SELECT

DownloadnetCDF files by variableby dataset

GenerateCharts

Symbolize

Summarize

Generate3D Point Mesh

ConvertNet CDFs toSingle CSV filesby dataset

Variable DatasetCSV

Import CSVas table tofGDB

Convert dateto standardformat

JAN

Make XYEvent Layer

Define LocalCoordinateSystem

Feature Class toFeature Class

Piecewise 3DInterpolation(each variable)

3D Visualization,Summary Plotsand Tables

Local EMUs

LocalizedEMUClusters

0

-1,000

-2,000

-3,000

-4,000

4 8 12 16 20 24

Dep

th

Temperature

Scatter Plot of Temperature by Depth

R2 = 0.76

A 250 square kilometer grid with 5,518 points

Counts of EMUs within the Study Area

Top 3 Prominent EMUs

Number of ObservationsTemperature: 4,983,203

Salinity: 2,953,802

Oxygen 96,824

30

200176

820

33 10

337 273

568

350

400

600

800

6 10 13 19 26 33 36 37

Co

unt

EMU Number0 13 1

EMU N

37 250500555500555050

NumberN

5

1N

7 84

99 26N bN

3

5

3

19N

7 8484

56855

100

568

3 33

56

33

56

1111100

3

64

36 37

0000000000000000000000000000000 646446446464664

Clustering resulted in 8 prominent clusters representing over 98% of the sample. Cluster 17 and 1 make up the majority of surficial waters.

The distinct surficial water masses are merging around Point Conception, and this is noticeable. The cooler water from the north is colliding with the southern waters and mixing in this region.

1

2 4

53

Create point mesh4

Populate point mesh with seasonal data by interpolating to a 3D mesh using the Linear Interpolation Method

5

Cluster using the 12 attributes6

Results

8

0

-1,000

-2,000

2 4 6 8 10 12 14 16

Distribution of attribute values

Scatter Plot of Temperature by Depth

R2 = 0.52

Plot of Temperature by Time

200,000

400,000

600,000

01896(2/1)

2016(2/1)

1916 1936 1956 1976 1996

Distribution of attribute values

Collection Date

Histogram of Temperature Values

MeanMedian

Standard Deviation1,000,000

2,000,000

3,000,000

0

6.3112.56

18.825.05 37.51 50.03

31.29 43.78 56.2762.52

68.7675.01

81.2587.5

99.99

93.740.07

Distribution of attribute values Temperature

Temperature

Use charts to help identify and remove outliers

Create charts for each one of the datasets1

2 Use convert time field to add months as an attribute to the table (converted from Date)

Oxygen Measurements by Month

10,000

20,000

30,000

01 2 3 4 5 6 7 8 9 10 11 12

Month

Salinity Measurements by Month

100,000200,000300,000400,000

01 2 3 4 5 6 7 8 9 10 11 12

Month

Temperature Measurements by Month

200,000

400,000

600,000

800,000

01 2 3 4 5 6 7 8 9 10 11 12

Tem

per

atur

eSa

linity

Oxy

gen

Month

3

Dec

Nov

Oct

Sep

Aug

Jul Jun

May

Ap

r

Mar

Feb

Jan

Temperature observations by month for past 10 years

0 - 426

2017201620152014201320122011201020092008

426 - 1,4941,494 - 3,2403,240 - 6,8426,842 - 16,163

Histogram of Temperature Values50,000

40,000

30,000

20,000

MeanMedian

Standard Deviation

10,000

0

2.424.35

6.278.19 12.04 15.89

10.12 13.96 17.8119.73

21.6523.58

25.527.42 31.27

29.350.5

Co

unt

Temperature

25 m

100 m Surface Area

Volume

Thickness

Unit Bottom

Unit Top

UnitMIddle

100 m5500 m

0 m

5500 m

0 m

10 m

5 m

EMU 3D Point MeshFramework

World Ocean Atlas EMU Points Feature Attributes

Depth LevelTemperatureSalinityDissolved OxygenNitrateSilicatePhosphate

MODIS ocean color

GeomorphologyBaseGeomorphologyFeaturesSurfaceAreaVolume, SpecialCases

PointIDQuarterIDUnitTop (m)UnitMiddle (m)UnitBottom (m)Thickness (m)ThicknessPos (m)EMUIDEMU Name

19

18

8

21 24

11

926

10

33

29

3 36

37

1413

5

2325

303135

00

Surface Area (km2)

Dep

th (m

)

AB

YSS

OPE

LAG

ICB

ATH

YPE

LAG

ICM

ESO

PELA

GIC

EPIP

ELA

GIC

100million

200million

300million

200

400

600

800

1000

1200

1400

1600

1800

2000

3000

4000

5000

5500

Hi-Resolution Local Ecological Marine UnitsA workflow for generating a three dimensional mapping of the ocean

WorkflowWhy? FINDINGS: Hawaii

FINDINGS:

Santa Barbara

Sayre, R.G., Wright, D.J., Breyer, S.P., Butler, K.A, Van Graafeiland, K., et al., 2017. Oceanography, 30(1), doi:10.5670/oceanog.2017.116

To maintain a healthy ocean and support sustainability, it is necessary to have a baseline for understanding the ocean’s ecosystems and a framework for detecting change.

In support of that goal, Sayre et al. (2017) developed a three-dimensional mapping of the oceans based on physiochemical environmental data and created a database of over 52 million points that depict the global ocean in x, y, and z dimensions. These points were clustered into 37 distinct volumetric region units, called ecological marine units (EMUs).

The horizontal resolution of the EMU mesh is a ¼° x ¼° grid (~27 km2 at the equator). In the vertical dimension, mesh points are located at variable depth intervals, ranging from 5 m increments near the surface to 100 m increments at depth. A total of 102 depth zones extend to 5,500 m.

While the EMUs have proven useful in a variety of global scale analyses, Sayre et al. recognize their limitations at other spatial scales stating, “global analyses are useful for macroscale comparisons of ocean regions, local management strategies and policies will require appropriately scaled geographic assessment and accounting units” (p. 91).

This poster describes a workflow that addresses the scalability issues of the global EMU dataset. The workflow shows how to build, visualize, analyze and share highly localized EMUs using data from both NOAA’s World Ocean Database and your local in situ observations.

In this example, the grid is not dense enough, and we want to cluster with seasonality.

Project coodinate system:

UTM Zone 10N

Number of global EMU’s within the Study Area: 57

Number of different EMU classifications: 9

Shallow, Cool, Normal Salinity, Moderate Oxygen, Medium Nitrate, Low Phosphate, Low Silicate

Moderate Depth, Very Cold, Normal Salinity, Very Low Oxygen, High Nitrate, Medium Phosphate, Medium Silicate

Moderate Depth, Cold, Normal Salinity, Very Low Oxygen, High Nitrate, Low Phosphate, Low Silicate

Use a definition query on the data to select seasons: https://www.timeanddate.com/calendar/aboutseasons.html

SQL Query = Sample_Month IN (3, 4, 5) for SpringSQL Query = Sample_Month IN (6, 7, 8) for SummerSQL Query = Sample_Month IN (9, 10, 11) for FallSQL Query = Sample_Month IN (12, 1, 2) for Winter

3 values: Temperature, Salinity, Oxygen, by 4 Seasons

These observations were taken at depths ranging from the surface to 4,572 meters between the years of 1928 and 2017. Nearly a century of data!

In this example we used the workflow to help understand Oxygen (203) Temperature (228,475) and Salinity (80,227)observations

Project coodinate system: UTM Zone 5N

Number of global EMU’s within the Study Area: 16

Number of different EMU classifications: 9

Data were obtained from the NOAA World Ocean Database through NOAA Select. In this workflow individual netCDF files are converted to a single feature class with attributes representing the sampled variable. Depth information is also translated to corresponding z values in the feature class.

We were also able to incorporate elevation (depth) data from NOAA's Coastal relief modeling to help derive the depths used to create a uniform 3D point Mesh.

https://www.ngdc.noaa.gov/mggcoastal/crm.html Volume 10- Hawaii

This particular point mesh has a 25 square kilometer horizontal cell size (user specified) using the WOA Standard (Oceanic) depth convention (also user specified) and is comprised of 164,019 points from the surface to 3,900 meters in depth

Data can be symbolized using Cluster ID

Now the temperature, salinity, and oxygen observations can be used to interpolate those values to the 3D point mesh.

Once the values are interpolated to the 3D Point Mesh, clustering can be used to group the data based on attribute values.

In this scenario the software decides the appropriate number of clusters (after data valuation). From here a table and supporting graphics can be generated showing the clustering results.

In this example we classified based on temperature, oxygen and salinity. 29 clusters were determined.

How?The EMU data set is an open-access resource available in both open data (e.g. an OGC geopackage) and standard commercial GIS formats (Esri file geodatabase).

We present a set of preconfigured steps which encapsulate a GIS workflow , a form of guided data analysis, which yields local EMUs.

The guided data analysis is flexible and provides measures that promote quality assurance and customization along the way.

Link to Project and Workflow: http://esriurl.com/LocalEMU

1 x 1 degree area off the west coast of Hawaii.

3 x 2 degree area off the coast of Southern California.

The Process

Data values are in Celsius. There are some apparent outliers that are identifiable in the graph and can be removed prior to clustering. You have the ability to trace these outliers back to the source dataset.

This graph shows the distribution of depth values associated with temperature. Most values are found between 0 and 1,400 meters. You have the ability to interact with the data and identify which values in the graph correspond with the spatial locations in the map.

This shows that there is a huge history of data here. It might be best to focus or work with the most recent data, collected within the last 30 years?