This is the authors' abstract. We don't add key points for this paper.
Non-Fickian breakthrough curves in porous media can be produced by spatially correlated velocity heterogeneity or by mass transfer to immobile regions, and a single breakthrough curve generally cannot tell these mechanisms apart. We show that a mobile–immobile model calibrated on the breakthrough curve of a strongly heterogeneous field (\(\:{σ\:}lnK²=4\)) at eight correlation lengths reproduces it over six orders of magnitude, yet mispredicts arrival-time densities by a factor of about two at 64 correlation lengths, while its fitted dispersivity and exchange rate drift with distance. We then exploit a property shared by all spatially Markovian transport models (advection–dispersion, mobile–immobile, multirate and uncorrelated continuous time random walk models): for flux-weighted injection, the breakthrough curve at distance \(\:2d\) equals the self-convolution of the curve at \(\:d\), so that all travel-time cumulants grow linearly with distance. Violations of this identity are a model-free signature of spatially variable transport. Using two-dimensional particle-tracking simulations, we construct a self-convolution test with controlled false-alarm rate and quantify its power for particle data and for noisy, censored concentration curves. We complement it with a flow-rate test based on the contraction of the scaled late-time tail expected when exchange is diffusion-controlled. Combined, the two tests classify heterogeneity-only, exchange-only and mixed regimes correctly in 86% of cases on average with 3000 samples per curve over a sweep of log-conductivity variance and capacity ratio. The results support the design of multi-distance, multi-rate tracer tests for mechanism identification.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Hoa et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: