses the study area and was interpreted as a relay fault segmented in three parts. The horizon picking points were then converted from two-way time to depth and integrated in the GeoModeller software for the development of the 3D geomodel. It was then used for first hydrothermal simulations in order to assess the theorical potential of the Dogger and Trias aquifers at Orléans metropolis. The 3D geomodel and first geothermal potential assessment have allowed defining areas of interest for geothermal development into the Dogger or Trias. However an initial exploratory drilling well or additional exploration techniques will be necessary to confirm/specify the reservoir properties (useful thickness, porosity, permeability) and the connectivity of the reservoir(s) and the flow rates that can actually be exploited, which cannot be predicted by the current geological model.

In Ile-de-France region, in the center of Paris Basin, geothermal energy contributes to a large extent to the supply of heating networks with about 50 of the 70 deep geothermal installations dedicated to district heating in France. Those installations mainly exploit the Dogger limestones between 1500-2000 m deep, which are present throughout the Paris Basin. In the case of Centre Val-de Loire region, south of Paris Basin, deep geothermal energy is very little developed, only one geothermal well is currently in operation and targeting the Triassic aquifer at Chateauroux on the southern edge of the basin. A former doublet had also targeted the Trias at Melleray (Orléans metropolis) in the 1980’s but was shut down after one year due to reinjection problem. In 2019, Orléans metropolis, in collaboration with BRGM, has launched a program in order to investigate its deep geothermal resources like the Dogger and Trias aquifers between 900 m and 1500 m deep. This action is in line with Orléans métropolis Territorial Climate Air Energy Plan (PCAET) and master plan for the heating networks adopted which foresee 65 000 additional dwellings to be connected using geothermal energy based heating networks. In order to reduce the risks of failure of deep geothermal drilling, one of the prerequisites is a better knowledge of the subsurface. This requires the development of an accurate 3D subsurface geomodel as well as the most reliable possible hydrodynamic and thermal parameters to assess the geothermal potential. The purpose of this work was to produce a 3D geological model of the Dogger and Triassic units, on the scale of Orléans Metropolis, based on hydrocarbon and geothermal well data as well as interpretation of 2D seismic data. Seismic data acquired in the 1960s and the 1980s were processed and interpreted. A particular attention was paid to the Sennely fault and its geometry. It crosses the study area and was interpreted as a relay fault segmented in three parts. The horizon picking points were then converted from two-way time to depth and integrated in the GeoModeller software for the development of the 3D geomodel. It was then used for first hydrothermal simulations in order to assess the theorical potential of the Dogger and Trias aquifers at Orléans metropolis. The 3D geomodel and first geothermal potential assessment have allowed defining areas of interest for geothermal development into the Dogger or Trias. However an initial exploratory drilling well or additional exploration techniques will be necessary to confirm/specify the reservoir properties (useful thickness, porosity, permeability) and the connectivity of the reservoir(s) and the flow rates that can actually be exploited, which cannot be predicted by the current geological model.

Deep neural networks (DNNs) have the potential to streamline the integration of seismic data for reservoir characterization by providing estimates of rock properties that are directly interpretable by geologists and reservoir engineers instead of elastic attributes like most standard seismic inversion methods. However, they have yet to be applied widely in the energy industry because training DNNs requires a large amount of labeled data that is rarely available. Training set augmentation, routinely used in other scientific fields such as image recognition, can address this issue and open the door to DNNs for geophysical applications. Although this approach has been explored in the past, creating realistic synthetic well and seismic data representative of the variable geology of a reservoir remains challenging. Recently introduced theory-guided techniques can help achieve this goal. A key step in these hybrid techniques is the use of theoretical rock-physics models to derive elastic pseudologs from variations of existing petrophysical logs. Rock-physics theories are already commonly relied on to generalize and extrapolate the relationship between rock and elastic properties. Therefore, they are a useful tool to generate a large catalog of alternative pseudologs representing realistic geologic variations away from the existing well locations. While not directly driven by rock physics, neural networks trained on such synthetic catalogs extract the intrinsic rock-physics relationships and are therefore capable of directly estimating rock properties from seismic amplitudes. Neural networks trained on purely synthetic data are applied to a set of 2D poststack seismic lines to characterize a geothermal reservoir located in the Dogger Formation northeast of Paris, France. The goal of the study is to determine the extent of porous and permeable layers encountered at existing geothermal wells and ultimately guide the location and design of future geothermal wells in the area.

In NW Europe, the Upper Carboniferous Variscan collision between Avalonia and the Armorica-Gondwana accretion complex led to the progressive tectonic inversion of the southern Avalonian margin and the development of a crustal-scale north-vergent thrust system propagating outward from the Late Mississippian to the Middle Pennsylvanian (330-305 Ma). The northern Variscan thrust front spreads over 2,000 km across NW Europe. In the Nord-Pas-de-Calais (NPC) coal district area (northern France), its 3D geometry and kinematics have been investigated through the reprocessing and interpretation of 532 km in length of industrial seismic reflection profiles acquired in the 1980s. The seismic interpretations point out the major compressional and extensional tectonic features affecting this fossil, deeply eroded, mountain front, highlighting its very atypical structure and kinematics. The deformation front is characterized by a main frontal thrust zone localizing most of the northward displacement (i.e. several tens of kilometers) of the Ardennes Allochthonous Unit above the slightly-deformed part of the Avalonian margin, referred to as the Brabant Para-autochthonous Unit. This large displacement induced the underthrusting of the molassic foreland basin (NPC coal basin) over nearly 20 km and was associated to the out-of-sequence dislocation of the mountain front. The underthrust Brabant Para-autochthonous Unit, made of both the Namurian-Westphalian (330-305 Ma) molassic foreland basin and the underlying Mid-Upper Devonian (390-360 Ma) and Dinantian (360-330 Ma) carbonate platform, is deformed by a series of second-order north-vergent thrust faults, often associated with ramp-related folds. These thrust faults are rooted in décollement zones located either at the transition between the Namurian shales and the Dinantian carbonates or in the Famennian shales. The 3D integration of the seismic interpretations led to the characterization of a major lateral ramp oriented NW-SE, affecting both the main frontal thrust zone and the basal thrust of some Overturned Thrust Sheets developed at its footwall. This lateral ramp represents a major zone of relay along the thrust front, in between two major segments, oriented respectively ENE-WSW to the east and WNW-ESE to the west. At the base of the underthrust Brabant Para-autochthonous Unit, the Mid-Upper Devonian platform is shown to be structured by synsedimentary normal faults responsible for the southward deepening and thickening of the southern Avalonian margin. These faults are oriented along two main directions i.e. N060-080° and N110-130°, that is the general orientation of the future Variscan structures. Overall, the results indicate that the Devonian pre-structuration of the southern Avalonian margin exerted a primary control on the dynamics and segmentation of the Northern Variscan Front in northern France by localizing both the frontal and lateral ramps within the thrust wedge.

On-line, real-time chemical and mineralogical analyses on drill cores are highly demanded by mining companies. However, they are a challenge because of drill core surface state and sample heterogeneities. We selected four rock samples: highly porous, siliceous breccia and serpentinized harzburgite coming from the base of a nickel laterite profile in New Caledonia which were sonic drilled, and fine grained, homogeneous sandstone and coarse grained granite which were diamond drilled and provided by Eijkelkamp Sonic Drill with unknown origin. The samples were analysed at five surface states (diamond or sonic drilled, cut as squares, polished at 6 and 0.25 mu m, powdered < 80 mu m) by portable XRF spectroscopy (pXRF) in mining and soil modes and portable infrared spectroscopy (pIR, Visible and Near Infrared-Short Wave Infrared range (VNIR-SWIR)). A total of 52 pXRF and 200 pIR analyses were performed per sample at each surface state. This study shows that the surface state has minor influence on the results of the portable instruments. By comparing pIR and pXRF results with laboratory devices (Raman spectroscopy, XRD with Rietveld refinement, XRF spectroscopy and ICP-AES), we evidence the lower and less accurate information obtained from handheld instruments in terms of chemistry and mineralogy. The porosity and grain size effect on the measurement need to be taken into consideration for on-line drill core analyses. We show that the combination of complementary analytical techniques helps to overcome the drawbacks of the core texture and of the precision of portable instruments in order to define the regions of interest (ROI) for mining companies. We also demonstrate that a precise pXRF calibration is mandatory and that the concentration of light elements (Si, Mg), even if not accurate, shows sufficient contrast along the lateritic profile for ROI definition.

Within the framework of responsible mining, a SOLSA project (www.solsa-mining.eu) to develop an in-situ tool allowing a quick mineralogical identification of site drill cores has been recently launched. Its objective is to develop new or improved highly-efficient and cost-effective, sustainable exploration technologies. It combines and integrates non-destructive sensors: X-ray fluorescence, X-ray diffraction, infra-red and Raman spectroscopy and 3D imaging. The challenge is to address mixtures of hard and soft rocks, as encountered in a lateritic environment. This paper focuses on the determination of spectral characteristics of laterite drill-cores in the visible to short wave infrared spectral range. One of the most important prerequisites is to study the influence of the surface roughness effect on infra-red spectroscopy analyses. For this purpose, four different rock samples: breccia, sandstones, granite and peridotite, each at five surface states have been considered: as-drilled, as-sawn, polished at 6 μm, polished at 0.25 μm and crushed to powder. The reflectance spectra have been acquired with an ASD Fieldspec 3® spectroradiometer with a contact probe at a sampling surface of 1.76 cm 2 , allowing a spectral analysis at wavelengths from 350 up to 2500 nm. The powder spectrum of breccia presents a higher reflectance than the four other spectra from the same material but weak absorption features. The as-sawn sample presents the higher absorption depth, followed by as-drilled sample and the two polished samples (figure 1). At wavelength 2219 nanometers, a peak of absorption is present. The presence of clay minerals is assumed like illite/sericite with more or less smectite, due to the relatively deep water absorption around wavelength 1900 nanometers.

In order to evaluate the instrumental parameters for the combined on-line-on-mine-real-time expert system SOLSA (http://www.solsa-mining.eu), portable and laboratory analyses were carried out on coarse granite, sandstone, serpentinized harzburgite and siliceous breccia. Each sample was studied at 5 different surface roughnesses (sonic or diamond drilled, cut, polished at 6 mu m and 0.25 mu m, sample powders). X-ray diffraction (XRD), portable Infra-Red (pIR) and X-ray-fluorescence (pXRF), and laboratory micro-Raman spectroscopy gave complementary and corroborating results. No major effect on the analyses was noted for the selected surface states. pXRF gave variable results except for the homogeneously serpentinized harzburgite, related to coarse or contrasting grain sizes or pores, small spot size (3 mm) and needs close-to-surface analyses. Portable IR (spot size 1.76 cm(2)) is carried out close to surfaces while Raman spectroscopy (1-2 mu m) is performed at distance. Sampling strategies have to be defined for each lithology. Major challenges for a combined on-line analysis are to adapt the specificities of the techniques to (1) analyse similar surface areas (from 2 cm(2) (pIR) to < mu m (Raman)), (2) smartly combine all the techniques into a single instrument, and (3) develop appropriate databases to reach a reliable "real-time" outcome results, which can be used for more precise geomodeling, and to rapidly define exploration and beneficiation parameters.

Seismic reflection imaging is a geophysical method that provides greater resolution at depth than other methods and is, therefore, the method of choice for hydrocarbon-reservoir exploration. However, seismic imaging has only sparingly been used to explore and monitor geothermal reservoirs. Yet, detailed images of reservoirs are an essential prerequisite to assess the feasibility of geothermal projects and to reduce the risk associated with expensive drilling programs. The vast experience of hydrocarbon seismic imaging has much to offer in illuminating the route toward improved seismic exploration of geothermal reservoirs - but adaptations to the geothermal problem are required. Specialized seismic acquisition and processing techniques with significant potential for the geothermal case are the use of 3D arrays and multicomponent sensors, coupled with sophisticated processing, including seismic attribute analysis, polarization filtering/migration, converted-wave processing, and the analysis of the diffracted wavefield. Furthermore, full-waveform inversion and S-wave splitting investigations potentially provide quantitative estimates of elastic parameters, from which it may be possible to infer critical geothermal properties, such as porosity and temperature.

Stephanian to Permian post-orogenic basins (SPB) outcrop in several limited locations in and around the present-day French Variscan basement. Little is known about their subsurface occurrences under the post-depositional sedimentary cover. Our work intends to decipher the structural evolution of the hidden SPB in the southwestern Paris Basin, where only a few regional studies have aimed to determine their location beneath the Mesozoic sedimentary cover. Our approach is based on the reprocessing and interpretation of 36 seismic lines (1480 km) acquired by the oil industry in the 1980s. We first obtain a comprehensive view of the geometry of the SPB (Figure 1), based on (i) the interpretation of the base of the Stephano-Permian surface (lower limit) and the erosional base of the surface of the Triassic layers (upper limit) and (ii) the recognition of specific internal geometries and seismic facies. The interpreted faults are grouped into different categories according to their period of activity, with a focus on synsedimentary faults related to thickness variations of the Stephano-Permian deposits. Figure 1-The Stephano-Permian basins beneath the Mesozoic sediments of the intracratonic Paris Basin

Dans le cadre de la reconnaissance geologique du sous-sol des domaines immerges francais, le BRGM a retraite en 2012 d’anciennes donnees sismiques petrolieres des annees 70 et 80 (Projet GEONORMER). Ces donnees recouvrent le domaine offshore de la Baie de Seine au bassin de Dieppe-Hampshire, bordure septentrionale du Bassin de Paris. Le Projet GEONORMER s’integre dans un ensemble de differents projets scientifiques regionaux focalises sur la geologie de la Manche et de la bordure septentrionale du bassin de Paris: 1) connaissance geologique du substratum de la Baie de Seine en collaboration avec l’Universite de Caen –M2C- UMR6143 [These de M. Benabdellouahed (2011)* et carte geologique a 1/250 000 de la France – feuille Baie de Seine-Pays de Caux (Paquet et al., RST 2016)] ; 2) connaissance geologique du bassin Dieppe-Hampshire en collaboration avec l’universite de Lille1-LOG-UMR8187 [These de M. Jollivet-Castellot (2015-2018)*]. Au cours de ces differents projets, de nombreuses acquisitions de sismique reflexion Tres Haute Resolution imagent les 100 a 200 premiers metres du sous-sol [Campagnes Baie de Seine, 2007, 2008 et CAROBSEINE, 2009 (Universite de Caen-BRGM); campagnes MERCAUX 2013, 2015 (BRGM et coll.,), campagne TREMOR, 2014 (Universite de Lille1 et Coll.)]. Au-dela, la structure plus profonde de ce domaine offshore est mal connue. Dans le cadre du projet GEONORMER, plus de 3000 kilometres de profils sismiques des campagnes 79MOE/79MOW et 80NCE ont ete traitees via une sequence de traitement pour ces donnees sismiques marines anciennes, construite a partir des outils open source Seismic Unix©. L’analyse des donnees sismiques a montre un niveau de bruit important et d’origine diverses, une attenuation rapide en profondeur des multiples et reflexions. Afin d’obtenir une image sismique interpretable en profondeur, la majeure partie des bruits a ete attenuee via une discrimination basee sur les vitesses et l’application d’un filtre dans le domaine F-K. Les multiples issus du fond de l’eau, peu profond, et les « peg leg » ont ete fortement attenues par l’application d’un filtre radon dans le domaine Tau-P. L’image finale obtenue est une migration pre-stack en temps, permettant de s’affranchir des conflits de vitesse de propagation des ondes et de mieux imager les structures complexes, jusqu’a 2,5 secondes temps double. Les resultats obtenus montrent la presence de discordances majeures, de structures synclinales et des structures de progradation du bassin. 1*. These de M. Benabdellouahed (2007-2010) de l’universite de Caen-BRGM (dirigee par B. Tessier et O. Dugue , Guennoc P. (BRGM DGR/GBS)) 2*. These de M. Jollivet-Castellot (2015-2018) de Universite de Lille-BRGM (dirigee par Gaullier V. – LOG-UMR8187 - Universite de Lille 1, Paquet F. – BRGM-DGR/GBS)