A modelling study of hydrodynamical and biogeochemical processes within the California Upwelling System
Karsten Alexander Lettmann, Florian Hahner, Vanessa Schakau, Tim Wüllner, Cora Kohlmeier
aa r X i v : . [ phy s i c s . a o - ph ] A ug A modelling study of hydrodynamical and biogeochemical processes within theCalifornia Upwelling System
Karsten Alexander Lettmann a, ∗ , Florian Hahner a , Vanessa Schakau a , Tim W¨ullner a,b , Cora Kohlmeier a a Institute for Chemistry and Biology of the Marine Environment, Carl von Ossietzky University Oldenburg, Germany b OFFIS - Institute for Information Technology, Oldenburg, Germany
Abstract
The ROMS modeling system was applied to the California Upwelling System (CalUS) to understand the key hydrody-namic conditions and dynamics of the nitrogen-based ecosystem using the NPZD model proposed by Powell et al. (2006).A new type of sponge layer has been successfully implemented in the ROMS modelling system in order to stabilize thehydrodynamic part of the modeling system when using so-called ”reduced” boundary conditions. The hydrodynamicperformance of the model was examined using a tidal analysis based on tidal measurement data, a comparison of themodeled sea surface temperature (SST) with buoy and satellite data, and vertical sections of the currents along thecoast and the water temperature. This validation process shows that the hydrodynamic module used in this studycan reproduce the basic hydrodynamic and circulation characteristics within the CalUS. The results of the ecosystemmodel show the characteristic features of upwelling regions as well as the well-known spotty horizontal structures of thezooplankton community. The model thus provides a solid basis for the hydrodynamic and ecological characteristics ofthe CalUS and enables the ecological model to be expanded into a complex ecological model for investigating the effectsof climate change on the ecological balance in the area investigated.
Keywords:
California Upwelling System, ROMS modelling system, biogeochemical modelling, NPZD model
1. Introduction
Eastern boundary upwelling (EBU) system belong tothe most productive regions of the word ocean, which isdue to the fuelling of the photic zone by cool and nutrientrich water masses from below based on offshore Ekmantransport in surface waters. These upwelling regions ac-count for only about 1 % of the global ocean, but produceabout 20 % of the global fish catch and are also knownto support sea birds and mammals such as whales andseals (see e.g. K¨ampf and Chapman (2016) for a generaloverview of global upwelling systems). The four main east-ern boundary systems are those off a) California / Oregon/ Washington in the North Pacific, b) Peru and Chile inthe South Pacific, c) off northwest Africa and Portugal inthe North Atlantic, and d) off South Africa and Namibiain the South Atlantic. Apart from these four major sys-tems, a number of other upwelling systems exist through-out the global ocean, some of which are year-round fea-tures, whereas others occur on a seasonal basis (K¨ampfand Chapman, 2016).Understanding the physical and biogeochemical pro-cesses in these upwelling systems is of great importance.And coupled modelling systems have been valuable tools in ∗ Corresponding author
Email address: [email protected] (KarstenAlexander Lettmann) the past to contribute to this understanding. Within thismanuscript, we want to focus on the California UpwellingSystem (CalUS) and have developed a coupled modellingsystem for that region. In detail, we use a 3D coastal oceancirculation model coupled to a lower trophic level nitrogen-based ecosystem model, which are part of the ROMS mod-elling system ( R egional O cean M odelling S ystem, see e.g.Haidvogel et al., 2000; Wilkin et al., 2005).The ROMS modelling system has been applied to theCalUS many times before (see e.g. Gruber et al., 2006;Song et al., 2011; Jacox et al., 2014), in order to study dif-ferent physical and biogeochemical processes. For exam-ple, the strong horizontal nutrient gradients and lateralhorizontal transports by filaments and mesoscale eddies,which are characteristic for EBUs, was nicely illustrated(see e.g. Marchesiello et al., 2003; Nagai et al., 2015).This manuscript describes the application of the ROMSmodelling system to the CalUS. The used lower trophiclevel nitrogen-based ecosystem model is based on the four-component NPZD model by Powell et al. (2006), whichitself is mainly based on the studies by Spitz et al. (2003)and Newberger et al. (2003). We show some validation ofthe physical and biological module.The interested reader will find a short overview of theCalifornia Current System in Section 2. In Section 3, themodelling System is described, and its validation is pre-sented in Section 4. Finally, the new sponge layer typeused to stabilize the ROMS modelling system when us- non peer-reviewed manuscript August 26, 2020 ng so-called reduced boundary conditions is presented anddiscussed in Appendix A.
2. The California Current System
As the hydrography and its variability of the Califor-nia Current System (CCS) has been described in the pastby many authors (see e.g. Hickey, 1979; Lynn and Simp-son, 1987; Strub and James, 2000; Centurioni et al., 2008;Checkley and Barth, 2009; Gangopadhyay et al., 2011;K¨ampf and Chapman, 2016, and references therein), weonly want to give a brief description of the CCS in order toprovide the background for evaluating the hydrodynamicmodel features presented below.The California Current System consists of different cur-rent features with different water mass characteristics dueto their source regions, that are located at surface or belowsurface, and which might show a northward or southwardnet flow structure (see e.g. Checkley and Barth (2009)Fig. 1 or Gangopadhyay et al. (2011) Fig. 3 for a gen-eral overview of the different current features). It extends,in the north, from the Transition Zone (50 ◦ N, separat-ing the North Pacific and Alaska gyres), where the east-flowing North Pacific Current (also called the West WindDrift, see e.g. Strub and James, 2000) approaches NorthAmerica, to subtropical waters off Baja California, Mex-ico ( ∼ ◦ N) in the south (Hickey, 1979; Checkley andBarth, 2009). Checkley and Barth (2009) and Gangopad-hyay et al. (2011) summarize diverse current features andprocesses on different spatial and temporal scales that oc-cur in the CCS: wind-driven upwelling, the geostrophicallybalanced California Current (CC), the coastal jet, the Cal-ifornia Undercurrent (CU), Inshore Countercurrent (ICC)(Davidson Current), jets (narrow high-speed flows) in gen-eral, squirts (localized energetic off-shelf flows), filaments,mushroom-head vortices, mesoscale and sub-mesoscale ed-dies, and finally large meanders. When describing thehorizontal position of these hydrodynamic features, we fol-low Lynn and Simpson (1987), who separate the CCS intoan offshore oceanic zone ( ≈
300 - 1000 km), a near-shorecoastal zone ( ≈ ≈
200 - 300 km) (this spatial division is also evidentin Fig. 1 of Checkley and Barth, 2009). These zones in-teract with each other by various mechanisms, and theirwidths are only a rough estimate and are not static.Concerning the California Current (CC), the classicalview depicts a slow and broad current that flows equator-ward within about 1000 km of the west coast of NorthAmerica connecting the eastward North Pacific Currentat approximately 50 ◦ N to the westward North EquatorialCurrent at approximately 20 ◦ N (Strub and James, 2000;Checkley and Barth, 2009). It is a year-round, and surface-intensified flow usually in the upper 500 m that carriesabout 10 Sv (Sverdrup et al., 1942; Checkley and Barth,2009), and, according to Lynn and Simpson (1987), themain core of the CC is located within the transition zone.However, this picture of the slow and broad current has changed over the last decades (see e.g. Davis, 1985; Huyeret al., 1998; Centurioni et al., 2008; Marchesiello et al.,2003). According to Checkley and Barth (2009) and refer-ences therein, the southward flow can be partly organizedin form of intense equatorward jets that are embeddedwithin the region of slower southward flow (Mooers andRobinson, 1984; Huyer et al., 1998). The intense jets havewidths of 50 - 75 km, speeds in excess of 0.5 m s − , com-prise up to half of the total CC transport, and are mainlylocated in or near the transition zone mentioned above.According to Collins et al. (2003) and looking at graphs inStrub and James (2000), the observed California Currentjets in and near the transition zone can be seen as the in-shore edge of the broader mean seasonal equatorward flowof the California Current. In some cases, these jets can betraced back to coastal upwelling jets that separate fromthe coast, merge offshore (to about 130 ◦ W) to become afree, open-ocean jet that maintains its identity as the CCcore during spring and summer (Barth et al., 2000; Struband James, 2000; Gangopadhyay et al., 2011).The near-shore coastal zone and the transition zoneshow a very complex dynamic, which changes during theyear, and with more mesoscale features present in latesummer to early fall (e.g. Strub and James, 2000). Withinthese zones, the upwelling regions, the coastal jet, the Cal-ifornia Undercurrent (CU), as well as the Inshore Coun-tercurrent (ICC) are located (Checkley and Barth, 2009).It is the source region of coastal and westward propa-gating cyclonic and anticyclonic mesoscale eddies (Kurianet al., 2011) and it is the region with all those smaller-scales, high-energetic features like those filaments men-tioned above. In addition, within the coastal zone, thereexist some frequent standing eddies, like the counterclock-wise Southern California eddy located south of Point Con-ception (Checkley and Barth, 2009), a counterclockwiseeddy off San Francisco and about half the distance to PointConception (Hickey, 1979), eddies near Point Arena, andthe Cape Mendocino eddy (Hayward and Mantyla, 1990).The coastal jet is generated in geostrophic balance dueto both a drop in coastal sea level and the presence ofthe cold water front near the coast (Checkley and Barth,2009), which generates a strong, equatorward coastal up-welling jet with speeds of up to 1 m s − . According to thethermal-wind relation, the coastal upwelling jet is verti-cally sheared, with strongest currents near the surface, be-cause temperature, salinity and, hence, density vary in thecross-shelf direction. The jet is also horizontally shearedand fastest near the strongest cross-shelf density differ-ence, i.e. the coastal upwelling front (Checkley and Barth,2009). Interactions of the alongshore flow with coastal andbottom bathymetric features (capes, banks, canyons), incombination with hydrodynamic instability, also leads tointense alongshore variability e.g. visible in transient andeven persistent meanders and eddies especially from springto early fall (Strub and James, 2000; Centurioni et al.,2008; Checkley and Barth, 2009; Drake et al., 2011).Within the coastal zone of the CCS, two narrow pole-2ard flowing boundary currents are found. These cur-rents, the Inshore Countercurrent (ICC) and the Califor-nia Undercurrent (CU), are distinguished from each otherby their water mass characteristics, their vertical location,and their temporal presence during the year (Collins et al.,2000).The CU appears as a subsurface maximum of flow be-tween 100 and 250 m depth over the continental slope andtransports warm, saline equatorial waters (Chelton, 1984;Lynn and Simpson, 1987; Hickey, 1998). It is consideredto originate in the eastern equatorial Pacific and to flowpoleward along the North American coast (Sverdrup et al.,1942; Lynn and Simpson, 1987). Thus, the CU can be seenas an example of poleward undercurrents also present inother major ocean basins, which are usually found overthe continental slope and which have typical alongshorespeeds of 0.1 - 0.3 m s − and a depth range of 100 - 300m (Pierce et al., 2000). In the mean, the main core of theCU is located at 250 m depth near the continental slopeand it does not extend beyond 100 km from the coast, al-though there is some seasonal variability of the height andstrength (mean ≈ − ) of the core (Lynn andSimpson, 1987; Collins et al., 1996). In detail, the CU hasbeen observed at locations ranging from Baja Californiato Vancouver Island (Hickey, 1998), and shipboard sur-veys along the West Coast of the U.S. show poleward flowover the upper slope at all latitudes (Pierce et al., 1996;Collins et al., 2000).According to Collins et al. (2000), the ICC has beenreported as a seasonal flow, appearing in fall and win-ter (Reid Jr. and Schwartzlose, 1962; Lynn and Simpson,1987). It is found over both the shelf and slope and trans-ports shallow, upper ocean waters, which mainly are de-rived from CC waters with some modification by coastalprocesses. North of Point Conception, the ICC is some-times called the Davidson Current or the Davidson InshoreCurrent (Reid Jr. and Schwartzlose, 1962; Hickey, 1979).
3. Description of the coupled modelling system
For studying the CalUS, we decided to use the ROMSmodelling system ( R egional O cean M odelling S ystem, seee.g. Haidvogel et al., 2000; Wilkin et al., 2005), as it hasbeen applied to this study region many times before.Here, within this modelling system a 3D coastal ocean cir-culation model is coupled to a lower trophic level nitrogen-based ecosystem model. In the following we will describethe two modules separately. The hydrodynamic component is built using an onlineone-way nested system (see model domains in Fig. 1 ) withthe coarser parent grid having a horizontal resolution ofabout 15 km and the nested child grid having a resolutionof about 5 km, which corresponds to a three-times nestingrefinement. In the vertical, 30 layers of terrain-following S-coordinates are used with a strong refinement near the sea o W 126 o W 117 o W 36 o N 39 o N 42 o N 45 o N Cape MendocinoSan FranciscoMontereyBayNew-port
M1B1 B2 B3B4 B5B6
Figure 1: The model area of the 3D modelling system applied to theCalUS. The small panel top right depicts the extent of the parentgrid (blue box) and the one-way nested child grid (green box), whosebathymetry is depicted within the larger panel. The symbols B1 -B6 denote the NDBC buoys (used for temperature validation). Thered line denotes the position of the Newport transect. Finally, themagenta symbol M1 denotes the position of the M1 buoy of MontereyBay Aquarium Research Institute (MBARI). surface to resolve the upper 500 m of the water column anda smaller one near the sea floor (see the corresponding pa-rameters in Tab. 1). To capture subgrid-scale vertical tur-bulence, a k-kl variant of the Generic Length Scale (GLS)turbulence scheme (Umlauf and Burchard, 2003; Warneret al., 2005) is chosen for estimating vertical mixing coef-ficients for the hydrodynamic and the biological module,whereas for the horizontal mixing of momentum and ac-tive/passive tracers harmonic diffusion with constant tur-bulent diffusivities is used (see Tab. 1). To deal with theturbulent vertical flux of horizontal momentum within thebottom boundary layer a quadratic bottom friction is se-lected.In order to reduce the model spinup-time, initial con-ditions of free-surface elevation, horizontal water veloci-ties, temperature and salinity for January 2012 are takenfrom the 1/12 degree global HYCOM + NCODA reanaly-sis data (HYbrid Coordinate Ocean Model, see e.g. Bleck,2002; Chassignet et al., 2006). This also means that thereis a sufficient amount of turbulence already present at thebeginning of the model simulation, which does not haveto be built over a spinup process of several years as e.g.described by Marchesiello et al. (2003).Hydrodynamic open-boundary conditions for the largerparent grid are obtained from two different data sources.In order to include tidal effects into the modelling sys-tem, tidal-harmonic constants are provided using the OSUTidal Data Prediction Software together with the OTIS http://volkov.oce.orst.edu/tides/otps.html Table 1: Model parameters used for the hydrodynamic module ofthe 3D coupled modelling system
General parameters:Number of vertical layers 30S-coordinate transformation equation 2S-coordinate stretching function 4S-coordinate surface control parameter 7.0S-coordinate bottom control parameter 0.5Quadratic bottom drag coefficient. 0.00253D velocity nudging time scale 1.0 dTracer nudging time scale 1.0 dFactor outflow/inflow nudging 10Parent grid:External/barotropic time step 1.5 sInternal/baroclinic time step 30 sHorizontal resolution ≈
15 kmHoriz. turbulent viscosity 300 m s − Horiz. turbulent tracer diffusivity 300 m s − Child grid:External/barotropic time step 0.5 sInternal/baroclinic time step 10 sHorizontal resolution ≈ s − Horiz. turbulent tracer diffusivity 100 m s − Figure 2: Schematic of the considered biological model after Powellet al. (2006): dissolved inorganic nitrogen (N), particulate nitrogen(detritus: D), phototrophic phytoplankton (P), and herbivorous zoo-plankton (Z). gree. Time-varying river runoff data for the year 2012from USGS, USA , is also included for the following largerrivers: Stikine River (Alaska), Columbia River (Oregon /Washington), Rogue River (Oregon), Klamath River (Ore-gon / California), Eel River (California), Sacramento River(California), San Joaquin River (California). For eachriver, its discharge and contribution to salinity has beenconsidered, whereas coastal temperatures and state vari-ables of the biological module are not affected by riverrunoff. The Powell et al. (2006) four-component NPZD model,which itself is mainly based on the studies by Spitz et al.(2003) and Newberger et al. (2003), is used as a sim-ple model with sufficient complexity to investigate biogeo-chemical conversion rates under the influence of turbulenttransport processes. The model parameters are mainlytaken as described within these articles (with one excep-tion, see below). Therefore, we skip a very detailed vali-dation of the biological model. Within this nitrogen-basedtrophic module, total nitrogen is partitioned between dis-solved inorganic nitrogen (N), particulate organic nitrogen(detritus: D), phototrophic phytoplankton (P), and herbiv-orous zooplankton (Z).The dynamics of each of these four components andtheir interactions are illustrated in Fig. 2 and can be de-scribed via a transport-reaction equation of the form: ∂C∂t + ∇ · ( v C ) = ∇ h · ( D h ∇ h C ) + ∂∂z (cid:18) D v ∂C∂z (cid:19) + R (1) U.S. Geological Survey, 2016, National Water InformationSystem data available on the World Wide Web (USGS Wa-ter Data for the Nation), accessed October 2018, at URLhttp://waterdata.usgs.gov/nwis/ igure 3: The contour plot denotes the anomaly (the spatial meanhas been subtracted) of the summer-mean modified geostrophicstream function S geo after Eq. (9) in 100 m depth, whereas the sum-mer mean Eulerian horizontal velocities at 100 m are denoted bythe black arrows. The red lines mark the location of the verticaltransects, for which normal velocities and temperatures are plottedwithin Fig. 4. where ∇ denotes the nabla-operator, v the water velocity( v = ( u, v, w )), D h the horizontal eddy diffusion coeffi-cient, D v the vertical eddy diffusion coefficient, and finally, R the net conversion rate for each species.In detail, the net biological conversion rates for eachspecies are (for more details see Powell et al., 2006): R N = δD + γ n GZ − U P (2) R P = U P − GZ − σ d P (3) R Z = (1 − γ n ) GZ − ζ d Z (4) R D = σ d P + ζ d Z − δD + w d ∂D∂z (5)The detritus rate term R D is augmented by the detritussinking rate (the last term on the right in Eq. 5), whichis actually not a biological conversion term. Furthermore,the following definitions are used: G := R m (1 − e − Λ P ) (6) U := V m Nk U + N αI p V m + α I (7) I := I par exp (cid:18) − k z z − k p Z z P ( z ′ ) dz ′ (cid:19) (8)Light attenuation is modelled by Eq. (8), where z denotesthe (positive) vertical distance between the sea surfaceand the position within the water column, I the variablesea-surface short wave radiation flux, and par the frac-tion of light that is available for photosynthesis (see e.g. Figure 4: Horizontal normal current velocity ([m s − ]) (panels a)- c)) and potential temperature ([ o C]) (panels d) - f)) on the ver-tical sections marked by the red lines in Fig. 3. Panel a) and d)correspond to the Newport transect, panel b) and e) to CalCOFIline 46.7 starting at Cape Mendocino, and panel c) and f) belongto CalCOFI line 60 starting at San Francisco. Normal velocities innorthward direction are denoted by positive (red) values.
Fennel et al., 2006). The parameters of this NPZD-modelare mainly identical to the values used in Powell et al.(2006) and are listed in Tab. 2. However, the parametersof grazing by zooplankton upon phytoplankton have beenchanged according to Fiechter et al. (2009), as this leads toa larger mortality of phytoplankton due to grazing, whichis more realistic for the CalUS (personal communicationwith Jerome Fiechter).At the open boundaries of the model domain, a nudg-ing method is used to force the biological variables to pre-scribed values. Apart from the DIN pool, all other biolog-ical variables are forced to zero for all times and all depthlevels. However, dissolved inorganic nitrogen (DIN) is setaccording to the depth-dependent annual-mean NO − con- Table 2: Model parameters used for the biological module of thecoupled 3D modelling system (mainly identical with Powell et al.,2006; Spitz et al., 2003).
Parameter Name Symbol ValueLight extinction coeff. k z − Self-shading coeff. k p µ M-N) − Initial slope of P-I curve α W − PAR-fraction par V m − Uptake half saturation k U µ M-NPhyto. senescence σ d − Zoop. grazing rate R m − Ivlev constant Λ 0.84 µ M-N − Excretion efficiency γ n ζ d − Remineralization δ − Detrital sinking rate w d − SS H [ m ] San Francisco
ROMS tide gauge SS H [ m ] Tofino / Canada
Figure 5: Comparison of modelled sea surface height (SSH) withtide gauge data at Fort Point, San Francisco, and Tofino (VancouverIsland, Canada) over a time-period of two weeks in April 2012. centration obtained from the Levitus data set (Levitus,1982). The initial conditions are set to a value of 1.0 mmol-N m − for all variables apart from the nitrogen pool, whichis set to 17.0 mmol-N m − . So, although DIN also includesother N-species like nitrite or amonium, the boundary andinitial conditions were mainly set to nitrate based num-bers.
4. ’Validation’ of the modelling system
After this technical description of the two coupled mod-ules, we want to demonstrate that the modelling system athand provides a sufficient first representation of the CalUSwithout reproducing all hydrodynamic and biogeochemicalfeatures of the CalUS.
Within this part, the hydrodynamic module is to bevalidated by means of temperature data, data of free-surface elevation, and current velocity.
Current Velocities.
First, we want to demonstrate that themodel captures the main features of the horizontal circu-lation within the CCS as described above. As an example,Fig. 3 depicts the summer mean horizontal Eulerian ve-locity in 100 m depth. In addition, this figure also depictsthe contour lines of the anomaly (the spatial mean hasbeen subtracted) of the 2012 summer-mean modified dy-namic height: S geo ( x, y, z ) := ζ ( x, y ) + 1 ρ Z ζz ρ ano ( x, y, z ′ ) dz ′ (9) Downloaded in March 2018 from http://iridl.ldeo.columbia.edu/SOURCES/.LEVITUS/index.html The mean is calculated as the mean of two-days averages over90 days during the months June, July, and August 2012.
Here, ζ denotes summer-mean sea surface height, ρ = 1000kg/m , ρ ano the density anomaly to 1000 kg/m obtainedfrom the summer mean distribution of potential temper-ature and salinity. As gS geo /f (with g the gravitationalacceleration, f the Coriolis parameter) is a geostrophicstreamfunction, S geo can be considered as some kind ofscaled geostrophic streamfunction, whose contour lines shouldalso be parallel to the geostrophic velocity.As is evident from Fig. 3, the mean horizontal currentvelocities are more or less parallel to the streamlines, whichis due to the validity of the geostrophic approximation.This picture shows the drop of S geo towards the coast;and the resulting coastal jet mentioned above is clearly vis-ible (especially between Newport and Cape Mendocino).Similar to observations, this jet shows strong meandering,and separates from the coast south of Cape Mendocinoas indicated in Fig. 3. of Gangopadhyay et al. (2011).While separating from the coast near Cape Mendocino,the coastal jet seems to split into two branches: one mov-ing offshore, one flowing further near the coast. Also simi-lar with this figure, within the south-west corner of Fig. 3,meanders and jets of the broader and slower CC seem tobe visible. In addition, there are also some strong closedcontours of the summer mean S geo , that might indicatestanding eddies within that region. Some of them mightcorresponds to mentioned features within the literature,as the one south of San Francisco near 36 N, that mightcorrespond to the San Francisco eddy mentioned in Hickey(1979); likewise the smaller one near Point Arena (near 39N) mentioned in Hayward and Mantyla (1990).The mentioned hydrodynamic features are also visibleon vertical sections of the horizontal current velocity de-picted in Fig. 4, and the positions of these transects of450 km length are marked by the red lines in Fig. 3. Thenorthern most transect is chosen according to the tran-sect presented in Powell et al. (2006). The two southernlines are chosen after the California Cooperative FisheriesInvestigations (CalCOFI) sampling grid. The line start-ing at Cape Mendocino corresponds to CalCOFI transectwith line coordinate The transformations between the CalCOFI sampling coordi-nates and geographic coordinates in latitude and longitude are per-formed after Weber and Moore (2013) using the Matlab softwarepackage by Robert Thombley and Augusto Valencia downloadedfrom http://calcofi.org/field-work/station-positions/calcofi-line-sta-algorithm.html. igure 6: SST at 16 June 2012 obtained from the 1/12 degree global HYCOM + NCODA reanalysis data (left), the hydrodynamic modelcomponent of the ROMS modelling system presented in this study (middle), and finally from L4-G1SST satellite data (left). upwards to 100 m, such that its northward flow velocityis also visible by the northward pointing velocity vectorsclose to the coast in Fig. 3. Table 3: Results of the tidal analysis for year 2012 obtained via theT-tide software package (Pawlowicz et al., 2002) at San Francisco andTofino (Vancouver Island, Canada) tide-gauge stations. The phasedifference is obtained by subtracting the phase value of the modelleddata from the phase value of the measured data. So, a negative phasedifference denotes the model lagging behind the data.
San Francisco Tofinodata model data model M amp. [m] 0.57 0.52 0.97 0.96 M ∆ φ [ ◦ ] 15.05 -0.50 M ∆ φ [min] 31.15 -1.04 S amp. [m] 0.13 0.12 0.28 0.29 S ∆ φ [ ◦ ] 10.0 -3.72 S ∆ φ [min] 19.99 -7.43 K amp. [m] 0.37 0.41 0.39 0.49 K ∆ φ [ ◦ ] 8.94 6.57 K ∆ φ [min] 35.66 26.20 O amp. [m] 0.23 0.26 0.24 0.31 O ∆ φ [ ◦ ] 8.07 7.24 O ∆ φ [min] 34.72 31.15 Sea Surface Height (SSH).
As the tides quite significantlycontribute to the short-term variability, we want to paysome attention to their representation within the model.As the model is forced along the open boundaries by tidalelevations as described above (here, model errors are onlydue to the errors within the forcing data), it might be ques-tionable if the (tidal) fluctuations of SSH in the centralpart of the model domain are of the right order. There-fore, tide gauge data of two stations located in the cen-tral area of the model domain have been considered formodel validation: from Tofino (Vancouver Island, Canada)within the coarser grid, and from San Francisco (FortPoint) within the finer child grid. As a first inspection,SSH time series for a two-weeks period in April 2012 aredepicted in Fig. 5. Although not matching perfectly, it isevident from this figure that the tides are represented ap-propriately with respect to amplitude and phase at thesetwo tide gauge stations. This is further confirmed by us-ing a Taylor diagram analysis (see Fig. 8) for the total year2012, which shows the good agreement in terms of stan-dard deviation and correlation. In addition, the harmonicanalysis (using the T-tide software package of Pawlow-icz et al. (2002)) presented in Tab. 3 shows a satisfactoryagreement of the tidal amplitudes of the harmonics M2,S2, O1, and K1 for both tide gauge stations. However,there is some phase lag in the order of some minutes of7 an Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec81012141618 [ ° C ] SST at buoy B1
ROMSbuoy data
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec6810121416 [ ° C ] SST at buoy B4
ROMSbuoy data
Figure 7: Comparison of modelled (blue line) and measured (greenline) sea surface temperature (SST) at buoys B1 and B4 for the year2012 (for the location of these buoys, please see Fig. 1). the M2 and S2 tide in San Francisco, which might be dueto small errors in tidal wave propagation. In the end, thisvalidation process demonstrates that the tidal dynamics inthe CalUS is represented to the right order of magnitudewithin the hydrodynamic module at hand.
Water Temperatures.
As signatures of upwelling processesare usually visible as specific features of sea surface tem-perature, we compare time series of modelled and mea-sured SST at different sites within the CalUS, for whichbuoy data have been available from the NDBC . The loca-tions of these buoys are shown in Fig. 1 and are denoted bysymbols B1 - B6, which correspond to the following buoynumbers: B1 - 46002; B2 - 46014 ; B3 - 46026 ; B4 - 46027;B5 - 46028; B6 - 46059.Fig. 7 depicts the 2012 time series of SST at two buoysites: one close to the coast (B4) and one more offshore(B1). This figure demonstrates that the model capturesthe long-term evolution of the SST quite well. It even re-produces the timing and strength of some upwelling eventsin May and June 2012, which can be seen from the lowerpanel in Fig. 7. However, is is also visible from the timeseries of the near-shore site that the model might overes-timate the strength of the upwelling process a bit, whichcould explain the underestimation of the near-shore SST.In addition, the SST comparisons for the other buoy sitesare depicted within the Taylor diagram in Fig. 8, fromwhich it is evident that the SST dynamics is captured bythe model in a sufficient manner.This impression of a slight overestimation of the up-welling process can also been reasoned from Fig. 6, whichshows a spatial plot of modelled SST for 16 June 2012 to-gether with SST fields obtained from the HYCOM model normalized standard deviation no r m a li z e d s t a nd a r d d ev i a t i on c o r r e l a t i o n c o e f f i c i e n t B1B2B3 B4 B5 B6SFTo
Figure 8: Taylor diagram for validation of modelled sea surface tem-perature at buoys B1 - B6 (for the location of these buoys, pleasesee Fig. 1) and SSH at tide gauge stations located in San Francisco(SF) and Tofino (To) (Vancouver Island, Canada). and satellite data. Although the position and extent of theupwelling region are quite similar to those features withinthe HYCOM data and the L4-G1SST satellite data , ourmodel seems to overestimate the upwelling process in thattime period to some extent, which is evident from thecolder SST values near the coast. However, looking at spe-cial features of SST at that day, the model clearly showsfilaments as well as mushroom-like and eddy structuresdue to the upwelling process as mentioned above.In addition, the near-shore upwelling of colder subsur-face waters can also be seen on the vertical CalCOFI sec-tions depicted in Fig. 4 d) - f), which are comparable tovertical temperature sections shown in Marchesiello et al.(2003) or Chenillat et al. (2013). To illustrate the performance of the biological mod-ule, monthly mean values of dissolved inorganic nitrogen(DIN), phytoplankton and zooplankton are plotted alongthe Newport transect (similar to Powell et al., 2006, redline in Fig. 1) for May 2012 in Fig. 9. The upwelling ofDIN-rich water mass is visible at the coast in panel Fig. 9a,as well as the DIN consumption within the euphotic zoneby growing phytoplankton. The growing of zooplankton isvisible in panel Fig. 9c, and the remineralization of detri-tus to DIN is evident from the increase of nutrients belowthe euphotic zone within the depth-interval between 50 -100 m in Fig. 9a. Data were obtained via the web-portal world-view.earthdata.nasa.gov. igure 9: Monthly mean values for May 2012 of the four speciesof the biological model on the vertical transect near Newport. Thelocation of this transect is depicted within Fig. 1. The interaction and the temporal succession of the N-species is also visible within a temporal snapshot of near-surface distributions of the N-species depicted in Fig. 10.This figure shows the horizontal mesoscale structure of thebiological variables, which is similar to the structures de-picted in Powell et al. (2006), and which also shows anincrease in nutrients towards the coast due to the pre-vailing upwelling dynamics. As an example, within thesouth-western region of the four panels of this figure, astretched structure is visible as a depletion of N, which isalso present as a build-up within the other panels for P, Zand D. Thus the temporal succession of nitrogen thoughthe four pools is visible.What is also visible within these horizontal distribu-tions, is the patchy structure of the zooplankton speciesthat is described e.g. in Messi´e and Chavez (2017). Messi´eand Chavez (2017) and Fiechter et al. (2020) investigatethe formation and occurrence of zooplankton hotspots withinthe CalUS. They attribute their formation and distribu-tion to regions of coastal nutrient upwelling as well as con-verging and diverging surface currents. The simple NPZDmodel at hand shows, to some extent, a similar behaviouras depicted in Fig. 11. Within that figure, the temporalevolution a zooplankton patch near Monterey Bay (de-noted as MB within that figure) is shown over a time pe-riod of nearly two weeks in June 2012. The simultaneousgeneration of zooplankton from phytoplankton within anupwelling centre as well as the horizontal transport of zoo-plankton within a narrow stripe to offshore locations isvisible from this figure.In order to investigate and validate the temporal dy-namics of this simple NPZD model, modelled near-surfaceDIN concentrations are compared to measured nitrate val-ues at a long-term time series station in Monterey Bay (M1
Figure 10: Exemplary horizontal distributions of the biological mod-ule at 16 June 2012 in a water depth of 15 m below sea surface. buoy, see Fig. 1 for the position of this buoy) (Sakamotoet al., 2017; Chavez et al., 2017). Concerning the originof upwelled water masses, Monterey Bay is mainly influ-enced by the upwelling center off A˜no Nuevo (to the northof the bay), from which cold and nutrient-rich water entersMonterey Bay (see e.g. Chavez et al., 2017), although up-welling and/or mixing occurs along the entire region fromA˜no Nuevo to Point Sur. Due to its wind-protection capa-bilities, its slower circulation and a warm and stable mixedlayer, Monterey Bay is a classical upwelling shadow envi-ronment that foster dense phytoplankton blooms (Chavezet al., 2017).As mentioned above, DIN includes more N-species apartfrom nitrate, such that a direct comparison of nitrate andDIN is problematic, However, the DIN is initialized andset at open boundaries to the order of long-term nitrateconcentrations. Therefore, the DIN within this manuscriptis of similar magnitude as nitrate. According to Chavezet al. (2017) the nitrate concentration at M1 buoy at theentrance of Monterey Bay is strongly influenced by coastalupwelling of deep nutrient rich waters. Riverine nitrate in-put is negligible at that site in general and only of someimportance during winter months (Sakamoto et al., 2017).The long-term (1988 - 2016) climatological nitrate con-centration shows a peak due to coastal upwelling duringMarch and July (Chavez et al., 2017), which is to someextent visible for the year 2012 in Fig. 12 (see green line inupper panel). It is evident from this figure that the mod-elled nitrate concentration (blue line in upper panel) is ofsimilar magnitude. However, it is also evident that the twotime series (modelled and measured) do not match. As anexample, the modelled time series depicts some peak at thebeginning of August 2012, whereas the measured nitratedata show a falling trend with low absolute values.In the lower panel of Fig. 12, the modelled (blue line)9 igure 11: Exemplary horizontal evolution of a zooplankton[mmol/m ] patch in a depth of 15 m below sea surface in the firsthalf June 2012. and measured (green line) SST is depicted, which showsthat the overall trend and the right order of magnitudeis captured by the model. However, some the short-termfluctuations are not always captured by the model, as cane.g. be seen in mid-May as well as at the beginning of July,when some cooler (an likely upwelling) events are missedby the model.In order get some idea, what causes these differences,we also compared the SST at this site, which seems tobe a good indicator of upwelling waters. According toSakamoto et al. (2017), as a first approximation, the ni-trate concentration (mmol/m ) can be estimated from seasurface temperature (SST, o C) by a simple linear regres-sion model (correlation coefficient r = 0.59) of the form:NO − = − . × SST + 43 .
59 (10)Thus, the top panel also shows the SST-based estimatednitrate concentrations using the modelled (cyan line) andmeasured (red line) sea surface temperature at buoy M1.It is evident that the model seems to capture an up-welling event with the largest measured nitrate concen-tration at the end of May, which is visible in all the fournitrate curves. However, the model shows high measurednitrate concentration at the beginning of July and end ofJuly, which only partially seem to be related to modelledupwelling events. Whereas the event at the beginning ofJuly shows some modelled upwelling signature (the cyancurve shows a small peak), the high nitrate concentrationat the end of July / beginning of August has no upwellingcounterpart.The reason for the occasional overestimation of themodelled nitrate data might be due to the initialization ofthe nitrate concentration within the model domain, whichalso puts hight nutrient loads in the upper waters of theeuphotic zone which is evident from Fig. 9, which shows
Figure 12: Time series of modelled and measured nitrate concentra-tions (upper panel) as well as sea surface temperature (SST, lowerpanel) at MBARI buoy M1 in 2012. (See Fig. 1 for the positionof this buoy in Monterey Bay.) Within the upper panel, the blueline denotes the modelled (m.) ’nitrate’ concentration, whereas themeasured is denoted by the green line (d.). Using a simple linearregression model based on measured SST, the other two lines de-note the estimated nitrate concentration using Eq. (10) (m. by SST- using modelled SST; d. by SST - using measured SST). nutrient concentration of about 15 mmol/m in offshorewaters. Therefore, in the model, in addition to upwellingof nutrients from below, horizontal currents could trans-port nutrients to buoy M1, which might explain the ni-trate peaks outside the upwelling season. Alternatively,the temporal dynamics of the uncoupled (i.e. without anyspatial transports) NPZD model could be visible on thisfigure.In order to improve the model, the initialization of nu-trients should be done in that way that the surface watersdown to a depth of about 200 m should be free of nutri-ents. Only deeper waters should contain initial nutrientloads that can be brought to the surface by coastal up-welling. As an alternative, the model should have spun upover several years to deplete offshore waters from nutrientsdue to detrital sinking.
5. Summary and conclusions
A coupled modelling system for the California UpwellingSystem has been presented and validated. The hydrody-namic performance of the model is investigated in moredetail by means of a tidal analysis against tide gauge data,a comparison of modelled sea surface temperature (SST)against buoy and satellite data, as well as vertical sectionsof along-shore currents and water temperature. Althoughthe upwelling dynamics along the coast might be a bitoverestimated as demonstrated by means of SST, in theend, the validation process demonstrates that the hydro-dynamic module used within this study is capable to re-10roduce the basic hydrodynamic and circulation featureswithin the CalUS.However, the usage of the simple NPZD module formodelling the real nutrient and planktonic dynamics inthat region is quite conceptual, as more sophisticated mod-els are standard and available (see e.g. Fiechter et al.,2020). On the other hand, the simple NPZD model athand is able to show some basic features of plankton dy-namics with the right order of magnitude within that re-gion.To conclude, the presented modelling system might bea valuable tool to investigate the physical and (to some ba-sic extent) the biogeochemical dynamics within the CalUS.However, although the physical module already seems tobe ’matured’ to a sufficient degree, the biological has to beimproved by means of more realistic initial and boundaryconditions as well as a more sophisticated structure of theunderlying ’food-web’ (e.g. using more planktonic speciesand higher trophic levels).
6. Acknowledgments
We are grateful to Klemens Buhmann, Matthias Schr¨o-der and Stefan Harfst for technical support. We wantto thank J¨org-Olaf Wolff and David M. Checkley Jr. forhelp and advice concerning the manuscript. The numer-ical simulations were performed on the high performancecomputing cluster CARL, financed by the Ministry for Sci-ence and Culture (MWK) of Lower Saxony, Germany, andthe German Research Foundation (DFG). VS was fundedwithin the research project ’ENVICOPAS - Impact of En-vironmental Chances on Coastal Pathogen Systems (EN-VICOPAS)’ by German Research Foundation (DFG) un-der grant number 283700004. FH was funded within theresearch project ’Macroplastics Pollution in the SouthernNorth Sea - Sources, Pathways and Abatement Strategies’by the Ministry for Science and Culture (MWK) of LowerSaxony, Germany.We finally want to mention some of our data sources:The 1/12 deg global HYCOM + NCODA Ocean Reanal-ysis was funded by the U.S. Navy and the Modeling andSimulation Coordination Office. Computer time was madeavailable by the DoD High Performance Computing Mod-ernization Program. The output is publicly available athttp://hycom.org. Atmospheric forcing data for estimat-ing the upwelling index were obtained from NCEP Reanal-ysis provided by the NOAA/OAR/ESRL PSD, Boulder,Colorado, USA, (see e.g. Kalnay et al., 1996). The L4-G1SST satellite data set, a blended Global 1-km Sea Sur-face Temperature Data Set for Research and Applicationswas provided by Yi Chao, Benyang Tang, Zhijin Li, PeggyLi, Quoc Vu, Jet Propulsion Laboratory. The atmosphericforcing for the hydrodynamic modelling system containsmodified Copernicus Climate Change Service Informationfor the year 2012 provided by ’Copernicus Climate ChangeService (C3S) (2017): ERA5: Fifth generation of ECMWFatmospheric reanalyses of the global climate. Copernicus −4 −3 −2 −1 0 1 2 3 4−4−3−2−101234 u i [m/s] u s i [ m / s ] Velocity characteristics in sponge layerc = 0.1 s/mc = 0.5 s/m
Figure A.13: Illustration of the sponge layer characteristics used atopen-boundary points to stabilize the vertical-mean velocity whenusing the sponge layer in combination with ’reduced’ boundary con-ditions. The green and blue line correspond to sponge-layer curvesfor c = 0. The black dashed line denotes the sponge-layer curve for c = 0. Climate Change Service Climate Data Store (CDS), down-loaded in August 2018 viahttps://cds.climate.copernicus.eu/cdsapp.Last but not least, we want to say ’thank you’ to Mon-terey Bay Aquarium Research Institute (MBARI) (by nameReiko Michisaki) for providing the ISUS nitrate time seriesdata (Chavez et al., 1994) obtained at their surface buoyM1. MBARI provides data ”as is”, with no warranty, ex-press or implied, of the quality or consistency. Data areprovided without support and without obligation on thepart of the Monterey Bay Aquarium Research Institute toassist in its use, correction, modification, or enhancement.
Appendix A. The proposed sponge layer type
A new type of sponge layer is introduced into the ROMSsource code to stabilize the vertical-mean velocity at openboundaries when so-called reduced boundary conditionsare used. In this case, the free-surface elevation is pre-scribed at open-boundary points and the vertical-mean ve-locities are derived from a simplified momentum balanceincluding pressure and Coriolis force. As this single pre-scription of the free-surface elevation and the derivationof the vertical-mean velocity might not be a consistentboundary condition, it is observed quite often that themodel becomes unstable at open-boundary points. There-fore, a new type of sponge layer has been introduced intothe ROMS source code, which has also been implementedinto the unstructured-grid ocean model FVCOM (see e.g.Chen et al., 2003, 2007; Qi et al., 2009) as explained inKirchner et al. (2020), and which we want to explain herefor completeness, as well.11o stabilize the vertical-mean horizontal velocity com-ponents ( u, v ) at open-boundary grid points, each velocitycomponent is modified according to the following equation: u si = u i c u i ] (A.1) u i denotes the eastward or the northward component, c denotes the sponge-layer parameter, with c = 0 switching-off the sponge layer. And u si denotes the velocity valuebeing further used at the open boundary.In Fig. A.13 two sponge-layer curves are depicted fortwo different sponge-parameter values. It is evident fromthis figure that the modification of the vertical-mean veloc-ity along the open boundary is very small for small veloc-ity values. In this case, the velocity is not affected by thesponge-layer. Depending on the value of the c -parameter,the sponge layer is only effective for larger velocity valueswhich e.g. might be encountered in case of instabilities.This sponge-layer characteristic after Eq. (A.1) is sym-metric around zero, which might be a disadvantage in caseof valid currents at open boundaries which enter or leavethe model domain. In this case, the mean velocity of theopen-boundary currents might be decreased too much bythe sponge-layer. In order to solve this issue, one coulduse a slightly modified version of Eq. (A.1). Suppose themodel should meet a background current velocity at theopen boundary denoted by u bi , then the following equationdampens differences to this background velocity wanted: u si = u i c ( u i − u bi )] (A.2)Again, within this study, the sponge-layer after Eq. (A.1)is used. References
Barth, J.A., Pierce, S.D., Smith, R.L., 2000. A separating coastalupwelling jet at Cape Blanco, Oregon and its connection to theCalifornia Current System. Deep Sea Research Part II: TopicalStudies in Oceanography 47, 783 – 810.Bleck, R., 2002. An oceanic general circulation model framed inhybrid isopycnic - Cartesian coordinates. Ocean Modelling 37, 55– 88.Centurioni, L.R., Ohlmann, J.C., Niiler, P.P., 2008. Permanentmeanders in the california current system. Journal of PhysicalOceanography 38, 1690–1710.Chassignet, E.P., Hurlburt, H.E., Smedstad, O.M., Halliwell, G.R.,Wallcraft, A.J., Metzger, E.J., Blanton, B.O., Lozano, C., Rao,D.B., Hogan, P.J., Srinivasan, A., 2006. Generalized vertical co-ordinates for eddy-resolving global and coastal ocean forecasts.Oceanography 19.Chavez, F.P., Herlien, R.A., Thurmond, G.D., 1994. Oasis - acquisi-tion system for moorings/drifters: Developed for interdisciplinaryscience, with multiple adaptable interfaces; tested on atlas moor-ings in monterey bay, california. Sea technology 35, 51–59.Chavez, F.P., Michisaki, R.P., Blum, M., Chavez, G.M., Friederich,J., Jones, B., Herlien, R., Kieft, B., Hobson, B., Ren, A.S.,Ryan, J., Sevadjian, J.C., Wahl, C., Walz, K.R., Yamahara, K.,Friederich, G.E., Messi´e, M., 2017. Climate variability and change:Response of a coastal ocean ecosystem. Oceanography . Checkley, D.M., Barth, J.A., 2009. Patterns and processes in theCalifornia Current System. Progress in Oceanography 83, 49 –64.Chelton, D.B., 1984. Seasonal variability of alongshore geostrophicvelocity off central California. Journal of Geophysical Research:Oceans 89, 3473 – 3486.Chen, C., Huang, H., Beardsley, R.C., Liu, H., Xu, Q., Cowles, G.,2007. A finite-volume numerical approach for coastal ocean circu-lation studies: comparisons with finite difference models. Journalof Geophysical Research 112, C03018.Chen, C., Liu, H., Beardsley, R.C., 2003. An unstructured, finite-volume, three-dimensional, primitive equation ocean model: ap-plication to coastal ocean and estuaries. Journal of Atmosphericand Oceanic Technology. 20, 159–186.Chenillat, F., Rivire, P., Capet, X., Franks, P.J.S.,Blanke, B., 2013. California Coastal Upwelling On-set Variability: Cross-Shore and Bottom-Up Propaga-tion in the Planktonic Ecosystem. PLOS ONE 8, 1–15.URL: https://doi.org/10.1371/journal.pone.0062281 ,doi: .Collins, C., Garfield, N., Rago, T., Rischmiller, F., Carter, E., 2000.Mean structure of the inshore countercurrent and California un-dercurrent off Point Sur, California. Deep Sea Research Part II:Topical Studies in Oceanography 47, 765 – 782.Collins, C., Paquette, R., Ramp, S., 1996. Annual variability of cur-rents at 350 m above the continental slope of Point Sur, California.CalCOFI Reports 37 , 251 – 257.Collins, C., Pennington, J., Castro, C., Rago, T., Chavez, F., 2003.The California Current system off Monterey, California: physi-cal and biological coupling. Deep Sea Research Part II: TopicalStudies in Oceanography 50, 2389 – 2404.Davis, R., 1985. Drifter Observations of Coastal Sur-face Currents During CODE: The Method and DescriptiveView. Journal of Geophysical Research 90, 4741–4755.doi: .Drake, P.T., Edwards, C.A., Barth, J.A., 2011. Dispersion and con-nectivity estimates along the U.S. west coast from a realistic nu-merical model . Journal of Marine Research 69, 1 – 37.ECMWF/ERA5, 2017. Copernicus Climate Change Service (C3S).ERA5: Fifth generation of ECMWF atmospheric reanalyses ofthe global climate. Copernicus Climate Change Service ClimateData Store (CDS). https://cds.climate.copernicus.eu .Egbert, G.D., Erofeeva, S.Y., 2002. Efficient inverse modeling ofbarotropic ocean tides. Journal of Atmospheric and Oceanic Tech-nology 19, 183 – 204.Fennel, K., Wilkin, J., Levin, J., Moisan, J., O’Reilly, J., Haid-vogel, D., 2006. Nitrogen cycling in the middle atlantic bight:Results from a three-dimensional model and implications for thenorth atlantic nitrogen budget. Global Biogeochemical Cycles 20.GB3007.Fiechter, J., Moore, A.M., Edwards, C.A., Bruland, K.W., Lorenzo,E.D., Lewis, C.V., Powell, T.M., Curchitser, E.N., Hedstrom,K., 2009. Modeling iron limitation of primary production in thecoastal Gulf of Alaska. Deep Sea Research Part II: Topical Studiesin Oceanography 56, 2503 – 2519.Fiechter, J., Santora, J.A., Chavez, F., Northcott, D., Messi, M.,2020. Krill hotspot formation and phenology in the california cur-rent ecosystem. Geophysical Research Letters 47, e2020GL088039.doi: .Gangopadhyay, A., Lermusiaux, P.F., Rosenfeld, L., Robinson, A.R.,Calado, L., Kim, H.S., Leslie, W.G., Haley, P.J., 2011. The Cal-ifornia Current System: A multiscale overview and the develop-ment of a feature-oriented regional modeling system (FORMS).Dynamics of Atmospheres and Oceans 52, 131 – 169.Gruber, N., Frenzel, H., Doney, S.C., Marchesiello, P., Williams,J.C.M., Moisan, J.R., Oram, J.J., Plattner, G.K., Stolzenbach,K.D., 2006. Eddy-resolving simulation of plankton ecosystem dy-namics in the California Current System. Deep-Sea Research I 53,1483 – 1516.Haidvogel, D.B., Arango, H.G., Hedstrom, K., Beckmann, A.,Malanotte-Rizzoli, P., Shchepetkin, A.F., 2000. Model evaluation xperiments in the North Atlantic Basin: Simulations in nonlinearterrain-following coordinates. Dyn. Atmos. Oceans , 239 – 281.Hayward, T.L., Mantyla, A.W., 1990. Physical, chemical and biolog-ical structure of a coastal eddy near Cape Mendocino. Journal ofMarine Research 48, 825–850. doi: .Hickey, B.M., 1979. The California current system - hypotheses andfacts. Progress in Oceanography 8, 191 – 279.Hickey, B.M., 1998. Western North America, tip of Baja Californiato Vancouver Island. The Sea, Wiley, New York 11, 345 – 394.Huyer, A., Barth, J., Kosro, P., Shearman, R., Smith, R., 1998.Upper-ocean water mass characteristics of the California current,Summer 1993. Deep Sea Research Part II: Topical Studies inOceanography 45, 1411 – 1442.Jacox, M.G., Moore, A.M., Edwards, C.A., , Fiechter, J., 2014. Spa-tially resolved upwelling in the California Current System and itsconnections to climate variability. Geophysical Research Letters ,10.1002/2014GL059589.Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D.,Gandin, L., Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y.,Chelliah, M., Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K.C.,Ropelewski, C., Wang, J., Leetmaa, A., Reynolds, R., Jenne, R.,Joseph, D., 1996. The NCEP/NCAR 40-Year Reanalysis Project.Bulletin of the American Meteorological Society 77, 437 – 472.K¨ampf, J., Chapman, P., 2016. Upwelling Systems of the World.Springer International Publishing.Kirchner, J.S., Lettmann, K.A., Schnetger, B., Wolff, J.O., Brum-sack, H.J., 2020. Carbon capture via accelerated weathering oflimestone: Modeling local impacts on the carbonate chemistry ofthe southern North Sea. International Journal of Greenhouse GasControl 92, 102855.Kurian, J., Colas, F., Capet, X., McWilliams, J.C., Chelton, D.B.,2011. Eddy properties in the California Current System. Journalof Geophysical Research: Oceans 116.Levitus, S., 1982. CLIMATOLOGICAL ATLAS OF THE WORLDOCEAN, NOAA Professional Paper 13. U.S. Government PrintingOffice, Washington D.C. , 173pp.Lynn, R.J., Simpson, J.J., 1987. The California Current system:The seasonal variability of its physical characteristics. Journal ofGeophysical Research: Oceans 92, 12947–12966.Marchesiello, P., McWilliams, J.C., Shchepetkin, A., 2003. Equi-librium structure and dynamics of the california current system.Journal of Physical Oceanography 33, 753–783.Messi´e, M., Chavez, F.P., 2017. Nutrient supply, surface currents,and plankton dynamics predict zooplankton hotspots in coastalupwelling systems. Geophysical Research Letters 44, 8979–8986.doi: .Mooers, C.N.K., Robinson, A.R., 1984. Turbulent Jets and Eddiesin the California Current and Inferred Cross-Shore Transports.Science 223, 51 – 53.Nagai, T., Gruber, N., Frenzel, H., Lachkar, Z., McWilliams, J.C.,Plattner, G.K., 2015. Dominant role of eddies and filaments in theoffshore transport of carbon and nutrients in the california currentsystem. Journal of Geophysical Research: Oceans 120, 5318–5341.Newberger, P.A., Allen, J.S., Spitz, Y.H., 2003. Analysis and com-parison of three ecosystem models. Journal of Geophysical Re-search 108.Pawlowicz, R., Beardsley, B., Lentz, S., 2002. Classical tidal har-monic analysis including error estimates in MATLAB using T-TIDE. Computers and Geosciences 28, 929 – 937.Pierce, S., Smith, R., Kosro, P., 1996. Observations of the polewardundercurrent along the eastern boundary of the mid-latitude Pa-cific. EOS, Transactions of the American Geophysical Union 77,F345 (abstract).Pierce, S., Smith, R., Kosro, P., Barth, J., Wilson, C., 2000. Conti-nuity of the poleward undercurrent along the eastern boundary ofthe mid-latitude north Pacific. Deep Sea Research Part II: TopicalStudies in Oceanography 47, 811 – 829.Powell, T.M., Lewis, C.V.W., Curchitser, E.N., Haidvogel, D.B.,Hermann, A.J., Dobbins, E.L., 2006. Results from a three-dimensional, nested biological-physical model of the CaliforniaCurrent System and comparisons with statistics from satellite im- agery. Journal of Geophysical Research 111, C07018.Qi, J., Chen, C., Beardsley, R.C., Perrie, W., Cowles, G.W.,Lai, Z., 2009. An unstructured-grid finite-volume surface wavemodel (fvcom-swave): Implementation, validations and applica-tions. Ocean Modelling 28, 153–166.Reid Jr., J.L., Schwartzlose, R.A., 1962. Direct measurements of theDavidson Current off central California. Journal of GeophysicalResearch (1896-1977) 67, 2491 – 2497.Sakamoto, C.M., Johnson, K.S., Coletti, L.J., Maurer, T.L., Mas-sion, G., Pennington, J.T., Plant, J.N., Jannasch, H.W., Chavez,F.P., 2017. Hourly In Situ Nitrate on a Coastal Mooring: A 15-Year Record and Insights into New Production. OceanographyURL: https://doi.org/10.5670/oceanog.2017.428 .Song, H., Miller, A.J., Cornuelle, B.D., Lorenzo, E.D., 2011. Changesin upwelling and its water sources in the california current systemdriven by different wind forcing. Dynamics of Atmospheres andOceans 52, 170–191.Spitz, Y.H., Newberger, P.A., Allen, J.S., 2003. Ecosystem responseto upwelling off the oregon coast: Behavior of three nitrogen-basedmodels. Journal of Geophysical Research: Oceans 108.Strub, P., James, C., 2000. Altimeter-derived variability of surfacevelocities in the California Current System: 2. Seasonal circulationand eddy statistics. Deep Sea Research Part II: Topical Studiesin Oceanography 47, 831 – 870.Sverdrup, H., Johnson, M., Fleming, R., 1942. The Oceans: TheirPhysics, Chemistry and General Biology. Prentice-Hall, UpperSaddle River, New Jersey .Umlauf, L., Burchard, H., 2003. A generic length-scale equation forgeophysical turbulence models. J. Mar. Res. 61, 235 – 265.Warner, J.C., Sherwood, C.R., Arango, H.G., Signell, R.P., 2005.Performance of four turbulence closure models implemented usinga generic length scale method. Ocean Modelling 8, 81 – 113.Weber, E.D., Moore, T.J., 2013. CORRECTED CONVERSIONALGORITHMS FOR THE CALCOFI STATION GRID ANDTHEIR IMPLEMENTATION IN SEVERAL COMPUTER LAN-GUAGES. CalCOFI Rep. 54.Wilkin, J.L., Arango, H.G., Haidvogel, D.B., Lichtenwalner, C.S.,Durski, S.M., Hedstrom, K.S., 2005. A regional Ocean ModelingSystem for the Long-term Ecosystem Observatory. J. Geophys.Res. 110, C06S91..Song, H., Miller, A.J., Cornuelle, B.D., Lorenzo, E.D., 2011. Changesin upwelling and its water sources in the california current systemdriven by different wind forcing. Dynamics of Atmospheres andOceans 52, 170–191.Spitz, Y.H., Newberger, P.A., Allen, J.S., 2003. Ecosystem responseto upwelling off the oregon coast: Behavior of three nitrogen-basedmodels. Journal of Geophysical Research: Oceans 108.Strub, P., James, C., 2000. Altimeter-derived variability of surfacevelocities in the California Current System: 2. Seasonal circulationand eddy statistics. Deep Sea Research Part II: Topical Studiesin Oceanography 47, 831 – 870.Sverdrup, H., Johnson, M., Fleming, R., 1942. The Oceans: TheirPhysics, Chemistry and General Biology. Prentice-Hall, UpperSaddle River, New Jersey .Umlauf, L., Burchard, H., 2003. A generic length-scale equation forgeophysical turbulence models. J. Mar. Res. 61, 235 – 265.Warner, J.C., Sherwood, C.R., Arango, H.G., Signell, R.P., 2005.Performance of four turbulence closure models implemented usinga generic length scale method. Ocean Modelling 8, 81 – 113.Weber, E.D., Moore, T.J., 2013. CORRECTED CONVERSIONALGORITHMS FOR THE CALCOFI STATION GRID ANDTHEIR IMPLEMENTATION IN SEVERAL COMPUTER LAN-GUAGES. CalCOFI Rep. 54.Wilkin, J.L., Arango, H.G., Haidvogel, D.B., Lichtenwalner, C.S.,Durski, S.M., Hedstrom, K.S., 2005. A regional Ocean ModelingSystem for the Long-term Ecosystem Observatory. J. Geophys.Res. 110, C06S91.