|1. Authors argued about trends in tank water extent that could not be explained by precipitation (page 7 line 28-29) without providing any evidences. The manuscript does not provide any time series plot of summarized tank water extent extracted from LS by the tank cluster in comparison with rainfall. A quick search reveals a paper by Suresh et al., 2010 that discussed inter-annual variability in rainfall around the Arkavathy watershed. If we assume a similar rainfall regime over Arkavathy, it resembles the tank water extent shown for at least 2 tanks in supplemental figure 4 and 5 for 1994, 2002, 2007 and 2008. Rainfall was well below the mean in 1994 and 2002 corresponding to the lowest water extent while the 2007 and 2008 had good rainfall also reflected in the increased water extent for those tanks compared to 1994 and 2002. It is possible that human activities are playing major role in declining tank water extent, but the temporal patterns also need to be evaluated with respect to changes in precipitation. Here is another example by Subash et al., 2014 where it shows rainfall is declining in Karnataka.|
2. In validation: adding analysis with respect to google earth images is good, however it provides some confusion, the max water extent shown in supplemental figure 6 does not in line with max extent shown in figure 6 in the main text. So are two sets of tanks presented in these two figures totally different? Speaking of water extent area, unlike what is mentioned in page 10 ln 19 and 22, 25 ha and 2.5 ha are equivalent to 276 and 26 pixels according to 30 m pixel size as suggested in Table 1 and section 2.2. Which one is correct? 25 ha or 27.8 pixels? if latter is true then i think it requires finer resolution than 30 m to effectively identify tanks of that size from the satellite image.
3. Why the image water extents were not compared against the tank area computed from the shapefile of KSRSAC or tank area from the topographic maps where possible? I would expect, the post-monsoon tank water extent would resemble the maximum water extent possible for each tank as it can be expected to be full after monsoon period.
4. The manuscript did not provide proper definition of what they mean by pre-monsoon, monsoon, post-monsoon, wet-monsoon, normal-monsoon (fig S6), dry season, monsoon year (page 8 line 18), normal year, wet year (page 10 line 26-27). In southern India there are two different monsoon seasons; southwest and northeast with two rainfall peaks june-july for southwest monsoon and October for northeast monsoon. The above mentioned terms are inconsistently used throughout the manuscript.
5. In the entire manuscript it was never mentioned what was the total water extent derived from images, all the tanks or by tank cluster for each year. Seems like it would be really very small.
6. Page 4 line 16-17 suggest study focus on Dec and Jan tank water extent estimates, however it is not clear how many data points were used in running the trend analysis. If only Dec and Jan estimates were considered then there would be only 18 data points between 1973 and 2010. Is it enough for a trend analysis? I do not think so.
7. Page 8 line 11-12, surface water extent was strongly related to precipitation metrics, what does this mean?
8. Page 8 line 31-34, how could the authors confirm tank water storage which is the volume of water dynamics from surface water extent which is an area.
9. Page 11 line 25-26, another example of no evidence. No observed streamflow data or analysis was conducted, but conclusion was drawn for observed streamflow. Throughout the manuscript the satellite derived extracted water extent is used vaguely as proxy to tank water storage and streamflow, which to me is inappropriate and incorrect.
10. Again out of context discussion of water table and drilling of wells in Page 11 line 26-30.
11. Page 12 line 4-7, no supporting evidence provided.
12. Page 12 line 19-25, if I understand correctly, what was done is, extract tank surface water extent area from satellite images over time and analyze the change that is present in the surface water extent area, then try to correlate these changes with changes in land use. There were no analyses done with streamflow or tank storage, now the authors started discussing about streamflow and tank storage.
13. I find the discussion section 4.1 incoherent given the analyses done. The discussion is all about streamflow while the analyses was on tank surface water extent extraction.
14. Figure 6 last line, that is why more than one index/method is needed to accurately identify tanks from multi-date satellite images.
15. The entire conclusion is incoherent and not supported by circumstantial evidence.
16. What is this supplemental figure 9 for? How do the lines drawn? Why the data points for the available images for December not shown in this plot?
17. Again what is this supplemental figure 7, 8 and 10 for? No mention of this figure in the main text. Graphs show the data and circumstantial evidence needed to support any conclusion or statement made in the text.
18. Very poor figures and maps, especially Figure 1, 4 and 7, no coordinates/index maps.
I think when reviewers sought major revision; it means there are substantial concerns about the method and results presented in the manuscript. During the revision those concerns need to be reevaluated and appropriate changes in the method or analysis needed to be made. In my opinion after the first major revision the flaws in the method and discussion were not addressed adequately which led to more incoherent discussion in the manuscript. Therefore I apologize for not being able to recommend this manuscript for publication.
Suresh, H. S., H. S. Dattaraja, and R. Sukumar. "Relationship between annual rainfall and tree mortality in a tropical dry forest: results of a 19-year study at Mudumalai, southern India." Forest Ecology and Management 259.4 (2010): 762-769.
Subash, N., and B. Gangwar. "Statistical analysis of Indian rainfall and rice productivity anomalies over the last decades." International Journal of Climatology 34.7 (2014): 2378-2392. (look for Table 5, data period 1974-2009)