A primary objective in exploration for and development of fluvial reservoirs is determining the thickness and width of sandstone-conglomerate bodies (mainly channel-belt deposits). Most of the existing techniques for estimating the dimensions of fluvial reservoirs have major drawbacks. A fresh approach to the problem is made using recent theoretical, experimental, and field studies. This new approach involves (1) new models for the lateral and vertical variation of lithofacies and petrophysical-log response of river-channel deposits with explicit recognition of the different superimposed scales of strata, (2) distinction among single and superimposed channel bars, channels, and channel belts, (3) interpretation of maximum paleochannel depth from the thickness of channel bars and the thickness of sets of cross-strata formed by dunes, and (4) evaluation of various methods for estimation of widths of sandstone-conglomerate bodies that represent either single or connected channel belts (outcrop analogs; correlation of sandstone-conglomerate bodies between wells; use of empirical equations relating channel depth, channel width, and channel-belt width; theoretical models; and three-dimensional seismic data).
Two fluvial reservoirs were reinterpreted using this new approach. In the first example from the Mesaverde Group, Colorado, maximum paleochannel depth had been underestimated because the degree of superposition of channel bars had been overestimated. As a result, channel-belt widths determined from empirical equations were underestimated. In the second example from the Travis Peak Formation, Texas, channel-belt width and connectivity of channel-belt sandstone bodies had been overestimated because of overzealous well-to-well correlation and inappropriate use of width and thickness data from supposed analogs. These examples demonstrate the potential value of this new approach in reservoir characterization and management.