Method for synergistically regulating and controlling chlorophyll-a concentration variability through vortex-upwelling

Abstract

The invention provides a method for cooperatively regulating and controlling chlorophyll-a concentration variability through vortex-upwelling, and relates to the technical field of ocean remote sensing and environmental monitoring. A multi-source satellite and reanalysis data set is obtained, and a stationary time sequence is constructed; identifying a dominant variation period in chlorophyll-a concentration, sea surface temperature and sea surface height anomalies, and extracting a monthly abnormal value sequence; performing complex empirical orthogonal function analysis on the sea surface height anomaly, extracting a dominant spatial variation mode of the sea surface height anomaly, and identifying a vortex distribution position and form; classifying and screening target vortexes according to vortex types and seasons, carrying out normalization processing on the chlorophyll-a concentration, aligning to vortex centers, and analyzing a spatial response relationship between the vortexes and the chlorophyll-a concentration; and drawing a density time-depth profile map and a CHL-SLA composite map by combining surface chlorophyll-a concentration, sea surface height anomaly and vertical Argo buoy data, and analyzing an influence mechanism of vortex-upwelling on chlorophyll-a horizontal distribution.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ocean remote sensing and environmental monitoring, in particular to a method for vortex-upwelling collaborative regulation of chlorophyll-a concentration variability. BACKGROUND

[0002] In the past two decades, the driving mechanism of chlorophyll-a concentration (CHL) variability in the Somali upwelling region has been a research hotspot in the field of ocean science, but there has been controversy. Previous studies have shown that factors such as increased wind stress, sea surface temperature changes, and mesoscale vortex structure have affected the spatial and temporal distribution of CHL in this region to varying degrees. However, the interaction mechanism between these factors is still unclear, especially the collaborative regulation of vortex and upwelling under the background of strong monsoon, which still lacks systematic quantitative research. Therefore, in-depth exploration of the dynamic process and spatiotemporal evolution characteristics of CHL variability has become the key to understanding the changes in regional ocean primary productivity.

[0003] Traditional remote sensing analysis methods have limitations in analyzing the coupling characteristics between mesoscale eddies and upwelling, making it difficult to fully capture their nonlinear propagation characteristics and spatial response patterns. Therefore, there is an urgent need to develop a new research framework based on multi-source data fusion and multi-temporal and spatial scale analysis methods to identify the collaborative action of vortex-upwelling and understand the process mechanism. SUMMARY

[0004] To solve the above technical problems, the present application proposes a method for vortex-upwelling collaborative regulation of chlorophyll-a concentration variability, which integrates wavelet and complex empirical orthogonal function (CEOF) collaborative mode extraction, chlorophyll-a concentration (CHL) anomaly analysis, vortex center CHL fusion, and horizontal-section data joint analysis methods, aiming to reveal the regulation mechanism of mesoscale eddies (especially anticyclones such as the Great Whirl, GW) and monsoon-driven coastal upwelling on the spatiotemporal distribution of CHL.

[0005] A method for vortex-upwelling collaborative regulation of chlorophyll-a concentration variability, comprising the following steps: (1) Obtain multi-source satellite and reanalysis datasets, including chlorophyll-a concentration (CHL), sea surface temperature (SST), sea level anomaly (SLA), wind speed, and Argo float data. Time fuse daily data of chlorophyll-a concentration, sea surface temperature, and sea level anomaly to generate monthly mean data, and construct a smooth time series; (2) Perform continuous wavelet transform on the smooth time series to identify the dominant variability period of chlorophyll-a concentration, sea surface temperature, and sea level anomaly. Use multiple seasonal-trend decomposition methods to separate long-term trend, seasonal period, and residual components, and extract monthly anomaly value sequences; (3) Applying complex empirical orthogonal function analysis to sea level anomaly, extracting its dominant spatial variation mode, identifying the position and shape of eddy distribution; (4) According to the classification of eddy type and season, the target eddy is screened, the chlorophyll-a concentration is normalized and aligned to the eddy center, and the spatial response relationship between eddy and chlorophyll-a concentration is analyzed; (5) Combined with surface chlorophyll-a concentration, sea level anomaly and vertical Argo float data, density time-depth profile and CHL-SLA composite map are drawn, and the influence mechanism of eddy-upwelling on chlorophyll-a horizontal distribution is analyzed.

[0006] Preferably, in step (3), the complex empirical orthogonal function analysis simultaneously analyzes the amplitude and phase information of sea level anomaly variation, and identifies the mesoscale eddy structure.

[0007] Preferably, when identifying the mesoscale cyclone and anticyclone eddy structure, the eddy center and polarity are determined by calculating the gradient and Hessian matrix of sea level anomaly, the eddy boundary is determined by finding the maximum closed contour, and the equivalent radius of the eddy is calculated.

[0008] Preferably, in step (4), the chlorophyll-a concentration is normalized and aligned to the eddy center, and is standardized according to the equivalent radius range of the eddy, so as to analyze the spatial distribution characteristics of chlorophyll-a concentration inside and around the eddy.

[0009] Preferably, in step (5), the horizontal-section data joint analysis includes: for Argo float trajectory, density time-depth profile is drawn, daily chlorophyll-a concentration and sea level anomaly corresponding to time and position are obtained, CHL-SLA spatial composite map is generated, Argo float trajectory is superimposed on CHL-SLA spatial composite map, and the influence of eddy-upwelling system on chlorophyll-a concentration horizontal distribution is comprehensively analyzed.

[0010] Preferably, when performing horizontal-section data joint analysis, Argo float profile data is used, which includes temperature, salinity, pressure and density, and mixed layer depth MLD is calculated according to density method: ; Wherein, z represents the depth, ρ(z) represents the density of seawater at depth z, ρ(z0) represents the density of reference depth, h represents the threshold.

[0011] Preferably, in step (2), continuous wavelet transform is used to extract periodic and cyclic components from stationary time series, and the data is decomposed into time-frequency space. For time series with uniform time step time series x n Wavelet coefficients, n=1,...,N for: ; Among them, wavelet coefficients This represents the time series at time point n and scale s. With wavelet basis functions The degree of similarity, For wavelet power spectrum, For the index of the summation variable.

[0012] Preferably, identifying cyclones and anticyclones through sea surface height anomaly contour lines and extracting vortex polarity, center position, and radius includes the following steps: ① Polarity determination: ; Indicates the center point of the vortex SLA scalar field, denoted by , H represents the Hessian matrix, and detH represents the determinant; Polarity determination: When the trace of H tr H < 0, it is a local maximum, indicating an anticyclone; when the trace of H tr H > 0, it is a local minimum, indicating a cyclone. ② Construct closed contour lines: Let S be the set of saddle points, and let s be the extreme values ​​of the boundary contour lines. ; For the center point of the vortex The SLA scalar field of the connected saddle points s; Boundary equivalence of the anticyclone center for: ; cyclone center boundary isopleth for: ; The vortex boundary is a closed contour line. : ; For internal connected components, Indicates boundary equality. Points on a closed contour line SLA scalar field; ③ Calculate the center point of the vortex : ; ④ Calculate the area A and perimeter L of the closed domain: , ; where a represents an area element of the internal connected domain, represents a unit length on the closed contour line.

[0013] Using the equal-area radius as the vortex radius : .

[0014] Compared with the prior art, the present application has the following beneficial effects: The present application proposes a comprehensive method combining wavelet analysis, modal extraction, anomaly identification, vortex center complexation and horizontal-section data joint analysis, which is used for revealing the action mechanism of mesoscale vortex and upwelling on the distribution of CHL in the sea. The present application can effectively reveal the coupling relationship between physical processes and biological responses by introducing Argo buoy section data to analyze the comprehensive influence of vortex and upwelling on MLD, SLA and CHL concentration. The present application has significant innovation and advantage in multi-source data fusion, complex dynamic process separation and ecological effect identification, and can more comprehensively and accurately depict the regulation mechanism of vortex and upwelling on the marine ecological environment. The method of the present application has good universality, can be applied to other typical upwelling and dynamic active sea areas in the world, and provides new technical means and scientific basis for marine ecological environment monitoring, resource assessment and climate change impact research. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 is the monthly time series data from 1997 to 2020, wherein, Figure 1 (a) is the time series graph of CHL from 1997 to 2004, Figure 1 (b) is the time series graph of SLA and SST from 1997 to 2024, Figure 1 (c) is the time series graph of the horizontal and vertical components of wind speed from 1997 to 2020, Figure 1 (d) is the time series graph of wind stress from 1997 to 2020; Figure 2 is the correlation relationship between monthly CHL and SLA, SST, wind speed U component and wind speed V component from 1997 to 2024; Figure 3 is the monthly average time series data of CHL, SLA and SST from 1997 to 2024, wherein Figure 3 (a) is the time series graph of monthly average CHL, Figure 3 (b) is the time series graph of monthly average SLA, Figure 3 (c) is the time series graph of monthly average SST; Figure 4Time series and energy spectrum of CHL, SST and SLA for the period 1998-2024, wherein Figure 4 (a)(d)(g) are time series of interannual monthly CHL, SST and SLA, Figure 4 (b)(e)(h) are energy spectrum plots of CHL, SST and SLA, Figure 4 (c)(f)(i) are average spectrum plots of CHL, SST and SLA; Figure 5 Scatter plots between monthly CHL anomalies and SST and SLA anomalies for the period 1997-2024, wherein Figure 5 (a) is a CHL-SST scatter plot, Figure 5 (b) is a CHL-SLA scatter plot, Figure 5 (c) is a SLA-SST scatter plot; Figure 6 SLA first mode in CEOF analysis; Figure 7 Spatial distribution of cyclonic and anticyclonic eddy frequency, wherein Figure 7 (a) is a spatial frequency distribution plot of cyclonic eddies; Figure 7 (b) is a spatial frequency distribution plot of anticyclonic eddies; Figure 8 CHL fusion plot at eddy center; Figure 9 Monthly mean MLD spatial distribution plot; Figure 10 Spatial distribution of monthly mean MLD in anticyclonic and cyclonic eddies; Figure 11 Time series of monthly mean MLD in anticyclonic and cyclonic eddies; wherein Figure 11 (a) is a time series of monthly mean MLD in anticyclonic eddies; Figure 11 (b) is a time series of monthly mean MLD in cyclonic eddies; Figure 12 Density time-depth profile plot based on Argo float trajectory and surface CHL and SLA spatiotemporal distribution plot. DETAILED DESCRIPTION

[0016] The application will be described in greater detail below with reference to specific embodiments. The following examples are helpful for those skilled in the art to further understand the application, but do not limit the application in any form. It should be noted that for those skilled in the art, without departing from the concept of the application, a number of changes and improvements can be made. These are within the scope of the application.

[0017] Example 1 1. Multi-source remote sensing and reanalysis data acquisition The present application adopts multi-source satellite and reanalysis datasets, in which the chlorophyll-a concentration (CHL) comes from the Ocean Color Climate Change Initiative (OC-CCI) of the European Space Agency, which is a three-level gridded product with a daily time resolution, a 4-kilometer spatial resolution, and a coverage period from 1997 to 2024; the sea surface temperature (SST) comes from the AVHRR remote sensing product of the National Environmental Information Center (NOAA NCEI), which has the same time and spatial resolution as CHL, with a time span from 1981 to 2023; the wind speed data is based on the ERA-Interim reanalysis product of the European Centre for Medium-Range Weather Forecasts (ECMWF), which selects the wind speed information at sea level 10 meters high, with a daily time resolution, a 0.12° spatial resolution, and a coverage period from 1997 to 2020; the sea level anomaly (SLA) comes from the Copernicus Marine Service, which is a four-level fusion product with a daily time resolution and a 0.25° spatial resolution, with a time range from 1993 to 2024; the Argo buoy data comes from the International Argo Data website, with a time resolution of about 10 days, a spatial resolution of 3°, and a time range from 1998 to 2025.

[0018] 2. Wavelet analysis and trend decomposition Firstly, the daily data of CHL, SST and SLA are time-fused to generate corresponding monthly average data. Then, the monthly data is de-trended to construct a stationary time series for subsequent spectral analysis. Based on the time series, the continuous wavelet transform (CWT) method is used to identify the dominant variation periods in CHL, SST, SLA and other variables, such as annual cycle, semi-annual cycle and short-term sudden fluctuations.

[0019] On this basis, further use the multiple seasonal-trend decomposition method (MSTL) to separate the long-term trend, seasonal cycle and residual components in the time series. This method can effectively strip the seasonal signal and extract monthly anomalies with independent significance. The final anomaly sequence can be used to analyze the correlation between CHL, SST and SLA, and provide theoretical support for subsequent collaborative regulation mechanism research.

[0020] 3. CEOF analysis and mesoscale eddy identification The CEOF analysis method is applied to SLA to extract its dominant spatiotemporal variation mode. Compared with the traditional EOF method, CEOF can simultaneously analyze the amplitude and phase information, and is suitable for revealing the propagation characteristics of mesoscale dynamic processes, especially having significant advantages in identifying rotating structures (such as eddies).

[0021] The first principal mode extracted by CEOF decomposition can effectively capture typical mesoscale anticyclonic structures in the Somali upwelling region, especially the persistent and large-scale GW. Simultaneously, this mode also reveals a hook-shaped transport path extending from the coast to the open ocean, reflecting the horizontal diffusion process of coastal upwelling under the guidance of the anticyclone.

[0022] 4. CHL Composite Analysis Method with Vortex Center This composite analysis method can systematically characterize the regulatory effects of cyclones and anticyclones on CHL concentration, providing important support for understanding the impact mechanism of mesoscale dynamic processes on surface primary productivity.

[0023] Target vortices were categorized and screened according to vortex type (cyclone / anticyclone) and season (June-September) to obtain their center coordinates and equivalent radii. For each vortex, normalization was performed, aligning the CHL image to the vortex center and standardizing it according to the vortex radius range. The spatial distribution characteristics of CHL for each type of vortex in different months were statistically analyzed and overlaid with SLA contour lines to analyze the spatial response relationship between vortices and CHL. The specific steps of this method are as follows: (1) Data acquisition and preprocessing Level 4 fused SLA data was acquired from the Copernicus Ocean Service, with a daily temporal resolution, a spatial resolution of 0.25°, and a time range of 1993–2024; this data was used for eddy identification. Simultaneously, Level 3 CHL data was acquired from the European Space Agency's Ocean Color Climate Change Programme, with a daily temporal resolution, a spatial resolution of 4 km, and a time range of 1997–2024. From the daily CHL data, monthly fused CHL data was first calculated. Then, the average of the monthly fused data for all years was calculated, and this average was subtracted from the monthly fused CHL data to obtain the monthly CHL anomaly data. Finally, the average of the monthly CHL anomaly data for all years was calculated to obtain the monthly CHL anomaly data.

[0024] (2) Vortex recognition and feature extraction Cyclones and anticyclones are identified using SLA contour lines, and vortex polarity (cyclone / anticyclone), center position, and radius are extracted. The specific calculation process is as follows: ① Polarity determination: ; Indicates the center point of the vortex SLA scalar field, Indicates the center coordinates, denoted by , H represents the Hessian matrix, and detH represents the determinant.

[0025] Polarity: When trH (trace of H) <0, local maximum, anticyclone; when trH >0, local minimum, cyclone.

[0026] 2. Boundary contour (maximal closed contour): Let S be the set of saddle points, and s be the extreme value of the boundary contour, which makes detH <0. s is a saddle point in the set S .

[0027] SLA scalar field of the relevant saddle point s connected with the vortex center point .

[0028] Anticyclone (center maximum): ; Boundary contour of the anticyclone center.

[0029] Cyclone (center minimum): ; Boundary contour of the cyclone center.

[0030] Vortex boundary is the corresponding closed contour: = interior connected domain; denotes the closed contour, denotes the boundary contour, SLA scalar field of the point on the closed contour.

[0031] 3. Calculate the vortex center point : ; 4. Radius (given by the closed domain geometry): Area and perimeter of the closed domain: , ; A represents the area, a represents the area element of the interior connected domain, L represents the perimeter, denotes the unit length on the closed contour.

[0032] Use the equal-area radius as the vortex radius: ; denotes the vortex radius.

[0033] 5. Horizontal-section data joint analysis method To clarify the influence mechanism of eddy-upwelling on CHL horizontal distribution, surface CHL and SLA data, and vertical Argo float data were used. For a specific Argo float trajectory, the density-time-depth profile was plotted using the Argo float pressure and density data. At the same time, the daily scale CHL and SLA data corresponding to each Argo float data were obtained. Thus, these data were compared. In addition, the starting time of this specific Argo float trajectory was found, and the composite graph of CHL and SLA on that day was plotted, and the trajectory of the Argo float was superimposed on the graph. From the graph, the horizontal distribution of CHL in the eddy and upwelling area, and the trajectory distribution of the Argo float can be seen.

[0034] Example 2 (I) Study area and data processing This embodiment takes the Somali upwelling area as the research object, and the time period is the southwest monsoon period (June-September) from 1997 to 2024. Based on OC-CCI, AVHRR, ERA-Interim and Copernicus Marine Service satellite and reanalysis data, CHL, SST, wind speed and SLA variables are obtained, and the CHL composite graph of the study area during the monsoon period is obtained.

[0035] It is found by superimposing the SLA contour lines that the coastal water body shows negative SLA and CHL concentration higher than 1 mg / m³, indicating the existence of upwelling-driven nutrient enhancement. In the open sea area, positive SLA corresponds to lower CHL concentration (<0.5 mg / m 3 ), reflecting the inhibition of biological productivity. This distribution pattern preliminarily characterizes the interaction between monsoon-induced upwelling and large-scale eddy dynamics, especially the important role of "Great Whirl" in regulating regional productivity.

[0036] (II) Analysis of temporal variability Using monthly time series, as shown in Figure 1 , the annual and seasonal variation trends of CHL, SLA, SST, wind speed components (U, V) and wind stress (curl τ ) were calculated respectively. Figure 1 Fig. (a) reflects that CHL showed an increasing trend from 1997 to 2004, a decreasing trend from 2004 to 2012, and a gentle trend from 2012 to 2024. Figure 1 Fig. (b) reflects that SLA and SST showed an increasing trend during 1997-2024, and had similar variation trends. Figure 1 Fig. (c) reflects that the vertical component of wind speed (V) has a weak decreasing trend in the whole interval, while the horizontal component of wind speed (U) has a relatively flat trend in the whole interval. Figure 1The middle (d) reflects that the wind stress did not show a significant trend throughout the period.

[0037] (III) Correlation coefficients of CHL with other parameters based on Figure 1 The time series data were used to calculate the correlation between CHL and SLA, SST, wind speed U component, and wind speed V component, such as Figure 2 As shown, the results reflect a significant negative correlation between CHL and SLA and SST (r = -0.83 and -0.59, respectively), indicating that SLA and SST are two major influencing factors of CHL.

[0038] (iv) Monthly variability of CHL, SST and SLA The above time series data is averaged by year to obtain the monthly average data, such as... Figure 3 As shown, the results indicate that CHL concentrations significantly increased during the southwest monsoon, peaking in September at over 0.5 mg / m³. SLA exhibited a strong positive anomaly, reflecting the dominant role of the large eddy in the upwelling region. Meanwhile, SST and CHL showed an inverse correlation, reaching their lowest values ​​in July (<20°C) and subsequently gradually increasing with the advance of the monsoon, which may be related to the strengthening of the large eddy.

[0039] (V) Wavelet Analysis and Dominant Mode Recognition Continuous wavelet transform (CWT) is used to extract periodic and cyclical components from time series. This method decomposes data into a time-frequency space, revealing the main variation modes and their evolution over time. It achieves a balance between temporal localization (suitable for low-frequency, slowly changing patterns) and frequency localization (suitable for high-frequency, rapidly changing oscillations). In other words, wavelet analysis can accurately identify the frequency of low-frequency events, while high-frequency oscillations have higher temporal resolution but relatively weaker frequency localization. Morlet wavelets are widely used in ecological research due to their good balance between time and frequency localization. For data with uniform time steps... time series (n=1,...,N), calculate wavelet coefficients : (1); wavelet coefficients This represents the time series at time point n and scale s. With wavelet basis functions The degree of similarity. Wavelet coefficients The square of the modulus This is the wavelet power spectrum, used to measure energy intensity at time point n and scale s. (Index of summation variables) In the convolution calculation, all N time points are traversed.

[0040] Due to the limited length of time series, significant edge effects occur at higher scales. The global power spectrum is the average of wavelet power over time, reflecting the strength of the main variability modes, and is scale-normalized. The statistical significance of wavelet power is tested by comparison with a first-order autoregressive (AR1) red noise background. In the present invention, the starting scale s0 is set to be twice the time resolution of the time series; the discrete scale interval d j Set to 1 / 12 (each octave is divided into 12 sub-octaves), the number of octaves is 4.7, and finally 39 scales are obtained, covering a time range from 2 months to 4.5 years.

[0041] The present invention first detrends the monthly data in Figure 1 to construct a stationary time series, such as (a) (d) (g) in Figure 4 , then uses the Morlet wavelet function to perform CWT transformation on the CHL, SST, and SLA sequences, extracts their power spectra, such as (b) (e) (h) in Figure 4 , and the normalized global power spectrum, such as (c) (f) (i) in Figure 4 .

[0042] Figure 4 The results in (a) show that the CHL concentration increased before 2004, then gradually decreased, and stabilized around 2012. Since then, the CHL concentration has generally remained below 0.3 mg / m³, with occasional peaks slightly above this threshold. SST does not show a clear long-term trend, but exhibits seasonal fluctuations, varying between 17-28°C, and is generally lower during periods of higher CHL concentration. This suggests that higher temperatures may inhibit phytoplankton growth, while lower temperatures may promote CHL increase by increasing nutrient supply. SLA is dominated by positive values, is significantly affected by GW, but fluctuates between -0.02 and 0.02 meters, reflecting complex ocean dynamic changes.

[0043] Figure 4 (b) (e) (h) in Figure 4In the diagram, (c), (f), and (i) represent the corresponding global power spectra, which are the results of averaging the power over time. The Y-axis represents the period (in years), corresponding to the cyclical frequency of change in each data set. Wavelet transform allows for multi-resolution analysis, enabling the simultaneous detection of short-term fluctuations and long-term trends. The shading in the figure represents power intensity, with dark red indicating stronger spectral signals and light red indicating weaker signals. The black contour lines represent a 95% confidence level, indicating statistically significant changes. The shaded area is the "cone of influence" (COI), where edge effects may influence the interpretation of results at data boundaries. The global power spectrum is used to identify the dominant modes of change; the red dashed line represents the 95% confidence level, which is the threshold for determining statistical significance.

[0044] The results show that seasonal cycles, especially annual and semi-annual cycles, are the dominant modes driving changes in CHL and SST. Before 2004, SLA gradually shifted from being dominated by an annual cycle to being dominated by high-frequency oscillations, a shift that may be related to the upward trend of CHL during that period. Overall, SLA exhibits alternating dominance of high-frequency oscillations and annual cycles in different time periods, reflecting its dynamic spatiotemporal interaction characteristics.

[0045] (vi) Outlier extraction and scatter analysis This invention employs a multiple seasonal trend decomposition (MSTL) method to remove the main periodic and cyclical components identified by CWT. MSTL uses the Loess (Locally Weighted Scatter Smoothing) method to decompose the time series, extracting the trend, seasonal, and residual terms. Loess is a non-parametric method that effectively estimates nonlinear trends and seasonal variations by fitting smooth curves to the data locally. MSTL is particularly suitable for time series data with multiple seasonal patterns, capable of simultaneously separating multiple periodic events. Furthermore, MSTL can capture seasonal variations over time, making it ideal for handling non-stationary data such as CHL, which are influenced by both long-term environmental changes and seasonal cycles. The residual term represents the portion that cannot be explained by trend and seasonality, reflecting anomalous signals in the time series. By estimating and separating multiple cyclical events, MSTL, combined with wavelet analysis results, helps identify these anomalies. Wavelet analysis provides time-frequency characteristics highlighting cyclical changes, while MSTL isolates long-term trends and seasonal components, thus providing a more comprehensive understanding of cyclical behavior and anomalous changes.

[0046] Outliers extracted using the MSTL method after wavelet analysis were analyzed using scatter plots, such as... Figure 5 The image shows a scatter plot of monthly outliers in CHL and outliers in SST and SLA from 1997 to 2024. Figure 5 (a) Scatter plot of CHL-SST; Figure 5(b) Scatter plot of CHL-SLA; Figure 5 (c) Scatter plot of SLA-SST. Both the SST-CHL and SLA-CHL scatter plots show a negative correlation between CHL and both SST and SLA, with a stronger correlation between CHL and SLA (-0.33) and -0.27 with SST. This further emphasizes the important role of SST and SLA as major drivers of CHL variation. Furthermore, a positive correlation exists between SST and SLA, indicating an interrelationship among internal ocean dynamic processes. In the scatter plots, the red line represents the regression line, and the gray shaded area represents the 95% confidence interval.

[0047] (vii) CEOF space modal analysis The CEOF algorithm was used to analyze the dynamic evolution characteristics of the SLA field and extract its dominant modes, such as... Figure 6 As shown in the figure, the results reveal the first mode in the CEOF analysis, which captures 29% of the variance in the dynamic changes of SLA. CEOF demonstrates a significant advantage in detecting moving features, revealing a vortex-like structure in the offshore region, a typical characteristic of GW. Furthermore, a hook-like structure extends from the nearshore area to the offshore waters north of the GW, indicating that nearshore waters are transported to the open sea. This mode is primarily shaped by the interaction between the GW and coastal upwelling, clearly reflected in the spatial distribution maps of CHL and SLA. The phase relationships between these dominant signals suggest that the GW plays a crucial role in enhancing upwelling, strengthening nutrient transport, and boosting regional biological productivity.

[0048] The CEOF results include multiple spatial modes, each with a variance contribution rate. The larger the value, the smaller the variance contribution rate, and vice versa, the lower the mode number. This example selects the first spatial mode, with a variance contribution rate of 29.02%. Figure 6 As can be seen, a vortex structure exists within the range of 50.5°E-55°E and 6.5°N-10°N. This structure coincides with a large vortex, indicating the dominant role of the large vortex in this region. Above the large vortex, a hook-shaped structure extends from the beginning of the upwelling region to the open sea. This embodiment does not provide a quantitative analysis of these results; it only reflects the influence of the vortex on the upwelling region, thus hypothesizing the influence of the vortex on the CHL distribution.

[0049] Previous studies have demonstrated the significant role of SLA in CHL variability. To further elucidate its dynamic mechanisms, future research will focus on the interaction between eddies (including cyclones and anticyclones, particularly GW) and coastal upwelling, exploring how these processes collectively influence short-term fluctuations and long-term changes in CHL.

[0050] (viii) Vortex type and frequency distribution The CEOF analysis highlights significant eddy activity in the remote upwelling region, which is dominated by GW. It can be confirmed by plotting the spatial frequency of eddies by type that the regions highlighted by CEOF are mainly affected by anticyclonic eddies, likely GW itself. In contrast, cyclonic eddies are more frequent in the northeast and northwest of the core of GW, which is consistent with the hook-like structure observed in the CEOF analysis. The spatial distribution of cyclonic and anticyclonic eddies in the Somali upwelling region shows that the darker the color, the higher the frequency of eddy occurrence. Near 8°N, 53°E, the frequency of anticyclonic eddy occurrence reaches a significant maximum, corresponding to the core region of GW. As shown in Figs. 6a and 6b, the spatial distribution of cyclonic and anticyclonic eddy occurrence frequency is as follows: Figure 7 Figure 7 Fig. 6a shows the spatial distribution of cyclonic eddies (CEs). Figure 7 Fig. 6b shows the spatial distribution of anticyclonic eddies (AEs).

[0051] (Nine) Composite analysis centered on eddies To assess the impact of eddies on CHL distribution, the present invention adopts a composite analysis method centered on eddies. Figure 8 The CHL fusion map centered on eddies in the Somali upwelling region from June to September is shown, with the left figure being an anticyclonic eddy and the right figure being a cyclonic eddy. CHL data is normalized according to the eddy radius and fused within a range of twice the eddy radius. Anticyclonic eddies have lower CHL in the core region (<0.5 mg / m³), while CHL is significantly higher in the western and northwestern peripheral regions (>0.6 mg / m³), which is highly consistent with the hook-like structure in CEOF. Cyclonic eddies exhibit higher CHL concentrations (>0.65 mg / m³), often exceeding 0.8 mg / m³ near the upwelling side, but their CHL spatial distribution is relatively scattered and incoherent over time.

[0052] This peripheral CHL enhancement is consistent with the CEOF modal feature derived from SLA, reflecting the interaction between GW and the upwelling. The hook-like structure in CEOF outlines the clockwise distribution of CHL along the north side of GW, indicating that high CHL water is transported to the open sea. In contrast, cyclonic eddies are characterized by higher CHL concentrations (>0.65 mg / m³) overall, often exceeding 0.8 mg / m³ near the upwelling side, but their spatial distribution pattern is not as coherent as that of anticyclonic eddies, with greater temporal fluctuations.

[0053] ​Two snapshots from September 2018 and 2019 highlight the transport of high-concentration CHL (approximately 1 mg / m³ or higher) along the periphery of the gas vortex (GW), while a significant increase in CHL concentration was also observed within the nearby cyclonic eddy, as shown. The GW clearly plays a crucial role in the redistribution of high-CHL water to the open sea. Compared to 2018, the GW structure was less pronounced in 2019, with a noticeably narrower high-CHL zone along the coast. This indicates that the GW not only plays a key role in upwelling but is also a significant driving mechanism for the outward transport of high-CHL water; while the cyclonic eddy may enhance CHL concentration by uplifting eutrophic water through isodense surface upwelling.

[0054] (x) Joint analysis of horizontal and vertical data To further understand the control mechanism of vortex-upflow on the spatial distribution of CHL in the horizontal direction, Argo buoy profile data, including temperature, salinity, pressure, and density, were used.

[0055] First, the MLD (Medium Depth) of the hybrid layer is calculated using the density method, as shown in the following formula: ; Where z represents depth (m, positive downwards). This represents the density of seawater at depth z (kg / m³). 3 ), Density at a reference depth (10m) The threshold value is 0.03 kg / m³. 3 ).

[0056] Then, the monthly average MLD for each Argo buoy in the study area was calculated, such as... Figure 9 As shown, the results indicate that the monthly average MLD is deeper during the monsoon season, especially the southwest monsoon season (June-September), suggesting that increased wind speed can deepen the mixing depth of water bodies. Furthermore, using the vortex center composite analysis method, the spatial distribution of MLD within one radius of the subcyclone vortex and the cyclone vortex was calculated separately, as shown... Figure 10 As shown in the figure. The results show that the MLD is deeper in the center of the anticyclone and shallower at the edge; the MLD is shallower in the center of the cyclone and deeper at the edge. To quantify the temporal distribution of MLD, the monthly average MLD in anticyclones and cyclones was calculated separately, as shown in the figure. Figure 11 As shown. The results show that, Figure 11 In the middle (a) region, the MLD is generally deeper in anticyclones, while... Figure 11 The shallower cyclone is in the middle (b) cyclone.

[0057] Furthermore, the relationship between CHL, MLD, and SLA is analyzed using specific Argo buoy trajectory data. For example... Figure 12As shown, Argo buoy trajectory data numbered 1900182 is used, and the trajectory starts at 9 September 2004, inside the anticyclone in the right panel. Around this time (July-October 2004), it can be seen from the density time-depth section that the water density is small and uniform from the sea surface to 300 meters below the sea surface (corresponding to a water pressure of 300 dbar), indicating a deep MLD, a high and positive SLA, and a small CHL. This indicates that the anticyclonic eddy deepens the MLD, prevents the bottom nutrients from reaching the sea surface, and thus inhibits phytoplankton growth. In contrast, in July-October 2005, the water pressure is high and uniform, consistent with the characteristics of an upwelling, corresponding to a low SLA and a high CHL. This indicates that the upwelling can bring the high-density and nutrient-rich seawater from the bottom to the sea surface, thus promoting phytoplankton growth.

[0058] The specific embodiments of the present application are described above. It needs to be understood that the present application is not limited to the specific embodiments described above, and various changes or modifications can be made by those skilled in the art within the scope of the claims, which do not affect the essential content of the present application. The embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily without conflict.

Claims

1. A method for vortex-upflow synergistic regulation of chlorophyll-a concentration variability, characterized in that, Includes the following steps: (1) Obtain multi-source satellite and reanalysis datasets, including chlorophyll-a concentration, sea surface temperature, sea surface height anomaly, wind speed and Argo buoy data. Perform time fusion on daily data of chlorophyll-a concentration, sea surface temperature and sea surface height anomaly to generate monthly average data and construct stationary time series. (2) Perform continuous wavelet transform on the stationary time series to identify the dominant variation period in the anomalies of chlorophyll-a concentration, sea surface temperature and sea surface height, and use the multiple seasonal-trend decomposition method to separate the long-term trend, seasonal period and residual components to extract the monthly outlier sequence. (3) Apply complex empirical orthogonal function analysis to the sea surface height anomaly to extract its dominant spatial variation mode and identify the distribution location and shape of the eddy; (4) Target vortices were screened according to vortex type and season, chlorophyll-a concentration was normalized and aligned to the vortex center, and the spatial response relationship between vortex and chlorophyll-a concentration was analyzed. (5) Combining surface chlorophyll-a concentration with sea surface height anomalies and vertical Argo buoy data, density time-depth profile and CHL-SLA composite map were plotted to analyze the influence mechanism of eddy-upflow on the horizontal distribution of chlorophyll-a.

2. The method for vortex-upflow synergistic regulation of chlorophyll-a concentration variability according to claim 1, characterized in that, In step (3), the complex empirical orthogonal function analysis simultaneously resolves the amplitude and phase information of the abnormal changes in sea surface height and identifies the mesoscale vortex structure.

3. The method for vortex-upflow synergistic regulation of chlorophyll-a concentration variability according to claim 2, characterized in that, When identifying the structure of mesoscale cyclones and anticyclones, the vortex center and polarity are determined by calculating the gradient of the sea surface height anomaly and the Hessian matrix, the vortex boundary is determined by finding the maximum closed contour line, and the equivalent radius of the vortex is calculated.

4. The method for vortex-upflow synergistic regulation of chlorophyll-a concentration variability according to claim 1, characterized in that, In step (4), the chlorophyll-a concentration is normalized and aligned to the center of the vortex, and then standardized according to the equivalent radius range of the vortex to analyze the spatial distribution characteristics of the chlorophyll-a concentration inside and around the vortex.

5. The method for vortex-upflow synergistic regulation of chlorophyll-a concentration variability according to claim 1, characterized in that, In step (5), the joint analysis of horizontal-profile data includes: drawing a density time-depth profile for the Argo buoy trajectory, obtaining the daily chlorophyll-a concentration and sea surface height anomaly at the corresponding time and location, generating a CHL-SLA spatial composite map, superimposing the Argo buoy trajectory on the CHL-SLA spatial composite map, and comprehensively analyzing the influence of the vortex-upwell system on the horizontal distribution of chlorophyll-a concentration.

6. The method for vortex-upflow synergistic regulation of chlorophyll-a concentration variability according to claim 5, characterized in that, The joint analysis of horizontal-profile data used Argo buoy profile data, which includes temperature, salinity, pressure, and density. The mixed layer depth (MLD) was calculated using the density method. ； Where z represents depth. This represents the density of seawater at depth z. Density representing the reference depth, This represents the threshold.

7. The method for vortex-upflow synergistic regulation of chlorophyll-a concentration variability according to claim 1, characterized in that, In step (2), continuous wavelet transform is used to extract periodic and cyclical components from the stationary time series, decomposing the data into a time-frequency space. For data with a uniform time step... time series x n Wavelet coefficients, n=1,...,N for: ； Among them, wavelet coefficients This represents the time series at time point n and scale s. With wavelet basis functions The degree of similarity, For wavelet power spectrum, For the index of the summation variable.

8. The method for vortex-upflow synergistic regulation of chlorophyll-a concentration variability according to claim 1, characterized in that, Identifying cyclones and anticyclones using isopleths of sea surface height anomalies, and extracting vortex polarity, center location, and radius, includes the following steps: ① Polarity determination: ； Indicates the center point of the vortex SLA scalar field, denoted by , H represents the Hessian matrix, and detH represents the determinant; Polarity determination: When the trace of H tr H < 0, it is a local maximum, indicating an anticyclone; when the trace of H tr H > 0, it is a local minimum, indicating a cyclone. ② Construct closed contour lines: Let S be the set of saddle points, and let s be the extreme values ​​of the boundary contour lines. ; For the center point of the vortex The SLA scalar field of the connected saddle points s; Boundary equivalence of the anticyclone center for: ; cyclone center boundary isopleth for: ; The vortex boundary is a closed contour line. : ； For internal connected components, Indicates boundary equality. Points on a closed contour line SLA scalar field; ③ Calculate the center point of the vortex : ； ④ Calculate the area A and perimeter L of the closed domain: ， ； Where 'a' represents the area element of the internal connected domain. Represents the unit length on a closed contour line; Using the radius of the equal area as the vortex radius : 。

Images