|I have reviewed the changes that the authors made, based on my original comments. I believe that the changes to the methodology and the explanation of the methodology are well done. There are improvements to the description of data and metadata used in the study, but I still have a major problem with how some of the data/metadata are used.|
I have extensive personal experience with USGS data collection techniques and don’t believe the authors have a good understanding of the way USGS collects and reports data and metadata, based on the use and interpretation of some of the metadata in their study.
-1- Change of base discharge and change of gauge datum do not indicate any change in flows or in measurement techniques. As I mentioned in my original review, “the ‘base discharge’ is a level set to allow 3-4 peaks per year on average to exceed this level. Instantaneous peaks above this level are then recorded. A ‘change of base discharge’ does not indicate any change in actual flows recorded or any anthropogenic change in watersheds above a gauge. “ This may have been unclear though, so I’ll add some more information. The base discharge level doesn’t affect USGS computation of flows, it only affects what gets reported as a peak above base discharge at a streamflow gauge. These peaks are sometimes used in peak-over-threshold studies (rather than using annual peaks). Change in base discharge should not be used in this study. It doesn’t indicate anything of use for the study and takes away from the other metadata, such as indicated regulation, that do mean something.
-2- Change in gauge datum also does not indicate any change in flows or change in measurement techniques. As I mentioned in my original review “The ‘change of gauge datum’ also does not indicate any change in flows or anthropogenic influence. It indicates only that the arbitrary zero gauge height for the rating curve has been changed, normally because of changes in the gauge control point on a river (the riffle or channel section that controls the relation between river height and flow at a gauge).”
To add more information, change in gauge datum does not represent a discontinuity in measurement techniques or flows. Just the opposite, it means that USGS is continually making flow measurements to verify the rating curve at a gauge. USGS has done this throughout its long history. Many control sections for gauges change because the riffle or channel controlling the relation between river height and flow can change over time. This is due primarily to channel/riffle changes, often caused by high flows. Sand channels are much less stable than bedrock channels. When a gauge datum is changed, it’s because the channel at the gauge (not necessarily other channel sections nearby) changed enough to bring the gauge height below the arbitrary zero point that was established when the gauge was created. The arbitrary zero point is then changed so there won’t be negative gauge heights, necessitating a change in the rating curve. There is no discontinuity of flow magnitudes on either side of this change. There is a discontinuity in the arbitrary height of flow that corresponds to the flows. This is not relevant change for your study. It would be relevant if trends over time in river heights were being studied.
There is a new statement in the report (p. 8, line 19) “Changes in the rating curve used to estimate streamflow from measured water levels are not recorded in the USGS notes but may be a significant source of variation in low flow values that is not accounted for”. This statement is not true. Changes in the rating curve are made to reflect the current relation between stream height and flow. As explained above, USGS regularly checks the rating curve to make sure the current relation is accurate. If the rating curve has changed based on coincident measurements of streamflow and river height, the rating curve is changed. Again this does not represent a discontinuity in the flow data. There would be a discontinuity in the flow data if this was not done regularly.
Change in gauge datum should not be used to classify trends. It is not a meaningful code for the purposes of the current study. I looked at the percentage of gauges noted as having change in datum, for the different categories in Table 1. Most of the categories (no trend, decreasing trend, increasing trend, etc.) have about 13% to 18% of gauges that had changed datums. In other words, all these groups have about the same percentage of gauges with changed datum. The exception is the “Step Change” category with only about 5% of gauges with a change in datum. This is likely explained by the fact that a lot of the change-of-datum gauges are in the mid-Atlantic region where there are a lot of sand channels (Figure 1), but there aren’t many step changes in this area (Figure 2).
This paper would be strengthened by using the metadata in the interpretation of results that has meaning (such as indicators of regulation) and removing ones with no meaning (change in base discharge and change in gauge datum) in all locations in the article. All text, tables, and figures that reference these two metadata codes should be changed to remove them from the study.