Modeling heterogeneity of gravel-sand, braided stream, alluvial aquifers at the facies scale

The coarse-grained alluvial aquifers are recurrently considered as almost homogeneous and isotropic media. Nevertheless, water flow through these sediments is controlled by internal architecture, therefore, a better knowledge is necessary to understand the effects of facies heterogeneity on the conductivity magnitude and anisotropy within these aquifers. In this work, we search interactions at the facies scale within a glacio-fluvial aquifer among sedimentary heterogeneity, distribution of permeability and anisotropy. The case study concerns Pleistocene sequences of the Ticino basin (Northern Italy), where three ‘model blocks’, dug into real sediments at a quarry site, have been investigated at the meter scale. These ‘model blocks’ represent the most diffuse facies associations: (1) sand-gravel dunes, (2) gravel bedforms, and (3) poorly sorted, mass gravity deposits. Facies maps with 2 cm resolution were obtained for each block and used to condition 3D geostatistical simulation of the sediment volumes with a hierarchical modification of SISIM. Conductivities of samples were either measured with laboratory tests or estimated with empirical formulas. Conductivity values were finally introduced as input data into the numerical flow model to compute the upscaled tensor of equivalent conductivity at the model block scale. The results of field infiltration tests were used to verify the outcomes of upscaling experiments, which identified the primary influence on the most permeable facies (open framework gravel) on both conductivity magnitude and anisotropy ratio. The latter property is unrelated to the abundance of this facies, but is controlled by its spatial distribution within the blocks. As the connectivity of open framework gravel increases, model blocks become more anisotropic. On the other hand, the low-permeable sand facies of dunes do not produce significant anisotropy at the block scale. These results show the importance of introducing the hierarchical assemblage of different-rank depositional units in conceptual aquifer models. In addition, we identified the most significant depositional units to forecast the spatial distribution of aquifer heterogeneity at different scales.