SEIA: a scale-selective eddy identification algorithm for the global ocean
- 1State Key Laboratory of Tropical Oceanography (LTO), South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, 510301, China
- 2University of Chinese Academy of Sciences, Beijing, China
- 3Southern Marine Science and Engineering Guangdong Laboratory (Guangzhou), Guangzhou, China
- 1State Key Laboratory of Tropical Oceanography (LTO), South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, 510301, China
- 2University of Chinese Academy of Sciences, Beijing, China
- 3Southern Marine Science and Engineering Guangdong Laboratory (Guangzhou), Guangzhou, China
Abstract. Automatic eddy identification algorithms are crucial for global eddy research. This study presents a scale-selective eddy identification algorithm (SEIA; https://github.com/Yk-Yang/SEIA) for the global ocean based on closed sea level anomalies (SLAs) that features two improvements in the detection and tracking processes. First, the scale-selective scheme replaces the previously used threshold for defining the eddy boundary and restricts the numbers of upper and lower grid points based on the data resolution and eddy spatial scale. Under such conditions, the eddy boundary is as large as possible, while the eddy region is not overestimated. Furthermore, a novel and effective overlap scheme is used to track eddies by calculating the intersecting ratio of time-step-successive eddies. SEIA generates 278,630 anticyclonic eddies and 274,351 cyclonic eddies from AVISO’s SLA dataset over a five-year period (2015–2019; http://www.doi.org/10.11922/sciencedb.o00035.00004; Yang et al., 2022). The global distribution of eddies, propagation speed, and eddy path characteristics in the Southern Ocean verify the validity of SEIA.
Yikai Yang et al.
Status: open (extended)
-
RC1: 'Comment on essd-2022-77', Anonymous Referee #1, 03 Mar 2022
reply
The manuscript is lack of innovations and highlights. It attempts to make improvements for eddy identifying and tracking, including the scale-selective scheme in eddy detection and the overlap scheme in eddy tracking. But there are still some major parts need to be improved in the paper. The suggestions are as follows:
- The scale-selective scheme corresponds to the radius of eddies ranging from 25 to 125 kmin the paper, which is incomprehensible. The formula as Line 161 shows present the scale-selective scheme, which means that you still select eddy contour by the threshold (Pmin-Pmax). However, the scale of eddy is changed with latitude, which should be considered in this paper.
- The validation part is not sufficient. Other source of data should be considered in the validation, like remote sensing data (sst, sss or oceanic chlorophyll) and in situ data (drifter or argo).
- In this paper, the data set of eddy during 2015-2019 are detected and tracked. The SLA data set from 1993 is available, which should be adopted to analysis. Meanwhile, the systematic comparison between eddy dataset in this paper and existing eddy datasets (like [1,2]) should be conducted in this work.
The manuscript is written very carelessly with many errors and unclear places. The level of English (grammar, style and syntax) throughout the manuscript does not meet the journal's required standard. I suggest rejecting the manuscript.
[1] Faghmous, J. H. et al. A daily global mesoscale ocean eddy dataset from satellite altimetry. Sci. Data 2:150028 doi: 10.1038/sdata.2015.28 (2015).
[2] Chelton, D., Schlax, M. & Samelson, R. Global observations of nonlinear mesoscale eddies. Prog. Oceanogr. 334, 328–332 (2011).
-
RC2: 'Comment on essd-2022-77', Anonymous Referee #2, 18 Mar 2022
reply
Nowadays, researches towards eddy are highly connected to the automatic eddy identification algorithms. The paper presents a new one based on the SLA dataset and rule of scale-selective, which is well-organized and in a good theme. The authors highlight two improvements featuring in the detection and tracking processes when compared to the previous methods. The apply of the new algorithm verify its validity.
Still, there were some questions which came to my mind while reading the paper and hence I suggest the authors to put some more effort and improve this work. There are several things need to be clarified before publishing, particularly about the preset parameters of the new algorithm. Therefore, I recommend this manuscript to be minor revision.
- In the section 2.2, the authors mentioned that “we consider only the simple contour condition with only one core (Fig. 1b): the concentric (Fig. 1c) and intersecting (Fig. 1d) closed types degrade to this type.” How is this “degrade” achieved? More explanation is needed.
- Among the process of “(i) Searching for SLA peaks”, a moving 3 3 grid window is addressed to search peaks of SLA. In Fig. 2a, there are so many peaks occurring nearshore. Is it rational, and how did the authors remove them? Also in this process, the authors mentioned that “SLA shallower than certain depth (50 m for SEIA) will be masked” because of the AVISO nearshore issue. Why did the depth set as 50 m. Any references?
- In the determination of the parameter Dt, the intersecting ratio, the authors illustrated that 0.39 is too ideal for a real scenario and set to 0.25. The value of 0.25 is based on what kind of consideration? There is no good explanation.
- “…when an eddy has a state of “-tracked-missing-” in two continuous time steps, all the missing information is temporally replaced with the former tracked information and seen as a complement state, allowing the tracking procedure to continue.” Is it reasonable to base eddy tracking on this complement eddy?
Yikai Yang et al.
Data sets
Five years (2015-2019) of global eddy product from SEIA Yikai Yang, Lili Zeng, Qiang Wang http://www.doi.org/10.11922/sciencedb.o00035.00004
Model code and software
A MATLAB- and SLA- based scale-selective eddy identification algorithm Yikai Yang, Lili Zeng, Qiang Wang https://github.com/Yk-Yang/SEIA
Yikai Yang et al.
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
255 | 51 | 10 | 316 | 4 | 5 |
- HTML: 255
- PDF: 51
- XML: 10
- Total: 316
- BibTeX: 4
- EndNote: 5
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1