the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Scaling and intermittent properties of oceanic and atmospheric pCO2 time series and their difference
Abstract. In this study the multi-scale dynamics of 38 oceanic and atmospheric pCO2 time series from fixed Eulerian buoys recorded with three-hour resolution are considered. The difference between these time series, the sea surface temperature and the sea surface salinity data were also studied. These series possess multi-scale turbulent-like fluctuations and display scaling properties from three hours to the annual scale. Scaling exponents are estimated through Fourier analysis and their average quantities considered globally for all parameters, as well as for different ecosystems (e.g. coastal shelf, coral reefs, open ocean). Sea surface temperature is the only parameter for which a spectral slope close to 5/3 is found, corresponding to a passive scalar in homogeneous and isotropic turbulence. The other parameters had smaller spectral slopes, from 1.18 to 1.35. By using empirical mode decomposition of the time series, together with generalized Hilbert spectral analysis, the intermittency of the time series was considered in the multifractal framework. Concave moment functions were estimated and Hurst index and intermittency parameters estimated in the framework of a lognormal multifractal fit. It is the first time that atmospheric and oceanic pCO2 and their difference ∆pCO2 are studied using such intermittent turbulence framework. The ∆pCO2 time series was shown to possess power-law scaling with an exponent of β = 1.32 ± 0.2.
- Preprint
(11829 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 10 May 2024)
-
RC1: 'Comment on npg-2024-7', Anonymous Referee #1, 23 Apr 2024
reply
The manuscript presents an analysis of the scaling properties of 38 oceanic and atmospheric pCO2 time series at 3-hr resolution to uncover similarities and differences between different ecosystems, as well as, to shed new insights on the role of turbulence (both of passive and active nature). The paper is well-written and also logically organized. It is appropriate for the scope of NPG. However, at the present stage I would not recommend it for publication but I would suggest for major revisions as outlined below.
Major comments
- The authors firstly present results based on Fourier PSD for spectral exponents, while later they use EMD/HSA to provide more insights on high-order statistics, as well as, to also reduce effects of periodicity in time series that could destroy scaling behavior. Why not to directly use EMD/HSA for also investigating spectral slopes by using the second-order moment of the generalized Hilbert spectra? This would directly overcome limitations provided by Fourier PSD. Concerning the second-order moment it is expected to observe an agreement between the spectral slope evaluated via Fourier PSD and those evaluated via EMD/HSA. This agreement seems to be missed if one looks at the second-order exponents for SSS in Figure 8 for Gulf of Maine and especially for pCO2 air time series.
- Concerning the scaling exponents of pCO2 air time series it seems that, especially for Gulf of Maine, a plateau is reached for high-order statistics that could be related to missing statistics for proper evaluation or to the choice of the range of scales where scaling exponents are evaluated. This needs to be fixed or explained.
- One of the main result is that multifractal intermittency is shown for the first time for oceanic and atmospheric pCO2 air time series. However, by looking at Table 5 I would say that intermittency is almost zero, considering average values of the µ parameter and its error range. Could this be related to the choice of the multifractal model, i.e., the log-normal one?
- A general comment on Figs. 4, 5, 7, 8: it would be desirable to add error bars on the estimated quantities (exponents) since this would allow directly to see if they are really different or can be comparable in the range of uncertainty. Furthermore, I would suggest to add in Fig. 4 both errors on ß as well as the indication of the range of frequencies where the spectral law is evaluated. The latter can be really useful to estimate the goodness of the fit, especially if looking at lower-middle panel when peaks appear at high frequencies.
- Figure 7. More than a comment it is a suggestion. It is an interesting result that seems to suggest that something occurs at northern mid latitudes. What about to see if there are variations in the slope of the low frequency regime that could be related to some large-scale forcing affecting that region? What about similar analysis for SST and SSS?
Minor comments (line by line)
- Line 16: "mitigated" seems to not be appropriate.
- Line 26: please clearly state which temporal and spatial scales are referring to.
- Line 29: "flux" -> "mixing"?
- Eq. (3): does it hold for current measurements? Is ß a function of the depth and of the frequency? Please clarify.
- Lines 124 and 125: "Pikes" -> "Peaks"?
- Figure 6: what are the dots?
- Line 199: linear interpolation could introduce some kinds of spurious high-frequency intermittent-like bursts where it is performed. Is it the case? If yes, these time intervals are removed for the evaluation of slopes?
- Eq. (6): there is a missing "Principal Value".
- Line 234: I do not see this agreement for pCO2 air time series (see Major Comments 1 and 2).
Citation: https://doi.org/10.5194/npg-2024-7-RC1
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
270 | 29 | 6 | 305 | 4 | 4 |
- HTML: 270
- PDF: 29
- XML: 6
- Total: 305
- BibTeX: 4
- EndNote: 4
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1