This manuscript is a novel contribution to studying diffusive processes in soil. A Lagrangian model is developed to account for heterogeneity in diffusive mixing in soil caused by pore structure. The pore size distribution is accounted for through varying diffusion coefficients. The model has first been tested by comparing it to the recent experiments on effectively non-Fickian diffusive mixing and shown to agree well in terms of time and concentration. Further, simulations were designed to combine diffusion and leaching in a saturated soil column and demonstrate the non-Fickian transport behaviour in this system. While the impact of physical or pore structure heterogeneity has commonly been shown to result in non-Fickian transport, this manuscript systematically increases the level of complexity in terms of diffusive processes studied and shows an interesting workflow to characterize and quantify the non-linear transport behaviour. I can therefore recommend it for publication after the following minor suggestions were considered and addressed:

• Pore space and water retention curve are subdivided using N=200. Did you investigate sensitivity to this number of bins?
• In relation to the previous question how is the bin size related to pore or throat radius that could have been actually measured in this type of soil?
• Can you comment on the expected impact of dimensionality and anisotropy?
• I would leave this to the authors to decide but to me the term “virtual experiment” is quite confusing. Why not simply say that you run your simulations to explore the model predictive capability in a more complex setting after you have demonstrated a good agreement with the experiment.

Minor:

• Line 22 instead of “… comparing…”   it should read as “…compared…”

Dear Editor, Thank you for sharing this manuscript with me.

Authors extended their previously developed Lagrangian Soil Water and Solute Transport (LAST) model; and presented a novel modeling approach of diffusive pore mixing (DIPMI). This modeling strategy is implemented for simulating reactive solute transport in partially water saturated soil domains. A key development of this modeling approach (with respect to their farmer LAST model) is that DIPMI gets rid of assuming perfect mixing of solutes among water particles. To this end, DIPMI is developed to account for the self-diffusion of water particles across a characteristic length scale of the pore space using pore-size-dependent diffusion coefficients.

Authors tested DIPMI approach to reproduce some experimental findings (from the literature) with diffusive mixing of water isotopes over the pore space of a fully saturated soil volume. They also performed simulation of mixing of a representative solute in a vertical - saturated soil column and compare results of the DIPMI approach against the ones that employed common perfect-mixing assumption.

Authors suggest that imperfect mixing of water and solutes in the pore space can result long tailing of corresponding solute breakthrough curves which agrees with experimental outcomes. Indeed, solute breakthrough curves of the simulations with the LAST approach of perfect-mixing assumption may exhibit clear differences to experimental outcomes.

This work can provide some insights to the simulation of imperfect subscale mixing in a macroscopically homogeneous soil matrix.

The original idea sounds interesting to me, and I believe this paper can be published in HESS with very mild revisions. In particular, I would suggest Author to extend the work by evaluating sensitivity of the simulation responses to the variation of the molecular diffusion coefficient values. Legends and axis of Figure 4 are hardly visible.

The authors have adressed my concerns satisfactorily.  I can recommend the manuscropt for publication.