See attached file.

The objective of the paper is to assess the ability of the ISBA_DF land surface scheme to simulate drainage at the bottom of the soil column by comparing the simulations with data from a set of lysimeters located in two places in France, and having various soil characteristics and vegetation cover. The data are also used to compare various models for the retention and hydraulic conductivity curves used in the modeling.

I had already revised an earlier version of this paper submitted to Hydrology and Earth System Sciences. The authors have addressed the comments I had on their manuscript and I have only a few additional comments to provide.

After reading comments on the earlier version of the manuscript and those already available on the present version, I have the feeling that the authors should better specify the context of their study in the introduction. Their context is the one of Land Surface Models (LSMs) that are applied regionally or at the global scale. This context explains why numerical simulations and parameter specification are generally simplified for the model to be applicable in different contexts, without specific calibration, except in the case of some validations using in situ data at specific sites.

The present study aim is to assess the ability of the current model to simulate accurately groundwater recharge. For that, the model is applied at the local scale, using data from several lysimeter experiments. But this is not the general application context of the model, where soil parameters are computed from soil texture using PedoTransfer Functions (PTF) and the soil profile is assumed homogeneous.

The interest of the study is to show that:

• When parameters and model configuration (in particular soil vertical heterogeneity, but also lower boundary condition) is specified using in situ data, the model performance is satisfactory
• In such configuration, some combinations of soil water retention and hydraulic conductivity models provide better simulations
• Model performance is significantly decreased when vertically homogeneous soil profiles are used or when PTFs are used

The study leads to interesting conclusions with regards to the specification of soil parameters in LSMs that could be better highlighted in the abstract and the conclusions.

Minor comments:

Line 313-314: the sentence is truncated.

Line 548: sensitivity instead of sensibility