|Review of: “Contextualizing time-series data: Quantification of short-term regional variability in the San Pedro Channel using high-resolution in situ glider data” by Teel et al. for Biogeoscience”|
This is my second review of the manuscript by Teel and Coauthors. As stated in my initial review, the Authors describe a nice new dataset of glider observations that they then use to investigate the representativeness of the SPOT time-series dataset. My main criticisms were: the lack of use of actual SPOT data, a confusing description of the PCA end member-definition, and a weak comparison with satellite-based chlorophyll observations. I am pleased to see that the Authors have taken these and another Reviewer’s comments seriously, and addressed them in revision. I appreciate the reorganization of the old figures, and a few new ones, and I find the revised manuscript clearer and easier to understand. The conclusions appear to be supported by the data, and there are a number of insights that should make the study valuable for the observational and modeling communities. Likewise, the glider observations will likely prove useful for anyone interested in the oceanography of the Southern California Bight, and in particular in the rich, long running SPOT dataset. I have a few minor comments that the Authors could further address; besides that, I am overall supportive of publication.
I would still carefully review the presentation of the two-step approach in the PCA analysis, to make sure it comes across as clear for the first-time reader. For example, in Section 2.4, the sentence starting in line 144 could further clarify the reason for a second PCA, the rationale for the 54 profiles selected, etc. Alternatively a reference could be added pointing to section 3.2, where more details are provided. Section 3.2 as well could benefit for an opening summary sentence, and could be clearly broken into two parts, one for the “original” PCA and the second for the “structured” PCA.
Figure 8, in particular panel (a) could be clarified. At first, I was confused by the number of dots of different colors, and it took me a while to figure out that longer re-sampling times increase the number of “sub-datasets” that can be extracted from the full timeseries, thus producing multiple dots. This could be explained in the caption. The Authors could also clarify in a sentence of two of text why the mean, S.D., and distance in PCA space are good measures of the ability of a sampling frequency to capture real variability. This is not completely obvious, given how abstract the PCA space is. I wonder if there is a better way to think about the impact of sub-sampling. For example, would a less frequent sampling strategy be able to recover the full end-members of the PCA (which include more rare, extreme states), rather than the mean state?
Abstract, line 22: I would change “was dominated by” with “could be described by a combination of” (or similar). Since most profiles are a combination of end members, none of the end members truly “dominates”.
Line 207: “meaningful distinction” seems vague; also what kind of “analysis”? Please be more specific.
Lines 222-227. I am still not convinced about the WSHC not being a true end member. After all, it is a real extreme in PCA space, and has warmer temperatures everywhere compared to other end members. Also, the deep chl maximum at ~30 m of WLC seems quite shallower than where I would expect a subtropical DCM (40-60m offshore the SCB). Finally, to the mechanisms listed in lines 224-227 for the DCM of the WSHC, I would add elevated NPP at depth along waters with high subducted NO3 (that is, nutrients, rather than just phytoplankton are subducted, stimulating growth at depth).
Lines 342, 344: “correlation” rather than “relationship”?
Line 388: “Significant” has a statistical meaning that doesn’t seem to apply to the eample of year 2001, which doesn’t seem statistically different from the mean or the other years. Maybe use “Larger”, or “Stronger” instead?
Please re-read carefully to fix some typos.