Analysis of the morphology of dissolution-bedform assemblages and hydraulically-transported sediments found within conduits in carbonate aquifers in North America and the British Isles has allowed the hydraulic conditions under which conduit flows occur to be established. Mean values of flow velocity, boundary-shear stress, conduit Reynolds number, conduit Froude number, boundary friction factor, boundary roughness and flow power have been calculated. The values obtained are in agreement with other evidence in the conduits and are comparable to those obtained from other, similar, hydraulic systems.

The regional movement of shallow groundwater in the fractured rock aquifer is examined through a conceptual-deterministic modeling approach. The computer program FRACNET represents the fracture zones as straight laminar flow conductors in connection to regional constant head boundaries within an impermeable rock matrix. Regional scale fracture zones are projected onto the horizontal plane, invoking the Dupuit-Forchheimer assumption for flow. The steady state flow solution for the two dimensional case is achieved by requiring nodal flow balances using a Gauss-Seidel iteration. Computer experiments based on statistically generated fracture networks demonstrate the emergence of preferred flow paths due to connectivity of fractures to sources or sinks of water, even in networks of uniformly distributed fractures of constant length and aperture. The implication is that discrete flow, often associated with the local scale, may maintain itself even at a regional scale. The distribution of uniform areal recharge is computed using the Analytic Element Method, and then coupled to the network flow solver to complete the regional water balance. The areal recharge weakens the development of preferential flow pathways. The possible replacement of a discrete fracture network by an equivalent porous medium is also investigated. A Mohr's circle analysis is presented to characterize the tensor relationship between the discharge vector and the piezometric gradient vector, even at scales below the representative elementary volume (REV). A consistent permeability tensor is sought in order to establish the REV scale and justify replacement of the discrete fracture network by an equivalent porous medium. Finally, hydrological factors influencing the chemical dissolution and initiation of conduits in carbonate (karst) terrain are examined. Based on hydrological considerations, and given the appropriate geochemical and hydrogeological conditions, the preferred flow paths are expected to develop with time into caves.

An analysis of the discharge and hydrochemical variations of contrasting springs at Crowsnest Pass showed they were part of a vertical hierarchy in the aquifer, in which underflow and overflow components play a dominant role. It was found that karst springs at Crowsnest Pass and elsewhere show a range between two end members. Thermal springs have long, deep flow paths, with high sulphate concentrations, low discharge variance and low flow velocities. Overflow springs have local shallow flow paths, low sulphate, high discharge variance, and high flow velocities. Intermediate between these end members are underflow springs; in the Rocky Mountains these are mostly aggraded, and give the sustained winter flow and high sulphate concentrations found in major rivers. It was found that underflow or overflow behaviour is able to explain most of the contrasts found between karst springs in discharge and sulphate concentrations. Conversely, differences in bicarbonate concentration are principally due to the ratio of allogenic to autogenic recharge to the aquifer. Hydraulic analysis showed that gradients decrease in the downstream direction, and are typically 0.0001-0.05 at maximum discharges, that friction factors vary by a factor of $>$1000, and that most active conduits have closed-channel flow and are in dynamic equilibrium with sediment supply. The analysis of the hydrological data from Crowsnest Pass and elsewhere has led to the development of a new conceptual model for groundwater flow in karst, in which the Hagen-Poiseuille flow net conditions the aquifer for conduit development, and determines where the conduits will be. The model explains why most conduits are in dynamic equilibrium with sediment supply, why temperate karst springs are mostly vauclusian, what the mean time for speleogenesis is, how $>$98% of the solution of limestone is in the surficial zone, and why there are karstic hot springs in the Rocky Mountains and elsewhere. The model enables predictions to be made of sink to resurgence flow velocities, of conduit depth below the water table, of the ratio of beds to joints used by conduits, of the spacing between cave tiers, and of the depth of vauclusian springs. This new understanding of how karstic aquifers develop and function gives a powerful predictive ability to karst hydrogeology.

This work, based on the study of several underground alpine networks, aims to propose some milestone in the history of these karstic regions.

The first part of the work is made up of three regional studies.

The Tennengebirge mountains are a massif of the limestone High Alps in the region of Salzburg in Austria. A cone karst close to the base level developed in the Neogene. Streams from the Alps fed the karst, resulting in the huge horizontal networks of which the Eisriesenwelt provides evidence. During the successive phases of upthrust, the levels of karstification, whether on the surface or deeper down, settled into a tier pattern, thus descending in stages from the base level. From the Pliocene era onwards, thanks to an increase in potential, alpine shafts replace the horizontal networks. The formation of these shafts is more pronounced during glaciation. The study of the Cosa Nostra - Bergerhöhle system developing 30 km of conduits on a gradient reaching almost 1 500 m provides a fairly full view of the karstification of this massif. It includes the horizontal levels developed in the Miocene and the Plio-Pleistocene, joined together by vertical sections. The most noteworthy features of the Tennengebirge, as in the neighboring massifs, lie first and foremost in the extreme thickness of the limestone which has recorded and immunized the differents steps of karstification. Secondly, the size of the networks can be, for the most part, accounted for by the contribution of allogenous waters from the streams of the Neogene and the glaciers of the Pleistocene. Generally sudden and unexpected, these flows of water engendered heavy loads (up to 600 m), simultaneously flooding several levels. To a lesser extent, the situation is similar today.

The Ile de Cremieu is a low limestone plateau on the western edge of the Jura. Due to its location in the foothills, the lobes of the Rhône glacier have covered it up, obliterating the surface karst. However, widespread evidence of anteglacial morphologies remains : paleokarst, cone karst, polygenic surface. Because of glacial plugging, access to the underground karst is limited. The main cavity is the cave of La Balme. Its initial development dates back to an early period. The morphological study has permitted the identification of several phases which go back to the Pleistocene and which are related to the Rhône glacier. The latter brought about modifications in the base level by supplying its merging waters as well as moraine material. These variations in the base level shaped the drainage structure. The underground glacial polishes are one of the noteworthy aspects recorded.

The massives of the Moucherotte and dent de Crolles belong to the northern French Prealps. They conceal large networks, respectively the Vallier cave and the Dent de Crolles. They were formed in the early Pliocene after the final orogenic phase and are in the form of horizontal conduits. The upthrust, which brought about the embanking of the Isère valley, left them in a perched position by taking away the basin which fed them. They were later, however, able to take advantage of waters from the Isère glacier during a part of the Pleistocene. The Vallier cave contains particularly glacio-karstic sediments of the lower Pleistocene, representing unique evidence of glaciation during this period. The vertical networks were put in place at the end of the Pliocene with the increase in karstification potential ; they underwent changes in the Pleistocene due to the effect of autochton and allogenous glaciers.

The second part of the work deals in general with the various forms and processes of karstification, sometimes going beyond the Alps. The study of cave deposits is a privileged tool in the understanding and reconstruction not only of the history of the networks but also the regional environment. The dating of speleothems by the U / Th method has very ofen given an age of over 350 000 years. The age of the networks is confirmed by the use of paleomagnetism which has yielded evidence of speleothems and glacio-karstic sediments anterior to 780 000 years. Anisotropic measurements of magnetic susceptibility have been used to distinguish the putting into place of glacio-karstic deposits by decantation.

Measurements of calcite rates lead to a typology of sediments based on their nature and carbonate content (rehandled weathered rocks, fluvial sands, carbonated varves, decantation clays). Granulometry confirms this differenciation by supplying precise details of transport and sedimentation modes : suspension and abrupt precipitation of clay, suspension and slow decantation of carbonated varves, suspension and rolling together with a variable sorting of sand and gravel. Mineralogical analyses oppose two types of detrital deposits. On the one hand, the rehandling of antequaternary weathered rocks extracted by the karst as a result of scouring during environmental destabilization and on the other hand, sediments characteristic of the ice age of the Pleistocene. The latter are not highly developed and their arrival in the karst is always later. Examination of heavy minerals, the morphoscopy of quartz grains and study of micromorphologies on thin blades provide precise details of conditions of evolution. The use of these methods of investigation allows for an accurate definition of the features of the evolution of the differents types of fillings, particularly speleothems, rehandled weathered rocks as well as carbonated varves. This wealth and complexity are emphasized by a detailed study of the sedimentary sequences of the Vallier cave and of the Bergerhöhle.
Speleogenesis is approached last of all in the light of above study. Emphasis is placed on the major part played by corrosion in the temporarily phreatic zone and on its many consequences (multi-level concept, simultaneous evolution of levels, origin of deep waterlogged karsts…).
Varia tions in the base level have induced karstification in contexts in which the potential was weak. These were followed by periods of increased potential to which were added the effects of glaciation. Perched horizontal levels belong to the first stages which ended in the early Pliocene, whereas alpine shafts developed in the second context. The role of structure and the parameters governing the shape of conduits (pits, meanders, canyons) are also dealt with. The different parts of the karst are borne in mind when dealing with the strength of karstic erosion during the ice age. It notably appears that it is weak on the crests and more or less non-existent in the deep parts of the karst which are liable to flooding. Finally, a preliminary analysis of an observation of neotectonic traces is presented.