Accessing BCO-DMO data
log in    
Brought to you by BCO-DMO    

ERDDAP > info > bcodmo_dataset_768655

Title Sum-
Institution Dataset ID
     data   graph     files  public Estimated nitrate d15N modeled using an ensemble of artificial neural networks (EANNs)    ?     I   M   background (external link) RSS Subscribe BCO-DMO bcodmo_dataset_768655

The Dataset's Variables and Attributes

Row Type Variable Name Attribute Name Data Type Value
attribute NC_GLOBAL access_formats String .htmlTable,.csv,.json,.mat,.nc,.tsv,.esriCsv,.geoJson
attribute NC_GLOBAL acquisition_description String For complete methodology, refer to Rafter et al. (2019). In summary:

Data Compilation:\u00a0Nitrate d15N observations were compiled from studies
dating from 1975 to 2018. This global ocean nitrate d15N database was
interpolated using an ensemble of artificial neural networks (EANNs). For the
compiled observed global ocean nitrate d15N data, see the related dataset:

Building the neural network model: We utilize an ensemble of artificial neural
networks (EANNs) to interpolate our global ocean nitrate d15N database,
producing complete 3D maps of the data. By utilizing an artificial neural
network (ANN), a machine learning approach that effectively identifies
nonlinear relationships between a target variable (the isotopic dataset) and a
set of input features (other available ocean datasets), we can fill holes in
our data sampling coverage of nitrate d15N.

Binning target variables (Step 1): We binned the nitrate d15N observations to
the World Ocean Atlas 2009 (WOA09) grid with a 1-degree spatial resolution and
33 vertical depth layers (0-5500 m). When binning vertically, we use the depth
layer whose value is closest to the observation's sampling depth (e.g. the
first depth layer has a value of 0 m, the second of 10 m, and the third of 20
m, so all nitrate isotopic data sampled between 0-5 m fall in the 0 m bin;
between 5-15 m they fall in the 10 m bin, etc.). An observation with a
sampling depth that lies right at the midpoint between depth layers is binned
to the shallower layer. If more than one raw data point falls in a grid cell
we take the average of all those points as the value for that grid cell.
Certain whole ship tracks of nitrate d15N data were withheld from binning to
be used as an independent validation set.

Obtaining input features (Step 2): Our input dataset contains a set of
climatological values for physical and biogeochemical ocean parameters that
form a non-linear relationship with the target data. We have six input
features including objectively analyzed annual-mean fields for temperature,
salinity, nitrate, oxygen, and phosphate taken from the WOA09
at 1-degree resolution. Additionally, daily chlorophyll data from Modis Aqua
for the period Jan-1-2003 through Dec-31-2012 is averaged and binned to the
WOA09 grid (as described in Step 1) to produce an annual climatological field
of chlorophyll values, which we then log transform to reduce their dynamic

The choice of these specific input features was dictated by our desire to
achieve the best possible R2 value on our internal validation sets (Step 4).
Additional inputs besides those we included, such as latitude, longitude,
silicate, euphotic depth, or sampling depth either did not improve the R2
value on the validation dataset or degraded it, indicating that they are not
essential parameters for characterizing this system globally. By opting to use
the set of input features that yielded the best results for the global oceans,
we potentially overlooked combinations of inputs that perform better at
regional scales. However, given the scarcity of d15N data in some regions, it
is not possible to ascribe the impact of a specific combination of input
features versus the impact of available d15N data, which may not be
representative of the region's climatological state, to the relative model
performance in these regions.

Training the ANN (Step 3): The architecture of our ANN consists of a single
hidden layer, containing 25 nodes, that connects the biological and physical
input features (discussed in Step 2) to the target nitrate isotopic variable
(as discussed in Step 1). The role of the hidden layer is to transform input
features into new features contained in the nodes. These are given to the
output layer to estimate the target variable, introducing nonlinearities via
an activation function. The number of nodes in this hidden layer, as well as
the number of input features, determines the number of adjustable weights (the
free parameters) in the network. For complete information, refer to Rafter et
al. (2019).

Validating the ANN (Step 4):To ensure good generalization of the trained ANN,
we randomly withhold 10% of the d15N data to be used as an internal validation
set for each network. This is data that the network never sees, meaning it
does not factor into the cost function, so it works as a test of the ANN's
ability to generalize. This internal validation set acts as a gatekeeper to
prevent poor models from being accepted into the ensemble of trained networks
(see Step 5). A second, independent or 'external' validation set, composed of
complete ship transects from the high and low latitude ocean were omitted from
binning in Step 1 and used to establish the performance of the entire
ensemble. Our rationale for using complete ship transects is the following. If
we randomly chose 10% of observations to perform an external validation, this
dataset will be from the same cruises as the wider data. In other words,
despite being randomly selected, the validating observational dataset will be
highly correlated geographically. Contrast this with validating the EANN
results with observations from whole research cruises in unique geographic
regions\u2014areas where the model has not \"learned\" anything about nitrate.
We therefore argue that these observations from whole ship tracks therefore
provide a more difficult test of the model.

Forming the Ensemble (Step 5):\u00a0The ensemble is formed by repeating Steps
3 to 4 (using a different random 10% validation set) until we obtain 25
trained networks for the nitrate d15N dataset. A network is admitted into the
ensemble if it yields an R\u00b2 value greater than 0.81 on the validation
dataset.\u00a0For complete information, refer to Rafter et al. (2019).
attribute NC_GLOBAL awards_0_award_nid String 766422
attribute NC_GLOBAL awards_0_award_number String OCE-1658392
attribute NC_GLOBAL awards_0_data_url String http://www.nsf.gov/awardsearch/showAward.do?AwardNumber=1658392 (external link)
attribute NC_GLOBAL awards_0_funder_name String NSF Division of Ocean Sciences
attribute NC_GLOBAL awards_0_funding_acronym String NSF OCE
attribute NC_GLOBAL awards_0_funding_source_nid String 355
attribute NC_GLOBAL awards_0_program_manager String Dr Simone Metz
attribute NC_GLOBAL awards_0_program_manager_nid String 51479
attribute NC_GLOBAL cdm_data_type String Other
attribute NC_GLOBAL comment String Global modeled nitrate d15N
PI: Patrick Rafter (UC Irvine)
Co-PIs: Aaron Bagnell (UCSB), Dario Marconi (Princeton), & Timothy DeVries (UCSB)
Version date: 28-May-2019
attribute NC_GLOBAL Conventions String COARDS, CF-1.6, ACDD-1.3
attribute NC_GLOBAL creator_email String info at bco-dmo.org
attribute NC_GLOBAL creator_name String BCO-DMO
attribute NC_GLOBAL creator_type String institution
attribute NC_GLOBAL creator_url String https://www.bco-dmo.org/ (external link)
attribute NC_GLOBAL data_source String extract_data_as_tsv version 2.3 19 Dec 2019
attribute NC_GLOBAL date_created String 2019-05-28T17:50:26Z
attribute NC_GLOBAL date_modified String 2019-06-17T20:08:40Z
attribute NC_GLOBAL defaultDataQuery String &time<now
attribute NC_GLOBAL doi String 10.1575/1912/bco-dmo.768655.1
attribute NC_GLOBAL Easternmost_Easting double 359.5
attribute NC_GLOBAL geospatial_lat_max double 83.5
attribute NC_GLOBAL geospatial_lat_min double -79.5
attribute NC_GLOBAL geospatial_lat_units String degrees_north
attribute NC_GLOBAL geospatial_lon_max double 359.5
attribute NC_GLOBAL geospatial_lon_min double 0.5
attribute NC_GLOBAL geospatial_lon_units String degrees_east
attribute NC_GLOBAL geospatial_vertical_max double 5500.0
attribute NC_GLOBAL geospatial_vertical_min double 0.0
attribute NC_GLOBAL geospatial_vertical_positive String down
attribute NC_GLOBAL geospatial_vertical_units String m
attribute NC_GLOBAL infoUrl String https://www.bco-dmo.org/dataset/768655 (external link)
attribute NC_GLOBAL institution String BCO-DMO
attribute NC_GLOBAL keywords String bco, bco-dmo, biological, chemical, d15, d15N, d15N_stdev, data, dataset, depth, deviation, dmo, erddap, latitude, longitude, management, oceanography, office, preliminary, standard, standard deviation, stdev
attribute NC_GLOBAL license String https://www.bco-dmo.org/dataset/768655/license (external link)
attribute NC_GLOBAL metadata_source String https://www.bco-dmo.org/api/dataset/768655 (external link)
attribute NC_GLOBAL Northernmost_Northing double 83.5
attribute NC_GLOBAL param_mapping String {'768655': {'latitude': 'flag - latitude', 'depth': 'flag - depth', 'longitude': 'flag - longitude'}}
attribute NC_GLOBAL parameter_source String https://www.bco-dmo.org/mapserver/dataset/768655/parameters (external link)
attribute NC_GLOBAL people_0_affiliation String University of California-Irvine
attribute NC_GLOBAL people_0_affiliation_acronym String UC Irvine
attribute NC_GLOBAL people_0_person_name String Patrick Rafter
attribute NC_GLOBAL people_0_person_nid String 615040
attribute NC_GLOBAL people_0_role String Principal Investigator
attribute NC_GLOBAL people_0_role_type String originator
attribute NC_GLOBAL people_1_affiliation String University of California-Santa Barbara
attribute NC_GLOBAL people_1_affiliation_acronym String UCSB
attribute NC_GLOBAL people_1_person_name String Aaron Bagnell
attribute NC_GLOBAL people_1_person_nid String 768632
attribute NC_GLOBAL people_1_role String Co-Principal Investigator
attribute NC_GLOBAL people_1_role_type String originator
attribute NC_GLOBAL people_2_affiliation String University of California-Santa Barbara
attribute NC_GLOBAL people_2_affiliation_acronym String UCSB
attribute NC_GLOBAL people_2_person_name String Timothy DeVries
attribute NC_GLOBAL people_2_person_nid String 766426
attribute NC_GLOBAL people_2_role String Co-Principal Investigator
attribute NC_GLOBAL people_2_role_type String originator
attribute NC_GLOBAL people_3_affiliation String Princeton University
attribute NC_GLOBAL people_3_person_name String Dario Marconi
attribute NC_GLOBAL people_3_person_nid String 768634
attribute NC_GLOBAL people_3_role String Co-Principal Investigator
attribute NC_GLOBAL people_3_role_type String originator
attribute NC_GLOBAL people_4_affiliation String University of California-Irvine
attribute NC_GLOBAL people_4_affiliation_acronym String UC Irvine
attribute NC_GLOBAL people_4_person_name String Patrick Rafter
attribute NC_GLOBAL people_4_person_nid String 615040
attribute NC_GLOBAL people_4_role String Contact
attribute NC_GLOBAL people_4_role_type String related
attribute NC_GLOBAL people_5_affiliation String Woods Hole Oceanographic Institution
attribute NC_GLOBAL people_5_affiliation_acronym String WHOI BCO-DMO
attribute NC_GLOBAL people_5_person_name String Shannon Rauch
attribute NC_GLOBAL people_5_person_nid String 51498
attribute NC_GLOBAL people_5_role String BCO-DMO Data Manager
attribute NC_GLOBAL people_5_role_type String related
attribute NC_GLOBAL project String Fe Cycle Models and Observations
attribute NC_GLOBAL projects_0_acronym String Fe Cycle Models and Observations
attribute NC_GLOBAL projects_0_description String NSF Award Abstract:
Tiny marine organisms called phytoplankton play a critical role in Earth's climate, by absorbing carbon dioxide from the atmosphere. In order to grow, these phytoplankton require nutrients that are dissolved in seawater. One of the rarest and most important of these nutrients is iron. Even though it is a critical life-sustaining nutrient, oceanographers still do not know much about how iron gets into the ocean, or how it is removed from seawater. In the past few years, scientists have made many thousands of measurements of the amount of dissolved iron in seawater, in environments ranging from the deep sea, to the Arctic, to the tropical oceans. They found that the amount of iron in seawater varies dramatically from place to place. Can this data tell us about how iron gets into the ocean, and how it is ultimately removed? Yes. In this project, scientists working on making measurements of iron in seawater will come together with scientists who are working on computer models of iron inputs and removal in the ocean. The goal is to work together to create a program that allows our computer models to "learn" from the data, much like an Artificial Intelligence program. This program will develop a "best estimate" of where and how much iron is coming into the ocean, how long it stays in the ocean, and ultimately how it gets removed. This will lead to a better understanding of how climate change will impact the delivery of iron to the ocean, and how phytoplankton will respond to climate change. With better climate models, society can make more informed decisions about how to respond to climate change. The study will also benefit a future generation of scientists, by training graduate students in a unique collaboration between scientists making seawater measurements, and those using computer models to interpret those measurements. Finally, the project aims to increase the participation of minority and low-income students in STEM (Science, Technology, Engineering, and Mathematics) research, through targeted outreach programs.
Iron (Fe) is an important micronutrient for marine phytoplankton that limits primary productivity over much of the ocean; however, the major fluxes in the marine Fe cycle remain poorly quantified. Ocean models that attempt to synthesize our understanding of Fe biogeochemistry predict widely different Fe inputs to the ocean, and are often unable to capture first-order features of the Fe distribution. The proposed work aims to resolve these problems using data assimilation (inverse) methods to "teach" the widely used Biogeochemical Elemental Cycling (BEC) model how to better represent Fe sources, sinks, and cycling processes. This will be achieved by implementing BEC in the efficient Ocean Circulation Inverse Model and expanding it to simulate the cycling of additional tracers that constrain unique aspects of the Fe cycle, including aluminum, thorium, helium and Fe isotopes. In this framework, the inverse model can rapidly explore alternative representations of Fe-cycling processes, guided by new high-quality observations made possible in large part by the GEOTRACES program. The work will be the most concerted effort to date to synthesize these rich datasets into a realistic and mechanistic model of the marine Fe cycle. In addition, it will lead to a stronger consensus on the magnitude of fluxes in the marine Fe budget, and their relative importance in controlling Fe limitation of marine ecosystems, which are areas of active debate. It will guide future observational efforts, by identifying factors that are still poorly constrained, or regions of the ocean where new data will dramatically reduce remaining uncertainties and allow new robust predictions of Fe cycling under future climate change scenarios to be made, ultimately improving climate change predictions. A broader impact of this work on the scientific community will be the development of a fast, portable, and flexible global model of trace element cycling, designed to allow non-modelers to test hypotheses and visualize the effects of different processes on trace metal distributions. The research will also support the training of graduate students, and outreach to low-income and minority students in local school districts.
attribute NC_GLOBAL projects_0_end_date String 2020-06
attribute NC_GLOBAL projects_0_name String Collaborative research: Combining models and observations to constrain the marine iron cycle
attribute NC_GLOBAL projects_0_project_nid String 766423
attribute NC_GLOBAL projects_0_start_date String 2017-07
attribute NC_GLOBAL publisher_name String Biological and Chemical Oceanographic Data Management Office (BCO-DMO)
attribute NC_GLOBAL publisher_type String institution
attribute NC_GLOBAL sourceUrl String (local files)
attribute NC_GLOBAL Southernmost_Northing double -79.5
attribute NC_GLOBAL standard_name_vocabulary String CF Standard Name Table v55
attribute NC_GLOBAL summary String We utilize an ensemble of artificial neural networks (EANNs) to interpolate our global ocean nitrate d15N database, producing complete 3D maps of the data. By utilizing an artificial neural network (ANN), a machine learning approach that effectively identifies nonlinear relationships between a target variable (the isotopic dataset) and a set of input features (other available ocean datasets), we can fill holes in our data sampling coverage of nitrate d15N.
attribute NC_GLOBAL title String Estimated nitrate d15N modeled using an ensemble of artificial neural networks (EANNs)
attribute NC_GLOBAL version String 1
attribute NC_GLOBAL Westernmost_Easting double 0.5
attribute NC_GLOBAL xml_source String osprey2erddap.update_xml() v1.3
variable latitude   double  
attribute latitude _CoordinateAxisType String Lat
attribute latitude _FillValue double NaN
attribute latitude actual_range double -79.5, 83.5
attribute latitude axis String Y
attribute latitude bcodmo_name String latitude
attribute latitude colorBarMaximum double 90.0
attribute latitude colorBarMinimum double -90.0
attribute latitude description String Latitude in degrees north
attribute latitude ioos_category String Location
attribute latitude long_name String Latitude
attribute latitude nerc_identifier String https://vocab.nerc.ac.uk/collection/P09/current/LATX/ (external link)
attribute latitude standard_name String latitude
attribute latitude units String degrees_north
variable longitude   double  
attribute longitude _CoordinateAxisType String Lon
attribute longitude _FillValue double NaN
attribute longitude actual_range double 0.5, 359.5
attribute longitude axis String X
attribute longitude bcodmo_name String longitude
attribute longitude colorBarMaximum double 180.0
attribute longitude colorBarMinimum double -180.0
attribute longitude description String Longitude in degrees East
attribute longitude ioos_category String Location
attribute longitude long_name String Longitude
attribute longitude nerc_identifier String https://vocab.nerc.ac.uk/collection/P09/current/LONX/ (external link)
attribute longitude standard_name String longitude
attribute longitude units String degrees_east
variable depth   double  
attribute depth _CoordinateAxisType String Height
attribute depth _CoordinateZisPositive String down
attribute depth _FillValue double NaN
attribute depth actual_range double 0.0, 5500.0
attribute depth axis String Z
attribute depth bcodmo_name String depth
attribute depth colorBarMaximum double 8000.0
attribute depth colorBarMinimum double -8000.0
attribute depth colorBarPalette String TopographyDepth
attribute depth description String Depth
attribute depth ioos_category String Location
attribute depth long_name String Depth
attribute depth nerc_identifier String https://vocab.nerc.ac.uk/collection/P09/current/DEPH/ (external link)
attribute depth positive String down
attribute depth standard_name String depth
attribute depth units String m
variable d15N   float  
attribute d15N _FillValue float NaN
attribute d15N actual_range float 0.98889, 26.663
attribute d15N bcodmo_name String dN15_NO3
attribute d15N description String modeled nitrate d15N
attribute d15N long_name String D15 N
attribute d15N units String per mil
variable d15N_stdev   float  
attribute d15N_stdev _FillValue float NaN
attribute d15N_stdev actual_range float 0.036923, 15.627
attribute d15N_stdev bcodmo_name String dN15_NO3
attribute d15N_stdev description String standard deviation
attribute d15N_stdev long_name String D15 N Stdev
attribute d15N_stdev units String per mil

The information in the table above is also available in other file formats (.csv, .htmlTable, .itx, .json, .jsonlCSV1, .jsonlCSV, .jsonlKVP, .mat, .nc, .nccsv, .tsv, .xhtml) via a RESTful web service.

ERDDAP, Version 2.02
Disclaimers | Privacy Policy | Contact