Heat as a tracer for examining depth-decaying permeability in gravel deposits.

Depth dependence of permeability can appear in any geologic setting; however, vertical trends in alluvial gravel deposits are poorly understood because of the high variability of hydraulic conductivity K in monotonic sequences. This paper examines the sensitivity of depth-decaying permeability through heat transport simulation around a river's losing reach in the Toyohira River alluvial fan, Japan. Observed variations in groundwater temperature indicate that heat fluxes are dominant in the shallow zone, despite a vertical hydraulic gradient. In eight cases with different conditions (presence or absence of exponential decay trend, large or small variogram range, and cell isotropy or anisotropy) 1000 K realizations are stochastically generated throughout a cross-sectional model. The groundwater flow and heat transport are transiently calculated, and the averaged root mean square error RMSE‾ is used for sensitivity comparison. The variance of RMSE‾ shows that small RMSE‾ realizations are effectively reproduced with vertical trend assumed. Plausible realizations of RMSE‾ below a given threshold were obtained only when a vertical trend was assumed. The most plausible realization almost completely matched the observations. However, the number of plausible realizations per case was ≤10 and the median RMSE‾ were insensitive to all the conditions. Statistical testing suggested that these plausible realizations may be statistically significant, aiding in generating a connected K zone for high heat flows. The cell anisotropy condition had the smallest effect on the simulation. Thus, effective modeling of the vertical trend contributes to heat transport; however, the model's efficiency is low without detailed information about the sedimentary structure.