This paper is an example of innovative research with a high practical impact on the water resources availability of millions of people. I found it interesting to review this paper. Despite the original thoughts and new insights, some corrections and amendments should be considered. While there is sufficient depth in the statistical analysis, the hydrological analysis requires more elaboration, especially the separation of UIB into different compartments with more local validation of stream flows and ET fluxes. With more proofs that the new rainfall correction makes sense, the whole work will get more attention and international support. The paper can be accepted for publication if the evidences are more localized and supported with other sources of information (local discharge measurements, local P-ET values that make sense, precipitable water column, atmospheric water balance).
The authors used APHRODITE as the main source of rainfall information. In general terms, this rainfall product is not regarded as being accurate. It can certainly not be used as a reference for various type of gridded rainfall products. I advise the authors to include an analysis of CHIRPS data, that with 5 km resolution meet the authors concerns regarding the large cell size of other existing rainfall products. CHIRPS has build in corrections for rain gauge measurements and elevation, and although not verified, I expect this to be more realistic than APHRODITE. The Ph.D. disseration of Cheema from the Delft University of Technology shows that TRMM measurements in the Karakoram requires a 200 mm correction. This level of correction is sufficient for letting P exceed ET values in areas with brooks and gullies, revealing drainage processes. Hence, the local distribution of ET can be used to explain the under-reporting of rainfall (in addition to snowmass).
While UIB may not host flux towers for evapotranspiration, other parts of the Tibetian plateau are equipped with several towers, and these measurements under similar high altitude conditions can be used to validate the evapotranspiration estimates from the four different products. High elevation ET modelling may have its own sources of error, and knowledge on the uncertainty envelope could help to include also the spatially distributed ET data for inverse modelling of rainfall.
The spatial variability of rainfall in UIB is significant, and the authors have made corrections over glacier areas. Did the corrections also include the forests and pastures due to the geo-statistical procedures applied ? The inflow into the main reservoirs (and especially for discharges > 1000m3/s) is governed by the rainfall surplus (P-ET) over alpine forest and pastures. A separated set of gauging stations near glaciers should be used to validate glacier mass balances. This is probably already done, but the results are presented together with lower altitude gauging stations in Figure 9. I am not convinced that the mass balance of glaciers is controlling the inflow in Tarbella and other reservoirs. But if presented in a local context, then it will demonstrate the role of glacier mass balances on runoff.
Another issue requiring more attention is the storage change in the unsaturated zone and in streams and lakes. The annual stream flow Q is not necessarily equal to P-ET+ MB due to storage changes. A longer period of several years is required for letting P-ET+MB coincide with Q.
While the extra precipitation and snowfall might be correct, it should also match with the atmospheric moisture balance. Can this significantly higher rainfall be explained by the advection process ? What is the source of this water ? How does a doubled rainfall regime impact the leeward rainfall patterns ? Is there continuous lateral transport of atmospheric moisture from evaporation of the plain areas, that sustain rainfall and snowfall ? A double amount of rainfall should have a source or origin, and perhaps the atmospheric moisture recycling could explain this process ? Adding some climatic analysis and explain the plausibility of double rainfall will improve the general acceptance of the results.
Where is the boundaries presented in Figure 1 based on ?
The average monthly rainfall and ET values for the various algorithms should be reported in tabular format
Can the elevation be added to Table 2 ?
A map with the location of the discharge gauging stations should be added
The abstract should more reflect the average condition (double rainfall and snowfall), and not focus too much on the extreme corrections of a factor 10. This is somewhat misleading
The MODIS data sets also contain a number of atmospheric parameters, including precipitable water amounts. Analysis of the atmospheric composition could facilitate the existence of higher rainfall intensities in certain areas.
Figure 10 is a graphical confirmation of the inconsistency in mass balance discussed before. It does not contribute to the overall findings of the paper.