1 Background

2 MPA Time Series

2.1 Overview

The Curated Data Views provide visualizations for time-series of oceanographic, climatological, and ecological data at three spatial scales: individual MPAs in the California network (155 in total), combined (aggregated) MPAs, and the reference bioregion. Whenever possible, oceanographic data products were developed through repeatable, community-developed access protocols and standards. Whenever possible, source data were accessed from NetCDF (Network Common Data Form, .nc) files, a binary data storage format that is widely used for oceanographic and climatological data. NetCDF, developed and maintained by Unidata, is a “self-describing” binary data format which allows for machine-independent portability. Tools for sharing and analyzing netCDF files. Ecological data products are derived from different long-term ecological monitoring datasets which are publicly accessible through the California Ocean Protection Council (OPC) Data Repository. Details of individual datasets and data variables are provided below.

2.2 Data and Source Code Access

Data are accessed and processed through the Research Workspace (Axiom Data Science), a web-based project management and data analysis platform that allows the execution of Python based Jupyter Notebooks. Notebooks run on the RW are in close proximity (infiniband connection) with the storage devices that hold local copies of large remote sensing and model output datasets. This allows users to easily and quickly access large datasets and minimizes internet bandwidth bottle necking. Processing scripts are maintained and documented on the Research Workspace.

2.3 Processing

Data are spatially masked and aggregated by the outline of each reference and MPA site extracted from polygon shapefiles (.shp) of individual MPAs, MPAs combined by bioregion, and bioregions. The shapefile of all individual MPAs was obtained as a publicly available dataset from the California Department of Fish and Wildlife (https://map.dfg.ca.gov/metadata/ds0582.html). MPA polygons from this shapefile were also aggregated by bioregion to create a shapefile of combined MPAs. For non-gridded datasets, a point-in-polygon algorithm is used to check if a coordinate pair is within an region of interest (ROI) using the Shapely and Geopandas Python packages. For gridded datasets, a spatial mask is made for the ROI using the Salem Python package .

After data are spatially organized, time series are extracted into monthly and annual metrics. Generally, the mean, 95th percentile, and the maximum were generated. Generated files are saved as comma separated value (.csv) text files.

2.4 Variables and Source Datasets


The CDIP Monitoring and Prediction (MOPS) model estimates nearshore wave conditions at nodes along the 15 meter isobath. Nearshore conditions are estimated using a non-stationary linear wave refraction model driven by offshore conditions that are measured in real time from in situ wave buoys. The model has been validated against in situ measurements and shown to represent nearshore conditions well in areas with relatively uniform coastal topography.

The model output includes wave spectral data and bulk statistics, including significant wave height (Hs) and dominant wave period (Tp). Wave power is approximated by the product of the squared significant wave height and the dominant period (Hs^2 * Tp). For each MPA and Reference site, MOPS nodes that fall within the boundaries are aggregated temporally into monthly and annual statistics (mean, maximum, and 95th percentile). If there are multiple nodes that fall within the boundaries of a region of interest, they are aggregated to calculate the spatial mean, minimum, and maximum at each time period.

2.4.2 Sea Surface Temperature

Sea Surface Temperature (SST, °C) data are obtained from the California Currently merged satellite-derived 1km dataset (http://spg-satdata.ucsd.edu/ca1km), which is converted and remapped from the global AVHRR OI dataset (Reynolds et al. 2007, http://podaac.jpl.nasa.gov/dataset/NCDC-L4LRblend-GLOB-AVHRR_OI). We downsample the daily dataset to generate monthly and annual temporal mean, maximum, and 95th percentile for each pixel using the resample() function in the python xarray library. We extract spatial subsets of these monthly and annual datasets based on shapefiles for each area of interest: the 155 MPAs in the California network, the combined MPAs for each of four designated bioregions, and the combined state waters for these bioregions. For each of these areas of interest, we calculate a spatial mean, maximum, and minimum SST value for each dataset time point.

2.4.3 COAMPS Gridded Wind Data

Coupled Ocean/Atmosphere Mesoscale Prediction Systems (COAMPS, version 3) is a mesoscale coupled operational atmospheric/ocean model developed and run by the Naval Research Laboratory Marine Meteorology Division. These data are ingested through the CeNCOOS cyber infrastructure and available as a CF-compliant netCDF file on the CeNCOOS THREDDS server. We obtain wind speed and direction variables from u and v wind vectors at 10 meters above sea level from the gridded nowcast output of the model. Scalar wind speed is calculated as the square root of the sum of squares of the u and v vectors and has units of meters per second. Wind direction is calculated using the trigonometric arctan function of the quotient of v divided by u and has units of degrees.

2.4.4 Kelp Canopy Area

Kelp area canopy area values (m2) for giant kelp (Macrocystis pyrifera) and bull kelp (Nereocystis luetkeana) are derived from Landsat 5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper Plus (ETM+), and Landsat 8 Operational Land Imager (OLI) satellite imagery at a 30 x 30 m pixel resolution for all coastal areas of California, including offshore islands. These area values are divided by the pixel area (900 m2) to obtain a percentage canopy cover. The data are publicly available from the Santa Barbara Coastal LTER on the Environmental Data Initiative (EDI) Data Portal.

2.4.5 Net Primary Productivity

Net Primary Productivity (NPP, mg C m-2 day-1) values are calculated from merged ocean color satellite datasets for Chlorophyll a and Photosynthetically Active Radiation (PAR) using the Behrenfeld and Falkowski model (Behrenfeld and Falkowski 1997), adapted for the California Current region (Kahru et al. 2009). The dataset has a daily temporal resolution and a spatial resolution of 4km. We downsample the daily dataset to generate monthly and annual temporal mean, maximum, and 95th percentile for each pixel using the resample() function in the python xarray library. We extracted spatial subsets of these monthly and annual datasets based on shapefiles for each area of interest: the 155 MPAs in the California network, the combined MPAs for each of four designated bioregions, and the combined state waters for these bioregions. For each of these areas of interest, we calculated a spatial mean, maximum, and minimum NPP value for each dataset time point.

2.4.6 KD490

The coefficient of diffuse attenuation of downwelling light at 490 nm (KD490) is a proxy for turbidity, and is obtained as a standard optical parameter from ocean color satellite sensors. These KD490 data are based on the ESA Ocean Colour Climate Change Initiative (OC-CCI) version 4.2 (Sathyendranath et al., 2019, https://esa-oceancolour-cci.org/), using the equation from Lee et al. (2005) equation and bbw from Zhang et al. (2009). The dataset has a daily temporal resolution and a spatial resolution of 4km. We downsample the daily dataset to generate monthly and annual temporal mean, maximum, and 95th percentile for each pixel using the resample() function in the python xarray library. We extracted spatial subsets of these monthly and annual datasets based on shapefiles for each area of interest: the 155 MPAs in the California network, the combined MPAs for each of four designated bioregions, and the combined state waters for these bioregions. For each of these areas of interest, we calculated a spatial mean, maximum, and minimum KD490 value for each dataset time point.

2.4.7 West Coast Upwelling Indices

The Coastal Upwelling Transport Index (CUTI) and the Biologically Effective Upwelling Transport Index (BEUTI) are two new upwelling indices that leverage state-of-the-art ocean models as well as satellite and in situ data to improve upon historically available upwelling indices for the U.S. west coast (Jacox et al. 2018 ). CUTI provides estimates of vertical transport near the coast (i.e., upwelling/downwelling) and was developed as a more accurate alternative to the previously available Bakun Index. BEUTI provides estimates of vertical nitrate flux near the coast (i.e., the amount of nitrate upwelled/downwelled), which may be more relevant than upwelling strength when considering some biological responses.

CUTI and BEUTI values are publicly available as daily indices at 1-degree latitude resolution. For each MPA, we identify the latitude bin that is closest to the center point of the MPA and extract the corresponding upwelling index values. For each bioregion, we calculate the mean of index values from all latitudes within the region.

2.4.8 Oceanographic Indices

The Multivariate El Niño/Southern Oscillation (ENSO) index (MEI.v2) is the time series of the leading combined Empirical Orthogonal Function (EOF) of five different variables (sea level pressure , sea surface temperature, zonal and meridional components of the surface wind, and outgoing longwave radiation) over the tropical Pacific basin)

The Pacific Decadal Oscillation (PDO) is described as a long-lived El Niño-like pattern of Pacific climate variability in the North Pacific basin. We use the NCEI PDO index, based on NOAA’s extended reconstruction of SSTs (ERSST Version 5). It is constructed by regressing the ERSST anomalies against the Mantua PDO index for their overlap period to compute a PDO regression map for the North Pacific ERSST anomalies. The ERSST anomalies are then projected onto that map to compute the NCEI index. It is publicly available as a monthly index.

The Extratropical-based Northern Oscillation Index (NOI) is an index of climate variability based on the difference in sea level pressure anomalies at the North Pacific High in the northeast Pacific and near Darwin, Australia, in a climatologically low SLP region (Schwing et al. 2002). It represents a wide range of tropical and extratropical climate events impacting the north Pacific on intraseasonal, interannual, and decadal scales. It is publicly available as a monthly index.

The Multivariate Ocean Climate Index (MOCI) is a regionally zonal indicator of ocean conditions that uses several of the above indicators including upwelling, sea level, wind, SST, and the MEI, PDO, NOI and NPGO indices (García-Reyes et al. 2017). This provides a climate index that bridges between global, statewide and bioregion-level estimates of variation. It is publicly available as a seasonal index.

2.4.9 CCFRP Angler Surveys

The California Collaborative Fisheries Research Program (CCFRP) monitors 14 MPAs and reference sites using fisheries-independent standardized surveys of catch per unit effort (CPUE) and fish sizes. Mean catch per unit effort (CPUE) for ecologically and recreationally important species is extracted for each MPA and its corresponding reference site

  • Combined Fish
  • Combined Rockfish (Sebastes spp.)
  • Halibut (Paralichthys californicus)
  • California Sheephead (Semicossyphus pulcher)

2.4.10 MARINe Rocky Intertidal Surveys

The Multi-Agency Rocky Intertidal Network (MARINe) conducts long-term monitoring and biodiversity surveys of rocky intertidal sites from Alaska to Baja California, including much of the California coastline. The data are publicly available on the OPC Data Repository. We extract density

  • Barnacle percent cover
  • Mussel percent cover
  • Sea star abundance
  • Black chiton (Katharina tunicata)

2.4.11 PISCO Kelp Forest Surveys

The Partnership for Interdisciplinary Studies of Coastal Oceans (PISCO) has conducted ecological surveys of nearshore kelp forest and rocky reef sites since 1999. SCUBA divers collect data on the density of kelp forest fish invertebrate, and macroalgae species using transect surveys at sites inside and outside MPAs.

Fish Variables Invertebrate Variables
Combined finfish Combined invertebrates
Combined Rockfish (Sebastes spp.) Combined Abalone (Haliotis spp.)
California Sheephead (Semicossyphus pulcher) California Spiny Lobster (Panulirus interruptus)
Combined Basses (Paralabrax spp. ) Sea urchins (Strongylocentrotus and Mesocentrotus spp.)
Combined Crabs

2.4.12 CRFS Recreational Catch Per Unit Effort

The California Recreational Fisheries Survey (CRFS) program is run by the California Department of Fish and Wildlife (CDFW), and collects fishery-dependent data on California’s marine recreational fisheries. Spatially explicit monthly catch and effort estimates are provided for the four fishing modes (private boats, Commercial Passenger Fishing Vessels (CPFVs), beaches/banks, and man-made structures). These data are spatially explicit on a grid of 1 x 1 mile ‘microblocks’ that subdivides the CDFW’s 10 x 10 nautical mile commercial fishing blocks to a finer spatial resolution.

We extract data from all microblocks within a 5km buffer zone surrounding each MPA (i.e., potential spillover areas) and within each bioregion. Mean catch per unit effort (CPUE) for ecologically and recreationally important species is calculated for the 5km zone around each individual MPAs, for the combined 5km zones around all MPAs in a bioregion, and for the entire bioregion.

  • Red Abalone (Haliotis rufescens)
  • Dungeness Crab (Cancer magister)
  • California Spiny Lobster (Panulirus interruptus)
  • California Sheephead (Semicossyphus pulcher)
  • Halibut (Paralichthys californicus)
  • Lingcod (Ophiodon elongatus)
  • Combined Rockfish (Sebastes spp.)

3 Ecological Indicators

3.1 Seascapes

Seascapes are coherent bodies of water at the landscape scale that are characterized by their physical, biological, and chemical properties. These properties drive planktonic presence and abundance that form the base of oceanic food webs which facilitate processes at higher trophic levels that are actionable for ocean and coastal management (Kavanaugh et al. 2016). Unlike static features that often characterize habitats, the variables that comprise seascapes are dynamic in time and space. Thus, seascapes reflect short- and long-term variability in coastal and ocean areas over time and have the potential to predict biological responses relevant to spatially-explicit or adaptive ocean management frameworks (Caldow et al. 2015, Lewison et al. 2015). As seascapes integrate local-scale data to understand landscape-scale processes, they can be used to improve our understanding of connectivity among MPA networks as well as how ecosystem-scale problems like climate change may impact a single MPA or MPAs connected by seascapes.

Seascapes are created from remotely sensed data as well as predictive models. As such, weather and climatic features like cloud cover can prevent us from visualizing parts of the ocean and their corresponding seascapes. This data includes: sea surface temperature (SST), photosynthetically active radiation (PAR), sea surface salinity, absolute dynamic topography (ADT), ice contribution, chromophoric dissolved organic material (CDOM), chlorophyll a (Chl a), MODIS normalized fluorescence line height (nFLH), and the nFLH:Chl a ratio. These data are then modeled using probabilistic self-organizing maps (PrSOM, Anouar et al. 1998) combined with a hierarchical agglomerative classification (HAC, Jain et al. 1987) to achieve a non-linear, topology-preserving data reduction to probabilistically fit one of 33 potential seascape categories, further described at (NOAA MBON, Kavanaugh et al. 2014). Seascapes are presented as either 8-day or monthly composites represented in geographic space at a 5km spatial resolution. The data used in this Shiny app are available in the links below:

If you use any of the seascapes data from this app, please cite: Maria T. Kavanaugh, Matthew J. Oliver, Francisco P. Chavez, Ricardo M. Letelier, Frank E. Muller-Karger, Scott C. Doney (2016) Seascapes as a new vernacular for pelagic ocean monitoring, management and conservation. ICES Journal of Marine Science.

3.2 C-HARM

The California-Harmful Algae Risk Mapping (C-HARM) Model generates nowcasts (same-day) and forecast (one to three day) predictions of harmful algal bloom (HAB) conditions through a combination of 1) circulation models that predict the ocean physics, 2) satellite remote-sensing data of the ocean “color” and chlorophyll patterns, and 3) statistical models for predicting bloom and toxin likelihoods. Specifically, the routine nowcast and forecast products of toxigenic Pseudo-nitzschia blooms and/or domoic acid events are produced by combining:

  1. empirical logistical models (GLMs) with
  2. existing hydrodynamic model simulations (CA-ROMS, 3 km)
  3. enhanced satellite imagery (MODIS-Aqua with gap-filling using Data Interpolating Empirical Orthogonal Functions - DINEOF)
  4. and community (HABMAP)/marine mammal observations (Anderson et al. 2016).

These predictions are generated daily to provide a nowcast and forecasts where you might encounter a Pseudo-nitzschia bloom and/or domoic acid event in real time up to three days in the future. The Ecological Models shiny product shows nowcast data. All C-HARM data can be accessed on the NOAA Coastwatch ERDDAP Server.

Pseudo-nitzschia Bloom Prediction shows the probability that the abundance of toxin-producing species of the diatom Pseudo-nitzschia is at or above the “bloom” threshold of 10,000 cells per liter. A value of 0.6, for example, means there is a 60% predicted probability of Pseudo-nitzschia blooms in that pixel. 0.6 was chosen as a threshold point where predictive accuracy and the probability of detection is optimized. This threshold is based on work in Trainer and Suddleson 2005, Lane et al. 2009, Anderson et al. 2009, 2011, 2016.

Domoic Acid Event Prediction (for particulate DA) shows the probability that the domoic acid concentration in the bulk phytoplankton pool is at or above 500 nanograms per liter (= 0.5 micrograms per L). A value of 0.6, for example, means there is a 60% predicted probability of a toxic event, although there is always the possibility that concentrations lower than 500 ng/L will lead to toxins in shellfish or strandings of marine mammals and birds.

Domoic Acid Toxicity Prediction (for cellular DA) shows the probability that the domoic acid concentration per Pseudo-nitzschia (i.e. how toxic are the algal cells themselves) is at or above 10 picograms per cell (pg/cell). To give a sense of the range, the highest cellular concentrations seen in the environment have not yet exceeded 200 pg/cell in the most toxic cells. A predicted probability value of 0.6, for example, means there is a 60% probability that a 10 pg/cell level of toxicity is present in the phytoplankton, although there is always the possibility that concentrations lower than 10 will lead to toxins in shellfish or strandings of marine mammals and birds. The thresholds for cellular and particulate DA are discussed further in Anderson et al. 2009, 2011, 2016.

3.3 EcoCast

EcoCast is a fisheries sustainability tool that helps fishers and managers evaluate how to allocate fishing effort to maintain target fish catch while minimizing bycatch of protected or threatened species. EcoCast is based on the concept of dynamic ocean management, a new management approach that uses real-time and near real-time data to support management responses that can change in space and time, at scales relevant for animal movement and human use.

Each day the EcoCast tool generates predictions of the spatial distributions of important migratory species, including those targeted for catch by fishers (i.e. swordfish) and those that may be unintentionally captured during commercial fishing activities (bycatch - i.e. leatherback sea turtle, sea lions, blue shark). Each distribution is then weighted, based on prevailing management strategies, to produce the EcoCast Map product.

EcoCast data sources include:

  • sea surface chlorophyll concentration
  • sea surface temperature
  • sea surface winds
  • sea surface height
  • eddy kinetic energy

These data sources are inputted into species distribution models to predict the spatial distributions of important migratory species. Species weightings, which range from -1.0 to 1.0, are set to reflect management priorities and recent catch and bycatch events. A species is assigned a more negative weighting to avoid catch and a more positive weighting when catch is desired.

The resulting EcoCast product is a daily color map. Better for fishing is indicated by increasingly darker purple colors, showing where the likelihood of target catch is good and the likelihood of bycatch is lower. Poorer for fishing is indicated by increasingly darker orange colors, showing where the likelihood of target catch is lower and the likelihood of bycatch is higher.

Metadata for EcoCast can be found here. EcoCast raw netCDF files can be found here.

3.4 Seascapes stacked plots and Shannon Diversity Index calculations

Stacked plots were created by subsetting the Seascapes dataset by MPA. Note that not all MPAs have Seascapes associated with them. The masking or subsetting function used was st_intersects in the sf package in R, which only includes an intersection if the center of the polygon of interest, in this case the Seascape pixel(s), is contained within the MPA polygon. This was not always the case considering how near shore all the MPAs are. Other reasons for missing data include cloud cover and missing salinity parameter in certain parts of the time series.

Shannon diversity indices are also dynamically calculated for the MPA and time period selected. Shannon Diversity Indices for Seascapes by MPAs are also available on a quarter-year time scale.

3.5 Seascapes time series and stacked plots comparisons for MPAs by bioregions

Time series bar plots show the seascape categories as different colored lines across time (x axis) by MPA. These can be visualized as individual MPAs by state or by bioregion. The stacked plots include all the categories of Seascapes by MPA aggregated together over the whole time series. These are also available as individual MPAs by state or bioregion.

3.6 C-HARM Hovmoller plots, heatmap plots, and summary tables

Hovmoller plots include the number of pixels within MPAs in a specific bioregion with high probability of cellular domoic acid (cDA), particulate domoic acid (pDA), and Pseudo nitzschia (PN) counts. Heatmap plots show the number of days per month that at least one pixel within the MPAs in each bioregion had cDA, pDA, and PN probability values surpass a threshold of 0.6, which is considered “high risk�?. The summary tables show the number of days with Seascape data, the number of days that at least one pixel in the MPAs in that bioregion was over the 0.6 threshold, and the normalized percentage of the year where at least one pixel was over the 0.6 threshold for pDA, cDA, and PN.

3.7 C-HARM and EcoCast Risk Maps

One year (2019-2020) time series maps of the areas where there are high HAB probability risk (p>0.6 threshold) and high relative abundance (p>0) of swordfish fisheries bycatch species (i.e. leatherback sea turtle, sea lions, blue shark).

4 Climate Model Outputs

4.1 Overview

The Climate Model Outputs tab provides visualizations of projected climate variables from a Regional Ocean Modeling System (ROMS) coupled with a biogeochemical model (NEMUCSC) based on the North Pacific Ecosystem Model for Understanding Regional Oceanography (NEMURO), and run with three downscaled Earth System Models: (GFDL), (IPSL), (HAD) (Pozo Buil et al. 2021). We obtain model outputs as monthly means of the climate variables from 1980 to 2100 on a (RESOLUTION???) spatial grid and aggregate them into 30-year mean values.

Model Output Variables include:

  • Sea Surface Temperature (deg C)
  • Surface Chlorophyll a (mg m-3)
  • Dissolved Oxygen (mg/L)

(other outputs such as pH and carbonate chemistry variables may be available at a later date)

4.2 Ocean Change Map

The model output datasets are masked to obtain the subset of data within the boundaries of the U.S. EEZ off the California coast. We generate spatial datasets of the change magnitude and percentage change in 30-year means for each climate variable from the projected values obtained from each of the three Earth System Models. Additionally, we generate datasets of standardized change relative to mean change across the California EEZ. We also calculate an “Ensemble Mean” of projected values from the three models, and the range of variation between the three models.

These spatial datasets of projected change are converted to rasters and used to generate a map of with the borders of individual MPAs, bioregions, and National Marine Sanctuaries to allow visualization of projected climate change in these regions of interest.

4.3 MPA Comparisons

We use the datasets of projected change detailed in the previous section to extract values of mean projected change in individual MPAs and other regions of interest. We extract spatial subsets of these datasets based on shapefiles for each area of interest: the 155 MPAs in the California network, the combined MPAs for each of four designated bioregions, and the combined state waters for these bioregions, and calculate the spatial mean and quartile values for these areas. For the three-model ensemble mean, we also calculate the range of the change values across the three Earth System Models.

4.4 Hotspots and Refuges

Here we define a ‘refuge’ to be the areas projected to experience the bottom 5th percentile of change within a reference region. Conversely, a ‘hotspot’ of change is defined as the areas projected to experience the upper 5th percentile of change within the reference region.

We use the datasets of projected change detailed in the “Ocean Change Map” section. The datasets are masked to obtain the subset of data within the area of interest: individual bioregions (North Coast, Central Coast, South Coast, Channel Islands), and the combined California State Waters. For each spatial subset, the threshold values corresponding to the upper and lower 5th percentiles of change values are calculated, and then used to obtain the spatial subset of values in these upper and lower percentiles.

Within each region of interest, we further subset the these ‘refuge’ and ‘hotspot’ datasets to the areas that occur within MPA boundaries. We divide the area of this subset by the area of the total upper and lower percentiles to obtain the percentage of refuge and hotspot areas that are protected by MPAs.

For the combined California State Waters dataset, we also calculate the percentage of refuge and hotspot areas that occur within each bioregion.

5 MPA Connectivity

5.1 WCOFS High-Resolution Nests

A physical ocean modeling system has been developed to investigate potential larval connectivity between and among greater Monterey Bay MPAs. Our implementation uses the Regional Ocean Modeling System (ROMS; http://www.myroms.org) in a triply-nested configuration based on NOAA’s West Coast Operational Forecasting System (WCOFS; https://tidesandcurrents.noaa.gov/ofs/dev/wcofs/wcofs_info.html). The outermost domain is the WCOFS 4 km resolution grid itself and its accompanying ocean state estimate that is updated on a daily basis. The 800 m middle nest spans the central coast from Pt. Buchon to Pt. Arena, and the innermost grid focuses on the greater Monterey Bay with a resolution of 160 m, extending from south of Pt. Sur to north of Pescadero. All domains use the same ROMS configuration and are forced by the same atmospheric fields derived from NOAA’s North American Model (NAM) and used for WCOFS. For the purposes of this effort, all domains are coupled in an online-nested fashion in which fields from each nest influence the solution in the adjacent nest across their coincident boundary on each time-step. Though the data assimilative WCOFS fields are discontinuous once each day, the solutions in the inner nests are continuous, meaning the initial condition on any given day is identical to the final condition of the previous day; WCOFS field discontinuities influence the middle model domain and in turn innermost domain through their boundary interactions.

By assimilating Sea Surface Height (SSH), Sea Surface Temperature (SST), and High Frequency (HF) Radar surface current estimates, the WCOFS fields are more consistent with observations than a free-running model that is not constrained by observation data. As a result, we chose to weakly nudge (with a 7-day relaxation time-scale) the middle and inner grids with oceanographic fields derived from the outermost WCOFS ocean state estimate that is run independently from the nests and referred to as the UCSC re-implementation of WCOFS. This procedure means that the middle and inner nests are only nearly free-running, constrained to ensure that their solutions maintain good agreement with available observations within their small footprint domains. The hydrodynamic model output is now being generated routinely with public availability planned via a UCSC THREDDS server in November 2021.

5.2 Particle Tracking Simulations of Connectivity

WCOFS became operational within NOAA on March 22, 2021. Because marine connectivity estimates are statistical in nature, they benefit from long simulations that exhibit greater variability in ocean conditions. To build more robust statistics, we augmented the operational WCOFS output with a pre-operational version starting on March 10, 2020. For this project, our connectivity calculations are based on 1.5 years of model simulations. The nested model physical simulations continue presently on an ongoing basis.

The ROMS configuration produces physical fields (temperature, salinity, SSH, 3-dimensional velocity, and turbulent mixing coefficients). This output then drives an offline particle tracking code built on the ROMS implementation and used in many previous studies (Drake and Edwards, 2009; Drake et al., 2011; Drake et al., 2013; Drake et al., 2015; Drake et al., 2018). Although the model resolves tidal motion, offline particles are driven by daily averaged currents with an imposed parameterization for the horizontal mixing associated with tides; this approach was chosen to reduce storage needs by a factor of 24. For each larval behavior, 3660 particles (representing larvae of various potential organisms) are released into the model domain in near coastal subregions and in MPA subregions every 12 hours over the 1.5 years of simulation. These organismal representations come with various particle lifetimes (e.g. pelagic larval duration [PLD]) and tendencies of remaining at the surface, or being entrained into the surface boundary layer or below it, or being neutrally buoyant and fully Lagrangian. This set of behaviors enables transport and connectivity exploration across a wide range of possibilities including many described in Drake et al. (2018) pertaining to kelp, rockfish, crabs, urchins, and abalone. Particles are transported by modeled ocean currents and tracked for 90 days. Connectivity statistics describe the probability that a particle leaves one subregion and enters another subregion within a competency window following a pelagic larval duration. Larval behaviors, such as maintenance within the surface mixed layer, maintenance below the surface mixed layer, and diel vertical migration, along with the no-behavior case, are included in our study. Connectivity statistics were calculated for a variety of release dates, pelagic larval durations, competency windows, and behaviors between MPAs, and between MPAs and more general coastal subregions. We assessed the differences in resolution solutions between the 3 km native WCOFS scale and the 800 m and 160 m scale nests.