Chlorophyll As A Reason of Climate Change

Global chlorophyll distribution induced by mesoscale eddiesAbstract：Mesoscale eddies can trap chlorophyll (CHL) and move it from one region to another. The concentration of oceanic CHL changes with depth, but satellites can only observe the CHL at the sea surface. Here, we estimate the eddy-induced CHL transport based on satellite observations, Argo float measurements, and global empirical models of vertical distribution of CHL. The combination of satellite altimeter data and Argo measurements is used to identify eddy boundaries by tracking the outmost closed contours of potential vorticity (PV) at depth; then, the surface CHL measured by satellite and the vertical distribution model of CHL are used to derive the CHL within eddy boundaries.

We find that the CHL trapped by mesoscale eddies can reach g, about half of the total CHL in the ocean; the global time-mean CHL eddy-induced zonal transport adds up to g/s westward.

The result shows oceanic mesoscale eddies play an important role in the elevated CHL in the interior of the ocean.

1. Introduction Unlike the large-scale circulation that continuously transports fluid, mesoscale eddies can discretely trap and transport fluid parcels and their associated water properties (Bibby et al., 2008; Chaigneau et al., 2011; McGillicuddy et al., 2007), including CHL, nutrients and zooplankton. They have been recognized as a key contributor in transporting heat (He et al., 2018; Jayne, S. R., Marotzke, J., 2002), dissolved carbon (Bidigare et al., 2003), and other tracers (Danabasoglu et al., 1994; Mei-Man et al., 1997). The eddy-induced transports of heat, salt (Dong et al., 2014; He et al., 2018) and water (Zhang et al.

Top Writers

Bella Hamilton

Verified writer

5 (234)

Samantha

Verified writer

4.9 (546)

Professor Harris

Verified writer

4.9 (457)

hire verified writer

, 2014; Yang and Xu, 2014; Meijers et al., 2007) have been shown to have a great impact on global climate change. The CHL is often used as a proxy to estimate primary production (Behrenfeld et al., 2006; Jenkins and Goldman, 2013). Mesoscale eddies are critical components of the ocean’s “internal weather” system. They are ubiquitous throughout the ocean and should exert a strong influence on CHL redistribution and ecosystems balance. Previous studies show that 10-50% of the new primary production is caused by eddy-induced nutrient fluxes (Huang et al., 2017; McGillicuddy et al., 1998; Oschlies et al., 1998; Lin et al., 2010; Ning et al., 2004). Upwelling of nutrient-rich water into the euphotic zone in cyclonic eddies can lead to enhanced CHL. The enhanced CHL concentration in cyclonic eddies have been confirmed in several studies (Chen et al., 2007; Falkowski et al., 1991).

Although the anticyclonic eddy is associated with downwelling, the reduction of nutrients is not that significant, this is because the downwelling is partially offset by ageostrophic upwelling caused by the anticyclonic eddy (Martin et al., 2001). A numerical study has shown that the CHL concentration in eddies can be much higher than the average value in the South China Sea (Xiu and Chai, 2011). Many mesoscale eddies originate from the eastern boundaries of the ocean basins where strong coastal upwelling gives rise to high CHL concentration, after reaching the interior of the ocean, eddies tend to contain higher CHL concentration than surrounding waters and therefore have a great impact on marine life (Gaube et al., 2013; Gaube et al., 2014). Although eddies play a significant role in distribution and transport of CHL in a certain region, the CHL trapped by global mesoscale eddies and its influence on CHL transport are still not clear. Although the volume or area of ​​a single eddy changes over time, some of the statistical properties of a large number of eddies are relatively stable. For example, mesoscale eddies with radius scales of order 100 km occupy about 25% of the ocean’s surface area at any given time (Chelton, et al., 2007). By the same token, although CHL trapped within a single eddy changes with time due to various active or passive processes (such as eddy stirring, eddy pumping, evolution of the eddy or phytoplankton production and losses), we assume that the total amount of eddy-trapped CHL in the global ocean or certain regions is relatively stable. Here we define eddy-trapped CHL as the amount of CHL trapped within eddy boundaries.

There are two major challenges for quantifying eddy-trapped CHL: one is the estimation of fluid volume trapped by eddies, and another is the determination of the vertical structure of CHL within the eddy interior. To estimate fluid volume trapped by eddies, it needs to define the boundary of an eddy. Potential vorticity is a conserved dynamical quantity (Rhines and Young., 1982), so the volume of fluid tends to be tracked within closed PV contours. Here we define the eddy boundary by the criterion of the outmost closed PV contours on isopycnals. The technique enables us to define eddy boundaries at depth from observed profiles of temperature and salinity and has shown promising capability in the study of transport of water mass, heat and salt trapped by mesoscale eddies (Zhang et al., 2014; Dong et al., 2014; He et al., 2018). The PV is estimated from the data set of Argo temperature/salinity profiles and sea level anomalies, and the boundary of the eddy is identified as the outermost closed PV contour on isopycnal surface at each depth. The accumulation of cruise and satellite observations provide us with opportunities to quantify the vertical structure of CHL. In the past, about 3806 situ CHL profiles widely distributed in the world ocean were collected during the cruise (Morel and Berthon, 1989). This makes it possible to establish an empirical model of the vertical CHL structure of the global ocean. The model displays a good capability to depict the change of CHL concentration with depth in the world ocean (Morel and Berthon, 1989; Uitz, Julia, et al., 2006). With the advances of Argo ocean observing network and satellites measurements, it allows us to reconstruct three-dimensional structures of eddies, and thus estimate the eddy-trapped CHL with the empirical model. Once the eddy-trapped CHL and the propagation speed of mesoscale eddies are known, the eddy-induced CHL transport can be obtained.

The paper is organized as follows: A summary of dataset and models of vertical distribution of CHL help to formulate requirements for calculating the eddy-trapped CHL (Section 2-3); Section 4 is devoted to focus on the quantification of eddy-trapped CHL distribution and eddy-induced CHL transport. In Section 5, we discuss and summarize the results. 2. Data 2.1. Altimetry dataset To improve the quality of our eddy identification and tracking, we use the AVISO “two-sat” merged dataset of global daily mean sea level anomaly (SLA) and sea surface height (SSH) with a spatial resolution from January 2003 to December 2014 (Pujol, et al., 2016). The datasets can be downloaded from (ftp.aviso.oceanobs.com). 2.2. Argo data Argo floats can provide the global temperature/salinity profiles. The profiles between 2003 and 2014 are available at http://www.usgodae.org/argo/argo.html. 2.3. CHL data The surface CHL with 8-days 4-km resolution is provided by MODIS-Aqua observations from year 2003 to 2014, which can be downloaded from (http://oceandata.sci.gsfc.nasa.gov/). The unit of CHL concentration is mg.m-3. 3. Methods 3.1. Identify mesoscale eddies from SLA data The Okubo-Weiss method is used to identify eddies from SLA data. The closed contours of the Okubo-Weiss parameter (Chelton et al., 2007) is given by (1) where the horizontal velocities are calculated by geostrophic balance (2) and(3) where is the sea level anomaly, is the gravitational acceleration, is the Coriolis parameter and the subscripts and represent the zonal and meridional spatial derivatives, respectively. The values of and are taken to detect eddies from SLA data. A wholly negative (positive) SLA within such contour indicates a cyclonic (anticyclonic) eddy. 3.2. Find the boundary of a single mesoscale eddy at each level After identifying the eddy from SLA data, the composite analysis approach (Zhang et al., 2014) will help us to calculate PV.

The details are as follows: the total density field is obtained from Argo data and the gridded climatological T/S field from the World Ocean Atlas 2005 data which can be downloaded from (http://www.nodc.noaa.gov/OC5/WOA05/pr_woa05.html), and then through hydrostatic balance with a reference level at , is obtained by (4) where is sea surface pressure anomaly, which can be calculated by the sea surface height anomaly , (5) is sea level anomaly at the center of an identified eddy. After calculating the eddy’s pressure field, the relative vorticity () in the Cartesian coordinates is calculated by (6) where QUOTE ρ is density field. Note that QUOTE ζx,y,z is then projected onto the isopycnal surfaces QUOTE ζ(x,y,ρ) . The potential vorticity is estimated by its definition (Rhines and Young, 1982) (7) where is the derivative of with respect to . With both relative vorticity and evaluated, the three-dimensional PV field can be computed. After projecting the PV distribution onto the isopycnal surface, the outermost closed contour on each isopycnal surface is regarded as the boundary of the eddy. So, the volume of an eddy can be evaluated, and the CHL concentration of an eddy will be calculated in the next section. 3.3. Eddy-trapped CHL First, as already stated, the in-depth boundaries of eddies are found (Section 3.1-3.2). Then, the surface CHL is collocated to the interior of mesoscale eddies with the time difference less than 4 days. The surface CHL data is linearly interpolated into SLA data grid, and then the eddy-induced surface CHL variability is isolated by high-pass filtering in time with periods longer than 500 days and in space with (Chelton et al., 2011a). Fig. 1 illustrates two examples of influence of eddies on the surface CHL. It shows some correlation between surface CHL and eddies. This echoes the results that ocean mesoscale eddies can affect surface CHL concentration. (Gao et al., 2017; Rubio et al., 2018; Werner et al., 2012; Ueno et al., 2010).

The model of vertical CHL distribution is based on the parameterization of extensive field data set, which was verified to be effective for simulating the global oceanic CHL profiles (Morel and Berthon, 1989; Sauzède, R., et al. 2015; Uitz, Julia, et al., 2006). The CHL concentration at depth can be calculated from (Morel and Berthon, 1989): (8) where is the background CHL concentration, is the CHL concentration at the depth of subsurface CHL maximum (), is the width of the subsurface CHL maximum. ,,and can be calculated from the CHL concentration observed by the satellite, and z depends on the euphotic depth, the euphotic depth is derived from the ocean color observation (Morel and Berthon, 1989). The global distribution of euphotic zone depth shown in Fig. 2a, which is obtained by averaging the euphotic zone depth within a bin around each grid point. We find that the euphotic zone depth mostly concentrated in meter, and the percentage of between 30m and 90m is (Fig. 2b). In other words, the eddy is inferred from SLA and Argo data, and the surface CHL concentration is obtained from ocean color observations over the area occupied by the eddy at approximately the same period. The selected surface CHL concentration combined with the Equation (8) can provide the CHL concentration at any depth within the eddy. We estimate the average value of depth-integrated to be (between 0 to). This value is basically consistent with Behrenfeld’s estimation in which the depth-integrated between 0 to euphotic depth from 1997 to 2006 varies from 4.2 to 5.0 Tg using the same model (Behrenfeld et al., 2006). The slight difference arises from the different integration depths and periods used in the estimation. The global depth-integrated CHL from 0 to accounts for 96.4% of the total depth-integrated CHL. Note that when the integral depth is greater than, a constant concentration which remains equal to the concentration at m is used. For the global analysis based on 12-years data record, we calculate the numbers of cyclonic eddies and anticyclonic eddies to be 67847 and 63514, respectively. We find that the number of cyclonic eddies is 6% more than anticyclonic eddies, the volume of cyclonic eddies is 20% larger than that of anticyclonic eddies, and the average CHL concentration in cyclonic eddies is 24% larger than anticyclonic eddies, and the average CHL concentration in cyclonic eddies is 24% larger than anticyclonic eddies. 4. Results 4.1. CHL trapped by mesoscale eddies In order to estimate the total CHL trapped by mesoscale eddies, we firstly identify eddies from the altimeter data (Section 3.1).

For each identified eddy, the concurrent Argo profiles of temperature and salinity are collocated to the interior of the eddy to construct the three-dimensional structure of the eddy (Section 3.2). After obtaining the volume trapped by the eddy and the concentration of CHL inside the eddy, we can estimate the eddy-trapped CHL. We assume a single mesoscale eddy moves coherently as bulk entity at all levels (Dong, et al., 2012; Dong, et al., 2014; Zhang, et al., 2013). For a single eddy with radial size and the CHL concentration , the CHL trapped by the eddy is calculated by (9) The upper limit of the integral is chosen to be 1.5 times the euphotic depth (). The vertical integral can be calculated by discrete summation as (10) QUOTE hz is the vertical interval between adjacent isopycnal surfaces, is the radius of eddy at the depth , is the average CHL concentration between adjacent isopycnal surfaces ( and ). In order to obtain the global distribution of eddy-trapped CHL, we set up a globally grid, for each grid, the time-averaged eddy-trapped CHL is calculated by (11) where is obtained by accumulating all the eddy-trapped CHL within a box centered at the grid point from January 2003 to December 2014, is the number of SLA snapshots. The eddy-trapped CHL is estimated to be g, accounting for about 56% of total CHL in the ocean. The total oceanic CHL can be estimated from the surface CHL and the model directly (Section 3.3).It can be seen from Fig. 3 that the eddy-trapped CHL is relatively small in tropical areas and larger in the mid and high latitudes except for the subtropical gyres and the “eddy deserts” regions. Regions of subtropical gyres were once considered oligotrophic, or nutrient poor because they are far from terrestrial runoff (Corno et al., 2007; Mcclain et al., 2004; Morris, et al., 1996).

The smaller eddy-trapped CHL near the equatorial region is attributed to the relatively small number of eddies, because the westward signal transmission is in the form of Rossby wave rather than eddies there (Chelton et al., 2011b). The larger eddy-trapped CHL in the mid and high latitudes arises from both large amount of westward propagating eddies and high CHL concentration caused by strong coastal upwelling near the east side of oceanic basins (Boyd et al., 2001; Brown and Fiechter, 2012; Gaube, et al., 2014). A notable feature is the existence of in relatively low eddy-trapped CHL in “eddy deserts” centered at about in the northeast Pacific which have been noted previously by Chelton et al (Chelton et al., 2011b). Compared to the oceans in the Northern Hemisphere, the eddy-trapped CHL is relatively large except regions of subtropical gyres and the southern edge of the Southern Ocean. The strong Antarctic Circumpolar Current (ACC) dominates the Southern Ocean’s large-scale circulation. Instabilities of the ACC jets release potential energy and produce large number of eddies (Cotroneo et al., 2013; Morrison et al., 2015; Thompson et al., 2012).

The strong upwelling in the Southern Ocean brings a rich source of the nutrients that supplies biological production. Enriched eddies and high CHL give rise to a high eddy-trapped CHL near 35o-50o S latitudes. A small number of eddies are identified at the southern edge of the Southern Ocean by satellite altimetry probably due to ice contamination and small size of eddies (Boyd et al., 2001; Cipollini, et al., 2017). Fig. 4 shows the distribution of observed surface CHL. Very high CHL appears near coastal regions, especially in the vicinity of eastern coasts, due to river runoff and strong coastal upwelling (Cipollini, et al., 2017). Coastal upwelling is rich in nutrients. These nutrients “fertilize” upper-layer waters, resulting in a high CHL. Fig. 4 and Fig. 3 show notable differences in CHL distribution in tropics and subtropical gyres. The westward equatorial current brings the CHL to the west, resulting in high surface CHL in eastern tropics (Gaube, et al., 2014; Killworth et al., 2004), but eddy-trapped CHL is low in tropics due to the deficiency of eddies. The low surface CHL in the subtropical gyres in Fig. 4 are related to dynamic feature of subtropical circulation.

The movement of ocean water within the Ekman layer of subtropical gyres forces surface water to sink, giving rise to the subtropical convergence near 20°–30° latitude where the nutrients are utilized but are not continuously resupplied by the cold, nutrient-rich water from below and therefore the surface CHL is minimized. However, due to the CHL trapped by mesoscale eddies in the subtropical gyres, the area of low CHL in subtropical gyres in Fig. 4 is largely reduced compared to that in Fig. 3. The differences in CHL variation in Fig. 3 and Fig. 4 indicates that mesoscale eddies exert a strong impact on the distribution of CHL.