An alternative approach to understanding groundwater flow in sparse channel networks supported by evidence from ‘background’ fractured crystalline rocks | Une approche alternative pour comprendre les écoulements d’eau souterraine dans les réseaux de chenaux épars, étayée par des preuves en contexte de roches cristallines fracturées Un enfoque alternativo Para entender el flujo de agua subterránea en redes de Canales dispersos respaldado por evidencia de rocas cristalinas fracturadas de “basamento” 由“背景”断裂结晶岩证据支持的了解稀疏通道网络内地下水流的可供选择的方法 Uma abordagem alternativa Para entender o fluxo de águas subterrâneas em rede de canais esparsos suportada pela evidencia do contexto de rochas cristalinas fraturadas

Size and shape of individual flow-features, and not their ‘organization’ in sets of predominant orientation, are the major influences on the ability of groundwater to percolate through sparse channel networks. Measurements in background fractured crystalline rocks proposed for nuclear waste repositories provide useful insight. Flow-features are observed as locations of increased transmissivity during packer or flow testing in boreholes. They are conceived here as channels on fracture surfaces. Findings are based on numerical modelling and a general formula by Barker (2018) for the percolation of two-dimensional (2D) objects in 3D space. Equidimensional shapes are found to be the least efficient at forming percolating networks. As discs are evolved into highly eccentric ellipses, percolation thresholds for number, area and intersection density decrease. At the same time, the percentage of features forming the active flow path declines from about 10% for discs to a few per cent for 50:1 ellipses. Compiling recent field measurements of area density of flow-features reveals low values within a limited range (0.01–0.8 m⁻¹). When this range is combined with practical values of likely channel width, long narrow flow-features are the only reasonable components of a sparse percolating network. Conventional equidimensional discrete fracture networks are considered unlikely. Innovative field investigation and modelling methods based only on hydrogeological measurements are suggested. It is concluded that this consideration of shape supports the approach, broadly termed the ‘long channel’ concept. Barker J.A. (2018) Intersection statistics and percolation criteria for fractures of mixed shapes and sizes. Comput Geosci 112:47–53.