Please find the review comments on the attached file

This manuscript presents an ambitious global modeling framework, describing terrestrial water and energy processes using many analytical solutions. It is a valuable modeling exercise. Reviewer #1 provided nice, detailed comments. Here my comments are relatively high level. 

1. "Calibration-free" in the title is misleading or at least questionable. In fact, the authors used the words "calibrate" and "calibration" several times in their own method description, e.g., L390, L403, L425, L439. Perhaps, "parameter-parsimonious" is more appropriate. 

2. Using analytical solutions to describe water/energy processes in global LSMs is attempting, but may not be feasible for all the processes that are necessary in LSMs. Most analytical solutions are based on strong assumptions, and these assumptions may be valid at certain locations or time scales, but not at others. For example, the Green-Ampt equation was originally derived as a point-scale theory, hence a key assumption underpinning it is the (horizontal) homogeneity of soil properties and other relevant variables. Here "point scale" is roughly at the order of 1-10 square meters. When applied globally, each spatial unit (say a lat/lon grid) is much larger than 1 square kilometer and thus contains strong spatial heterogeneity. It is quite often that, within the same lat/lon grid, at the same time, there are patches already saturated, and others are unsaturated. Whether the Green-Ampt equation can be applied at such a spatial scale is highly questionable. The runoff scheme used in the VIC model, nonetheless, is not empirical but at least partially physically based. It explicitly embraces the spatial heterogeneity of soil moisture deficit levels in a spatial unit, hence theoretically more suitable to be used in LSMs for regional or global applications. The authors should at least point out these theoretical limitations (hence potential limitations in SPLASH applications) in their introduction and discussion. 

3. More clarifications need to be provided on the comparison between SPLASH and VIC-3L at the global scale. Are they applied at the same spatiotemporal resolutions? Has VIC-3L been properly spun up and calibrated? Why not use more hydrologic data to compare these two models, e.g., observed streamflow data? 

4. Many observed data are already available globally in a spatially distributed fashion (e.g., SWE). Showing the comparisons in the form of spatial maps might be more informative than climatic zones. For global LSMs, one main objective is to capture the spatial variability in the variables of interest.