CRWR Online Report 2000-5
A GIS Based Analysis of the Benthic Community in the Western
Arctic Ocean
by
J?na Finnd?s J?nsd?ttir, M.S.E.
Graduate Research Assistant
and
David R. Maidment, PhD.
Kenneth H. Dunton, PhD.
Principal Investigators
August 2000
CENTER FOR RESEARCH IN WATER RESOURCES
Bureau of Engineering Research  The University of Texas at Austin
J.J. Pickle Research Campus  Austin, TX 78712-4497
This document is available online via World Wide Web at
http://www.crwr.utexas.edu/online.html
A GIS Based Analysis of the Benthic Community in the Western
Arctic Ocean
by
J?na Finnd?s J?nsd?ttir, B.S. Geophysics
Report
Presented to the Faculty of the
Department of Environmental and Water Resources Engineering
of The University of Texas at Austin
in Partial Fulfillment
of the Requirements
for the Degree of
Master of Science in Engineering
The University of Texas at Austin
August 2000
This report is dedicated to my family, for always supporting me
my mom and dad, Bj?rg Bjarnad?ttir and ?rni J?nsson
my sisters, Anna Margr?t, Bjarney, Fr??a and Lena
my nieces, ?sd?s Bj?rg, Lilja D?gg and ?rn? R?n
and my late grandfather, J?n Gu?mundsson
iv
Acknowledgements
I want to thank all those who helped me throughout this research. Thanks
to my advisor Dr. David Maidment, Dr. Kenneth Dunton and Susan Schonberg
for their friendship, guidance and for the opportunity to work on this exciting
research.
Thanks to all my friends at UT that have been my family for the past two
years. Thank you Connie, Gerardo and Seung Hi, my best Texan, Mexican and
Korean friends. Thank you Mary and Teri for the exciting traveling, Big Bend,
Guadalupe Mountains, Guatemala, and Copper Canyon. Thank you Katherine for
being an awesome housemate and a friend.
My Icelandic friends here in Austin have given me the chance to jump into
a little Icelandic world when home seems so far away. Takk Rakel, Siggi, Gy?a,
J?n, Gr?ta, Brigitta, Stebbi og Maja.
Finally, I?m forever grateful to everyone at home for all the e-mails and
phone calls. Iceland is a lot closer to my Texas world when I can keep track of
what y?all are doing.
This research was funded by the National Science Foundation.
August 7, 2000
v
Abstract
A GIS Based Analysis of the Benthic Community in the Western
Arctic Ocean
J?na Finnd?s J?nsd?ttir, M.S.E.
The University of Texas at Austin, 2000
Supervisor: David R. Maidment
Knowledge of the spatial distribution of benthos in the Western Arctic
Ocean gives valuable information of the effects of global change on the ecology
of the Arctic Ocean. The goal of this research was to construct a geospatial
database and analyze the benthic community in the Western Arctic Ocean using
Geographical Information Systems (GIS). The study area includes the Beaufort,
Chukchi, East Siberian and Northern Bering Seas. Benthic sample data were
assembled in a relational database and analyzed in ArcView and Arc/Info. A
Geostatistical Analyst extension to ArcMap was used to interpolate benthic
biomass with kriging techniques. The kriging model shows regions of high
benthic biomass in the Bering and Chukchi Sea, where the warm nutrient rich
waters of the Pacific Ocean mix with the Arctic Ocean. The benthic biomass is
significantly lower in the Beaufort Sea, while regions of relatively high benthic
biomass appear to exist in the vicinity of fresh water inflow from major rivers.
vi
Table of Contents
List of Tables......................................................................................................ix
List of Figures......................................................................................................x
Chapter 1: Introduction ........................................................................................1
1.1 Background ...........................................................................................1
1.2 Study Objectives ...................................................................................2
1.3 Study Area ............................................................................................4
1.4 Report Outline.......................................................................................5
Chapter 2: Literature Review ..............................................................................6
2.1 Benthic Community studies...................................................................6
2.1.1 Southwestern Beaufort Sea, Broad .............................................6
2.1.2 Southwestern Beaufort Sea, Carey .............................................7
2.2 Geostatistical Analyst ............................................................................7
2.2.1 Deterministic Interpolation.........................................................8
2.2.2 Geostatistical Techniques...........................................................9
Chapter 3: Data.................................................................................................12
3.1 Benthic Data........................................................................................12
3.1.1 Benthic data from NODC.........................................................13
3.1.2 Benthic Data from other Sources..............................................15
3.2 Digital Bathymetry..............................................................................19
3.2.1 ETOPO5 5-minute Gridded Elevation Data..............................19
3.2.2 Global Bathymetric Prediction for Ocean Modelling and
Marine Geophysics ..................................................................20
3.2.3 International Bathymetric Chart of the Arctic Ocean
(IBCAO)..................................................................................22
3.2.4 Global Digital Elevation Model, GTOPO30.............................23
3.3 GIS Files for Orientation .....................................................................24
3.3.1 The NOAA Nautical Charts .....................................................25
vii
3.3.2 Digital Chart of The World ......................................................26
3.4 Ice-Ocean Model for the Arctic Ocean.................................................26
Chapter 4: Methodology....................................................................................29
4.1 Map Projection....................................................................................29
4.2 Preparation of GIS Layers for Orientation ...........................................31
4.2.1 Building a Frame and a Latlong Grid .......................................31
4.2.2 Layers from Digital Chart of the World....................................33
4.2.3 NOAA Nautical Charts ............................................................34
4.3 Digital Bathymetry/Elevation Basemap ...............................................36
4.3.1 Importing the Bathymetry Grids...............................................36
4.3.2 Assembling the Basemap .........................................................38
4.4 Processing the Arctic Ice-Ocean Model Output....................................41
4.5 The Benthic Relational Database .........................................................45
Chapter 5: Results.............................................................................................50
5.1 Digital Bathymetry Model...................................................................50
5.2 Interpolation of Benthic Biomass.........................................................51
5.2.1 Interpolation Using Topogrid ...................................................53
5.2.2 Interpolation Using the Geostatistical Analyst..........................54
5.2.3 Interpolation of the Entire Benthic Biomass Data Set ...............61
5.2.4 Interpolation of the Benthic Biomass Data in Two Sets............65
5.2.5 Interpretation of the Interpolated Benthic Biomass Maps .........69
Chapter 6: Conclusions .....................................................................................77
6.1 Introduction.........................................................................................77
6.2 Preparation of Geospatial Database .....................................................77
6.3 Interpolation of Benthic Biomass in the Western Arctic Ocean............79
6.4 Recommendations for Future Work.....................................................80
viii
Appendix A: Scripts/Programs and Procedures ..................................................82
Appendix B: Data Dictionary...........................................................................105
References .......................................................................................................108
Vita..................................................................................................................111
ix
List of Tables
Table 3.1 Description of the Original IBCAO Grid ............................................23
Table 4.1 Lambert Azimuthal Equal-Area Projection Parameters .......................31
Table 4.2 Comparison of the Deviations of the Three Bathymetry Grid Values
from Measured Depth.................................................................................40
Table 5.1 General Statistics of the Benthic Biomass Samples.............................52
Table 5.2 Predicted Benthic Biomass and Standard Error for Selected Locations70
x
List of Figures
Figure 1.1 The Study Area, the Western Arctic ....................................................5
Figure 3.1 Benthic Data Sampling Location along with DCW Coverages...........13
Figure 3.2 The Station table ...............................................................................14
Figure 3.3 The Site table ....................................................................................14
Figure 3.4 The Sample table...............................................................................15
Figure 3.5 The Taxon table.................................................................................15
Figure 3.6 The Tables from the hx59 and hx85 Cruises......................................16
Figure 3.7 The Tables from the hx139 Cruise.....................................................17
Figure 3.8 The Tables with Carey?s Data ...........................................................18
Figure3.9TheTableWacasey ...........................................................................18
Figure 3.10 The Table Feder..............................................................................19
Figure 3.11 The ETOPO5 Bathymetry / Elevation Grid .....................................20
Figure 3.12 Global Bathymetric Prediction Built on Gravity Information...........21
Figure 3.13 The IBCAO Bathymetry Grid..........................................................23
Figure 3.14 The GTOPO30 Digital Elevation Grid.............................................24
Figure 3.15 A Sample of the NOAA Nautical Charts .........................................25
Figure 3.16 The Mean Surface Temperature for July ?93 (Ice-Ocean Model) .....27
Figure 3.17 Mean Surface Salinity for July ?93 (Ice-Ocean Model)....................28
Figure 4.1 The Fivedeg Layer Overlying the Study Area....................................33
Figure 4.2 The NOAAmesh Shapefile................................................................35
Figure 4.3 The NOAAmesh Properties Box .......................................................36
Figure 4.4 Comparison of the a) IBCAO and the b) Gravity Grids (Bering Strait)
...................................................................................................................38
Figure 4.5 Difference Between Measured Depth and the ETOPO5 Grid Value...39
Figure 4.6 Difference Between Measured Depth and Gravity Based Grid...........40
Figure 4.7 Difference Between Measured Depth and the IBCAO Grid Value.....40
Figure 4.8 Question Boxes that Appear During Execution of findflux.ave...........44
Figure 4.9 Currents in the Chukchi Sea Overlying a Temperature Grid, July 1993
...................................................................................................................45
Figure 4.10 The Tables in the Arctic Database and Their Relationships .............46
Figure 4.11 The Format of an Append Query .....................................................46
Figure 4.12 The Benthos, a Table Containing All the Benthic Data ....................47
Figure 4.13 The Configuration of the ODBC drive.............................................47
Figure 4.14 The SQL Connect Window .............................................................48
Figure 5.1 The Basemap, a Digital Bathymetry/Elevation Model??????.52
Figure 5.2 The Distribution of the Benthic Biomass Samples .............................52
Figure 5.3 Interpolation of Biomass with Topogrid,a)RawData,b)Transformed
Data ...........................................................................................................54
xi
Figure 5.4 A Trend Analysis of the Benthic Biomass .........................................55
Figure 5.5 Comparison of the Biomass Distribution to a Normal Distribution ....56
Figure 5.6 Histograms of the Biomass Samples, Raw Data and Log-Transformed
...................................................................................................................56
Figure 5.7 A Geostatistical Method Selection.....................................................57
Figure 5.8 A Detrending Surface........................................................................57
Figure 5.9 A Semivariogram ..............................................................................58
Figure 5.10 Biomass Interpolation and Standard Error Contours ........................60
Figure 5.11 A Cross Validation of Model Output ...............................................60
Figure 5.12 Prediction Error Comparison of Three Ordinary Kriging Models.....62
Figure 5.13 Interpolated Benthic Biomass and Prediction Standard Error
Contours??????????????????????????.65
Figure 5.14 Probability of Biomass Exceeding 50 g/m
2
......................................65
Figure 5.15 The Ordinary Kriging Model for the Beaufort Sea...........................66
Figure 5.16 Interpolation of Benthic Biomass in Two Sets and Prediction
Standard Error Contours?????????????????...??.68
Figure 5.17 Probability of Biomass in Beaufort Sea Exceeding 30 g/m
2
.............68
Figure 5.18 Probability of Biomass in Chukchi and Bering Sea Exceeding 250
g/m
2
...........................................................................................................69
Figure 5.19 The Location of the Example Points, A, B, C and D ........................70
Figure 5.20 The Benthic Biomass Prediction at Point A .....................................71
Figure 5.21 The Benthic Biomass Prediction at Point B .....................................72
Figure 5.22 The Benthic Biomass Prediction at Point C .....................................72
Figure 5.23 The Benthic Biomass Prediction at Point D .....................................73
Figure 5.24 The Benthic Biomass Prediction at all Four Points ..........................73
Figure 5.25 The Benthic Biomass Prediction at Point D, Beaufort Sea Model ....74
Figure 5.26 The Benthic Biomass Prediction at Point D, Beaufort Sea Model ....75
Figure 5.27 Comparison of the Two Models at Point C ......................................76
1
Chapter 1: Introduction
1.1 BACKGROUND
The goal of the National Science Foundation?s (NSF) Arctic Systems
Science (ARCSS) program is to establish knowledge of the physical, geological,
chemical, biological and sociocultural processes of the Arctic system. The
program supports large integrated research projects to establish an understanding
of the Arctic Ocean and its response to global warming. Knowledge of the Arctic
Ocean can then be used to create policy response options to global climate change
(NSF, 1999).
One of four ongoing components of ARCSS is the Ocean/Atmosphere/Ice
Interaction of the Arctic System (OAII) program. OAII seeks to enhance the
understanding of the continuous link between water, ice, air, biota, dissolved
chemicals, and sediments in the Arctic and adjacent oceans, in order to get a
clearer picture of the Arctic?s role in global climate change (ARCSS, 2000).
The Western Arctic Shelf Basin Interactions (SBI) project is one of the
ARCSS/OAII projects. It emphasizes the Chukchi and Beaufort Sea region in the
Western Arctic Ocean and focuses on the connection between the Arctic shelves,
slopes and deep basins with regard to biological, physical and geochemical
processes (WASBI, 2000).
Arctic Benthic Biota - Links to Global Change is a research project within
the scope of SBI. It concentrates on the benthic community in the western Arctic
and its goal is to link patterns of the benthos to physical and biological processes,
2
and thereby establish knowledge of the effects of global change on the ecology of
the Arctic Ocean. Benthic organisms are often considered integrators of the life
in the water column above. Since the water column filters out high frequency
seasonal variations to some extent, the abundance of benthic organisms does not
change drastically with seasons. Therefore with sufficient measurements they are
good indicators of long-term changes.
Mapping of the geographic extent and character of benthic organisms in
the context of other features such as ocean depth, freshwater inflow from rivers,
etc. is essential to comprehend the information available and to connect different
physical and biological factors applicable in the Arctic Ocean. The Geographical
Information Systems (GIS) framework can efficiently handle this sort of
investigation. The ArcView and Arc/Info GIS platforms, developed by the
Environmental Systems Research Institute (ESRI), are capable of visualization
and various analyses of point, vector and raster data.
1.2 STUDY OBJECTIVES
The objectives for this first year of the Arctic Benthic Biota - Links to
Global Change study were devoted to the preparation of data and the
development of methods for further analysis. The work was carried out as a joint
effort of the Marine Science Institute (MSI) and Center for Research in Water
Resources (CRWR). Most of the effort has been dedicated to the GIS component
of the project while a search for adequate benthic data has been extensive and is
not yet completed. The study objective consists of six tasks:
3
? Benthic data compilation. Sample data of the benthic community structure
are available from the National Oceanographic Data Center (NODC) and
other sources. The NODC data are reformatted and compiled into a relational
database (Microsoft Access). In Access, data selection is made and records
are imported to ArcView.
? Digital bathymetry map. Several sources of digital bathymetry/elevation are
available for the study area. Comparison of the grids in ArcView reveals
differences in the resolution and the interpolation techniques of the various
grids. Therefore an evaluation of what data sources are best for the project is
carried out in this research.
? Orientation database construction. Several GIS datalayers are available or
were developed in order to place other layers in a better geographical context.
These involve nautical charts from the National Oceanic and Atmospheric
Administration (NOAA), coastlines and rivers from the Digital Chart of the
World (DCW), and others.
? Importation of oceanographic model results. An Ice-Ocean Arctic Ocean
model has been developed at the Naval Postgraduate School (NPS) in
Monterey California (Maslowski, 2000). By changing the file format of the
models output, benthic data can be compared to ocean currents, temperature
and salinity. Extrapolation of benthic data may be interconnected with those
physical features.
? Geostatistical Analysis. A Geostatistical Analyst extension of Arc/Info 8 can
interpolate point data with deterministic and geostatistic techniques. Thereby
4
it has the potential to create gridded maps of benthic patterns, and patterns
associated with physical and biological processes.
1.3 STUDY AREA
The Arctic Ocean, the smallest of the worlds four oceans (after the Pacific,
Atlantic, and the Indian ocean), has an area of approximately 14 million sq km.
An ice sheet generally covers the area above 75? N, while further south, ice
occurrence is seasonal. The major water exchange with other oceans is through
circulation into and from the Atlantic Ocean through the Fram Strait between
Greenland and Svalbard, where the cold water of the Arctic blends with the
warmer water of the Atlantic. The Arctic Ocean connects to the Pacific Ocean
through the Bering Strait where the warm, nutrient rich waters of the Pacific come
up through the Bering Strait (Smith 1990). North of the Bering strait, between
Siberia and Alaska, lies the Chukchi Sea. To the east of Chukchi Sea, north of
Alaska, is the Beaufort Sea (Figure 1.1).
The study area of this research extends from 62? to 75? North and from
160? East to 128? West (Figure 1.1), encompassing the northern part of the Bering
Sea, the Bering Strait, the Chukchi Sea and the Beaufort Sea. The area is
characterized by high benthic productivity in the shallow Bering and Chukchi Sea
shelf, where the ocean depth is generally less than 60 m. Nutrient flux into the
Beaufort Sea is very limited and consequently, the productivity in the Beaufort
Sea is considerably lower. The Mackenzie River does, however, carry large
5
amounts of sediment and nutrients to the Beaufort Sea, and higher benthic
productivity is identifiable in the vicinity of the river?s mouth.
Figure 1.1 The Study Area, the Western Arctic
1.4 REPORT OUTLINE
This report is divided into six chapters. The first chapter introduces the
study while the second presents an existing literature review on related subjects.
The third chapter is dedicated to the description of the data used in the project
while chapter four describes the methods and procedures used. The results of the
study are presented in the fifth chapter and conclusions are drawn in Chapter six.
Appendix A presents details of procedures and ArcView Avenue scripts.
6
Chapter 2: Literature Review
2.1 BENTHIC COMMUNITY STUDIES
Several research groups have studied the benthic community of the
western Arctic Ocean. Broad (1981), Carey (1984) and Wacasey (1974)
conducted extensive work in the southern Beaufort Sea, while Feder (1982) and
Grebmeier (1987) focused on the Chukchi and Northern Bering Sea.
2.1.1 Southwestern Beaufort Sea, Broad
Broad et al. (1978-1981), explored the nearshore benthos of the Beaufort
and Chukchi Sea, during the summers of 1975-1980. He found that the species
biomass was very low from 2 m depth to the shoreline, due to annual shore-fast
ice. The biomass increased at water depths between 2 m to 20 m, yet were not as
high as those measured in offshore regions by Carey (1984). The principal
nearshore faunal species were polychaete worms, bivalve mollusks, and
crustaceans. An area of higher biomass was recorded just north of Barter Island.
The ?Boulder Patch?, an approximately 40 km
2
area of cobbles and boulders with
attached kelp and invertebrate life, was discovered in Stefansson Sound, near
Prudhoe Bay, Alaska. The Boulder Patch is a rare feature of the Alaskan Beaufort
Sea shelf, which is blanketed predominantly by silty sands and mud. This area
supports a high biomass kelp community dominated by three species of red and
brown algae and by a diverse assortment of invertebrates representing every
major taxonomic phylum.
7
2.1.2 Southwestern Beaufort Sea, Carey
Carey?s (1984) final report, ?Selected Ecological Studies on Continental
Shelf Benthos and Sea Ice Fauna in the Southwestern Beaufort Sea?, describes his
exploration of the southwestern Beaufort Sea. Five quantitative benthic samples
were taken at each location with a 0.1 m
2
Smith-McIntyre bottom grab. Carey
also draws conclusions from the studies of the past years that started in 1971. The
conclusions imply that benthic macrofauna biomass is in general at a maximum at
the edge of the continental shelf and at the upper continental slope. The temporal
variability in the benthic community is greatest at the shelf?s edge and is
correlated with depth across the shelf. The extent of ice cover in the summer
months affects the penetration of light through the water column, thereby
affecting primary production of phytoplankton. Where the shelf is at a depth of
20?25m, the benthic macrofauna seems lower than in the surrounding areas,
probably caused by ice gouging the bottom of the ocean during the winter.
2.2 GEOSTATISTICAL ANALYST
Benthic data cannot be sampled on a spatially continuous scale but rather
at point locations. A continuous surface would however be the optimal result of
an exploration. Therefore, ESRI?s beta product, the Geostatistical Analyst
Arc/Info 8 extension was used to interpolate the benthic data. The extension was
studied and the Geostatistical Analyst Tutorial (2000) reviewed. The tutorial
included nine chapters describing geostatistical interpolation techniques and
application of them in Arc/Info 8. The tutorial is written in an ?easy to
8
understand? manner while an appendix contains the complete mathematical
formulas, which the geostatistical analysis is based upon.
Several methods of gaining familiarity with the data are available before
interpolation takes place. Histograms let the user examine the frequency
distribution of the data while Normal Quantile-Quantile plots (QQ-plots) look at
whether the data can be assumed to have a normal distribution. The Trend
Analysis option in the extension allows the user to identify nonrandom trends in
the data. This data exploration can assist the user in an identification of outliers
and a selection of interpolation technique.
Two groups of interpolation techniques are available in the Geostatistical
Analyst extension, deterministic interpolation and geostatistics.
2.2.1 Deterministic Interpolation
Deterministic interpolation techniques are built on the basic law of
geography that things close together are more similar than things that are far
apart. The Geostatistical Analyst provides four deterministic interpolation
techniques, Inverse Distance Weighted, Global Polynomial, Local Polynomial,
and Radial Basis Functions.
The Inverse Distance Weighted Function calculates the value at every
point by taking a weighted average of the point data in the neighborhood. The
weight of a measured data point is inversely proportional to the distance from the
measured point to the prediction point. Therefore measurements close by have a
greater weight than those further away.
9
The Global Polynomial Function fits a mathematical function
(polynomial) to all the measured points. The process selects the function that
minimizes the squared differences between measured and predicted values. The
result is a smooth mathematical surface that shows gradual trends in the study
area.
The Local Polynomial Function also fits a polynomial to the data but
allows for more local variation. While the global interpolation tries to fit all the
measured values with one function, the local function uses many polynomial
functions to fit the area. Each function applies to a specified neighborhood, but
areas overlap to avoid sharp breaks.
The Radial Basis Function fits a smooth surface through every measured
point and minimizes the surface curvature. It differs from the polynomial
functions since the surface goes through every measured point, while giving a
more gently varying surface than the inverse distance weighting method.
2.2.2 Geostatistical Techniques
The geostatistical methods are based on statistics. They have two major
tasks, calculating the appropriate statistics and creating a prediction surface. A
great advantage of these methods is that they can give an estimate of how well
they are predicting the surface, i.e. error of prediction, whereas the deterministic
methods cannot. They are all in the kriging family and are generally referred to as
kriging methods or cokriging methods for a multivariate case.
Many of the kriging methods assume a normal distribution of the point
data. Therefore, it is important to identify whether the data must be transformed
10
or whether a trend must be removed in the interpolation process. The QQ plots
are graphs on which quantiles from two distributions are plotted relative to each
other. The normal QQ plots and histograms can show whether the data are
normally distributed. The Geostatistical Analyst extension can transform data and
remove trends before interpolation and then add these effects back to the
interpolated surface.
The kriging technique consists of evaluating the spatial autocorrelation of
the data and assigning weights based on the distance from a measured location to
the prediction location. Pairs of measurements are grouped together in bins,
defined by a distance range and an angle tolerance if anisotropy is considered. An
average of squared difference in values for each bin is plotted against the distance
that separates the points (an empirical semivariogram). Data pairs located close
together should have a smaller difference in observed values than data pairs that
are far away from each other. A line is then fit through the points on the
semivariogram. That line is then assumed to represent the spatial autocorrelation
of the data, and it defines the weight of measured values in a predicted surface.
Anisotropic models create separate models for each direction interval
defined by the angle tolerance. An anisotropic model can be good to account for
directional characteristics of the data. However if data are sparse, empty bins can
exist and bins with few points for averaging may create undesirable variations.
While the kriging technique uses information on one variable, the
cokriging process uses information on several variables. It estimates the
autocorrelation of the main variable of interest and the crosscorrelation between
11
the main variable and all other variables. This is done in order to use more widely
sampled variables and their correlation to the variable of interest. However, if
there is however no correlation between a pair of variables, this method can
introduce more errors, since more statistical constants are estimated.
Several different kriging and cokriging methods are available in the
Geostatistical Analyst extension, including ordinary, simple, universal,
probability and disjunctive kriging and cokriging.
? Ordinary kriging assumes an unknown constant mean ?, but adds an error
factor, ?(s) evaluated based on autocorrelation. Z(s)=? + ?(s), where s is
a spatial location
? Simple kriging assumes a known constant mean, ?, but adds an error
factor, ?(s) evaluated based on autocorrelation. If the mean is exactly
known this model gives more exact results than the ordinary kriging where
? and therefore ?(s) had to be estimated. Z(s)=? + ?(s)
? Universal kriging assumes a mean, ?(s), where ?(s) is a deterministic
polynomial function. Z(s)=?(s)+ ?(s)
? Indicator kriging is used for binary data that are created either from a
threshold in continuous data or if the data are all 0 or 1. Apart from that,
themodelisthesameastheordinarykrigingmodel. Z(s)=? + ?(s)
? Probability and disjunctive kriging are intented to do the same thing as
indicator kriging. Disjunctive kriging is a generalization of the indicator
kriging, but requires a bivariate normality assumption of errors.
12
Chapter 3: Data
A major objective of this research project is to obtain quantitative benthic
data for the Beaufort Sea, Chukchi Sea, and northern Bering Sea and to create
representative surfaces from the point data. Investigators at the Marine Science
Institute (MSI) and the author at the Center for Research in Water Resources
(CRWR) have invested great effort to finding benthic data from various sources to
import into a relational database. However, benthic data cannot tell the whole
story alone. Therefore digital data sources have been explored in order to find
helpful data layers, such as bathymetry grids, nautical charts, coastlines, and
major rivers. Also the output of the NPS Ice-Ocean Arctic model has been
adapted to the ArcView framework.
The data were derived from various sources, and were in various file
formats and projections. This chapter describes the data and their sources.
3.1 BENTHIC DATA
Several groups have researched the benthic community structure and
biomass in the study area. Many of the data used in this study were compiled at
the National Oceanographic Data Center (NODC), while other data came from
literature or were obtained as spreadsheets from researchers. Figure 3.1 shows the
locations of the benthic sampling. The point locations are divided up colors
according to the tables their source.
13
Figure 3.1 Benthic Data Sampling Location along with DCW Coverages
3.1.1 Benthic data from NODC
Broad (1981) researched the Western Beaufort Sea and submitted results
to the National Oceanographic Data Center (NODC). Most of the NODC data for
the study area are from his research. Benthic data were requested from the
NODC for the study area and the data were supplied in a heavily coded text file.
The translation of the coded file to a spreadsheet text file required extensive
computer script writing effort, which was carried out at the MSI. The resulting
tables are four; Station, Site, Sample and Taxon.
A station is the specific location where the sampling boat stops at on a
specific date. The Station table (Figure 3.2) includes the number of the station,
the date on which the samples were taken, the station?s exact location and habitat
composition. The location is given as latitude and longitude in decimal degrees.
14
Figure 3.2 The Station table
At each station there are many sites where samples of the seafloor are
collected. The Site table (Figure 3.3) lists every record of a site, the site?s station
number, its temperature (?C), salinity (ppt) and depth (m). It also lists the type of
grab that was used for sampling.
Figure 3.3 The Site table
At each site many different types of animals are collected in the grabs and
they are listed in the Sample table (Figure 3.4). Therefore, there are many
samples for each site. In the Sample table, there is a record for every animal
found at each site. It lists the taxonomic number of each animal, the site and
station of where it was found, and its abundance and biomass. The abundance is
15
the number of individuals found while the biomass is the wet weight of all the
individuals (g/m
2
).
Figure 3.4 The Sample table
Finally, the Taxon table(Figure3.5)pairsthetaxonomicnumbersinthe
Sample table with the name of each animal.
Figure 3.5 The Taxon table
3.1.2 Benthic Data from other Sources
The search for quantitative benthic data is still going on. As of now the
database includes Broad?s (1981) NODC data, two collections submitted by
Grebmeier (1987, 1985) and data from investigations carried out by Carey (1984),
Wacasey (1974) and Feder (1982). All data used in this study required location
data (lat, long) and wet weight biomass (g/m
2
) while other fields give various
16
additional information. All the tables are consistent in measurement units and can
therefore be merged.
Grebmeier?s data from the Chukchi and Northern Bering Sea were
submitted in four spreadsheets. Data from the hx59 and hx85 cruises are in two
tables (Figure 3.6). Thesisjmg gives the general information for each station,
location, date, depth. temperature, salinity etc. while the hx59-hx85-benthic table
lists each animal type, its biomass and abundance. Data from the hx139 cruise are
set up in a similar way (Figure 3.7). The table hx1391 gives general information
for the station and each grab, including latitiude, longitude and date, while the
table hx139benthic gives abundance and biomass of the animal species found in
the grab.
Figure 3.6 The Tables from the hx59 and hx85 Cruises
17
Figure 3.7 The Tables from the hx139 Cruise
Data from the Western Beaufort Sea were extracted from reports by Carey
(1984). Four tables with Carey?s data are in the database (Figure 3.8). Data from
a research done in 1971 are in two tables, Websec-71-600 and Carey1971. They
give the location, date and biomass. The Outer Continental Shelf (OCS) project
was carried out in the years 1975-1978 by Carey (1984); the data from those
cruises are in the tables CareyOCS and CareyOCS_bio. CareyOCS gives the
location, date, grab number and depth while the CareyOCS_bio table lists the
species, and the biomass, found in every grab listed in the CareyOCS table.
18
Figure 3.8 The Tables with Carey?s Data
Wacasey?s (1974) report provides general information on stations in the
southern Beaufort Sea, along with the benthic biomass and abundance. Those
data are gathered in the table titled Wacasey (Figure 3.9).
Figure3.9TheTableWacasey
Information from Feder?s (1982) report on the Chukchi and the northern
Bering Sea are gathered in the table titled Feder, which gives general information
on stations, biomass and abundance (Figure 3.10).
19
Figure 3.10 The Table Feder
3.2 DIGITAL BATHYMETRY
Several sources of gridded bathymetry/elevation are available for the
study area. The grids differ in their data sources, resolution and accuracy.
Therefore the available literature associated with the grids was reviewed. A
description of these grids and their sources are given below. The sources
considered for this project are the following:
? ETOPO5, a global 5-minute bathymetry grid (NGDC, 2000)
? A global bathymetry map built on marine gravity along with ship sounding
(Sandwell, 2000)
? IBCAO, a bathymetry grid of the Arctic Ocean (Jacobson, 2000b)
? GTOPO30, a global 30 arc second elevation grid (EROS, 2000)
3.2.1 ETOPO5 5-minute Gridded Elevation Data
The information site for the ETOPO5 (NGDC, 2000) gives some details
about the global digital bathymetry/elevation dataset. The grid was assembled
from various data sources available in 1988, resulting in a 5-minute degree cell
grid. It is not clear whether the data sources are raw ship soundings, contours, or
20
grids. Since little detail were available for oceanic data shallower than 200m, the
interpolation algorithm used to interpolate contour charts was set to an arbitrary
cut of ?10m to prevent overshooting, i.e. creation of imaginary islands.
ETOPO5 can be downloaded from a NGDC Internet website,
http://www.ngdc.noaa.gov/mgg/global/seltopo.html. The corner points of a
region of interest are given in geographic coordinates. Then a server extracts the
correct area and creates an ASCII delimited text file, either as an XYZ list (long,
lat, depth/elevation) or as a matrix of depth/elevation. A negative value indicates
a depth while a non-negative value is elevation above mean sea level. Data for
the study area were extracted and the resulting grid can be seen in Figure 3.11.
Figure 3.11 The ETOPO5 Bathymetry / Elevation Grid
3.2.2 Global Bathymetric Prediction for Ocean Modelling and Marine
Geophysics
In an article by Sandwell and Smith (1996) a global digital bathymetric
map is described. This map has a resolution of 3-10 km and is created by
combining ship soundings from the past 30 years with high-resolution marine
21
gravity information provided by the Geosat, ERS-1/2, and Topex/Poseidon
satellite altimeters. Previous global bathymetry maps have relied on only the ship
soundings. Ship soundings are used with precaution in this map; older soundings
are not as reliable as the most recent Global Positioning System (GPS) navigated
multibeam data. Gravity anomalies are highly correlated with seafloor
topography, therefore the ship soundings are used to calibrate the gravity
information, i.e. to find the topography/gravity ratio. While this bathymetry map
does not agree with earlier maps such as ETOPO-5 bathymetry model, the high
accuracy and spatial resolution of gravity information reveal many bathymetric
features previously unrecognized.
Areas of interests from the global topography dataset are available up to
latitude of 72?N as seen on Figure 3.12. They can be downloaded as an ASCII
XYZ text file at the Internet website in geographic coordinates,
http://topex.ucsd.edu/cgi-bin/get_data.cgi.
Figure 3.12 Global Bathymetric Prediction Built on Gravity Information
22
3.2.3 International Bathymetric Chart of the Arctic Ocean (IBCAO)
A technical reference and user?s guide along with the Beta Version of the
IBCAO digital grid became downloadable from the Internet site in the spring of
2000. According to the guide (Jakobsson, 2000a), the grid was built from all
available contour, grid, point, and track data north of 64?N, and will be further
improved once additional data are collected. Gridding of the source data to a cell
size of 2.5 km was done using curvature splines in a tension algorithm using the
Generic Mapping Tools (GMT) program Surface with a tension parameter set to
0.35. Subsequently the effects of randomly spaced ship tracks were smoothed out
using a weighted average convolution.
The guide contains several application suggestions and information on
how the grid can be viewed and used; it does however not mention ArcView or
Arc/Info applications.
The IBCAO grid can be downloaded from the Internet site:
http://www.ngdc.noaa.gov/mgg/bathymetry/arctic/ibcaobetagrid.html. It is only
available in its complete version, i.e. regions of interest cannot be selected. The
data are available as an ASCII or a netCDF file and Table 3.1 describes the
original grid while Figure 3.13 shows the extent of the grid in the study area.
23
Table 3.1 Description of the Original IBCAO Grid
Figure 3.13 The IBCAO Bathymetry Grid
3.2.4 Global Digital Elevation Model, GTOPO30
The GTOPO30 is a global data set of elevation developed by U.S
Geological Survey?s Earth Resources Observation Systems (EROS) Data Center
(EDC). A Readme text file gives information about the dataset (EROS, 2000).
Elevations in GTOPO30 are given at 30-arc second regularly spaced intervals
over the whole land surface of the globe. The data set is available at the Internet
site http://edc.usgs.gov/landdaac/gtopo30/gtopo30.html where the zipfiles,
24
E140N90, W180N90 and W140N90, encompass the study area (Figure 3.14).
The Digital Elevation Model is provided as 16-bit signed integer data in a simple
binary raster; a Readme file provides instruction on how to translate the file into
an ArcInfo grid. The horizontal reference system is referenced to the WGS84
spheroid; latitude and longitude are given in decimal degrees. Ocean areas have
been given the nodata value ?9999.
Figure 3.14 The GTOPO30 Digital Elevation Grid
3.3 GIS FILES FOR ORIENTATION
Several GIS data layers were collected to give the benthic information a
better geographical background. These involve nautical charts from the National
Oceanic and Atmospheric Administration (NOAA), as well as coastlines and
rivers from the Digital Chart of the World (DCW). A description of these data is
in the following two sections.
25
3.3.1 The NOAA Nautical Charts
Digital raster charts of the entire NOAA chart collection are now available
through Maptech?s mapserver (Maptech, 2000). The charts are scanned version
of the NOAA nautical charts and include every detail found in the paper version.
The digital charts can be imported into ArcView using a Chartviewer extension
downloadable from a NOAA (2000) Internet site along with a Chart Reprojector
extension. The charts are in various projections but the Chartviewer extension
reads the projection of each file and sets the view projection while the Reprojector
projects the charts permanently.
Charts of the study area, (R 36 CD) Alaska Norton Sound to Beaufort Sea
CD-ROM, were purchased from Maptech. A sample of the maps is shown in
Figure 3.15.
Figure 3.15 A Sample of the NOAA Nautical Charts
26
3.3.2 Digital Chart of The World
The Digital Chart of the World (DCW) was originally developed for the
US Defense Mapping Agency (DMA) by ESRI. The source is DMA 1:1,000,000
scale aeronautical charts. A detailed description of the dataset can be found at the
Pennsylvanian State University Libraries Internet site (2000). Several datasets are
available from the DCW. However, for this research, only the coastlines and river
coverages were downloaded from the Internet site (PSU, 2000). The Arc/Info
coverages downloaded were for the areas: East Russia, Alaska, the Yukon
Territory and the Northwest Territories. All the coverages, (Figure 3.1) are in
geographic projection.
3.4 ICE-OCEAN MODEL FOR THE ARCTIC OCEAN
An ice-ocean model with a cell size of 18 km has been developed for the
Arctic Ocean in an attempt to explain the Arctic Ocean response to global
variations (Maslowski, 2000). The model consists of an ocean general circulation
model coupled to a sea ice model calibrated with 20 years of atmospheric data
from the European Center for Medium-range Weather Forecasts (ECMWF). The
grid cell size of the model is 1/6-degree, a little more than 18 km, and the grid
consists of 368x304 cells. The output is produced in four ways, as follows:
? 30-day averages of sea surface height and 3-D fields of temperature,
salinity, and velocity vectors (called ocean t-files)
? 30-day snapshots of similar fields plus fresh water and Atlantic water
tracers (h-files)
27
? 3-day snapshots of selected 2-D fields (T,S,u,v and tracers) (movie-files)
? 3-day averages of most sea ice predicted fields (thickness, concentration,
growth/melt, etc.) (ice t-files)
Ocean t-files for the year of 1993 were made available for this study, they
included 12 monthly mean files (t.c6.93-*.cdf.gz) a land mask file (kmp.levels),
bathymetry file (bath.levels) and a model grid file (latlonh.dat). Each of the 12
monthly mean files is in a netCDF format and includes matrices (368 columns x
304 rows) for several variables. The variables include sea surface height - AH1
(cm), velocity ? AU, AV (cm/sec), temperature ? ATEMP (?C) and salinity -
ASALT (ppt). An example of the surface temperature and salinity grids is shown
in Figure 3.16 and Figure 3.17.
Figure 3.16 The Mean Surface Temperature for July ?93 (Ice-Ocean Model)
28
Figure 3.17 Mean Surface Salinity for July ?93 (Ice-Ocean Model)
29
Chapter 4: Methodology
The approach of this study is directed towards the application of ESRI?s
Arc View and Arc/Info platforms. This study?s main focus is to illustrate current
state of knowledge of the Western Arctic benthos and to provide a basis for
further analysis. This is done by plotting quantitative benthic data from earlier
studies, and tying them to other available information, such as bathymetry,
temperature, salinity and freshwater inflow. The GIS framework can incorporate
data from various sources for nice visualization, and also allow the user to do
queries and calculations. Therefore, the ArcView and Arc/Info platforms are
optimal for this research.
Numerous organizations have digital data available. The data, however,
come in different formats and map projections; much work is involved in
reformatting and converting the data. Procedures have to be developed and
computer scripts written in order to make further analysis possible.
This chapter describes the general methods and procedures used in the
project. Procedure details, Arc Macro Language (AML) and Avenue scripts, can
be found in Appendix A. All data layers are distinguished by a bold font.
4.1 MAP PROJECTION
Geographic information systems use spatially referenced data, i.e. data are
viewed according to their (x,y) coordinates. In order to create a flat map sheet
from the curved Earth, its three dimensional surface has to be transformed. The
30
most commonly known way of displaying geographic data is to assign the
longitude to the x-axis and the latitude to the y-axis. This geographic coordinate
system distorts shapes and other properties of the data and is therefore not useful
for an analysis of the data. However, various mathematical transformations,
called map projections, can convert geographic locations to reference systems that
preserve some of the data?s practical information. Every projection distorts either
one or more of the four general properties, shape, area, distance or direction. A
choice of projection must be made with regards to the type of analysis intended
and the location of study area (Snyder, 1987).
The Polar Stereographic projection is a widely used projection in the polar
regions; it is practical for navigational purposes since it preserves direction, i.e.
local angles are accurate everywhere. This project, however, deals with an
interpolation of benthic biomass; therefore preservation of area is more important
than direction. The Lambert Azimuthal Equal-Area projection retains the true
Earth surface area at the same scale, while other properties are somewhat
distorted.
The specific parameters of the Lamberts Azimuthal Equal-Area were
chosen with regards to the study area. The chosen radius of sphere of reference is
the default radius in Arc/Info. The longitude of center of projection was chosen
so that the extent of the study area is vertically symmetrical, while the North Pole
was selected as the longitude of center of projection. False easting and northing
were chosen so that all points in the study area have positive locations.
31
All data used in this research have been projected to the Lambert
projection; the specific parameters can be found in Table 4.1
Table 4.1 Lambert Azimuthal Equal-Area Projection Parameters
The specific parameters were chosen with regards to the study area. The
chosen radius of sphere of reference is the default radius in Arc/Info. The
longitude of center of projection was chosen so that the extent of the study area is
vertically symmetrical, while the North Pole was selected as the longitude of
center of projection. False easting and northing were chosen so that all points in
the study area have positive locations.
4.2 PREPARATION OF GIS LAYERS FOR ORIENTATION
The general frame of the study area gives the user a frame of reference,
i.e. knowledge of what part of the world is being viewed. That general
information consists of coastlines, a latitude/longitude grid, nautical charts etc.
4.2.1 Building a Frame and a Latlong Grid
The study area reaches from latitude 62? N to 75?N and 160?E to 128?W.
The area encompasses the northern part of the Bering Sea, the Chukchi Sea, The
Projection: Lambert Azimuthal Equal-Area
Units: Meters
Radius of sphere of reference: 6370997 m
Longitude of center of projection: -165? 00' 00''
Latitude of center of projection: 90? 00' 00''
False easting: 2000000 m
False northing: 4000000 m
32
Beaufort Sea and the East Siberian Sea. It was selected so that in reaches from
the continental shelf to the deeper basin of the Arctic Ocean.
On a sphere the 180?W longitude is the same as the 180?E longitude.
However, when ArcView and Arc/Info handle geographic coordinates, 180?W is
given the number ?180 while 180?E is +180. Since the study area reaches over
180? E/W longitude the area is not spatially continuous in geographic coordinates.
Layers for the study area in geographic coordinates are therefore divided up in
two parts, one west of 180?E/W longitude and another east of the 180?E/W
longitude. However layers in projected coordinate systems are spatially
continuous.
When the study area has been defined, the relevant sections of maps have
been chosen. A shapefile consisting of the outline of the study area was created
by which other layers could subsequently be clipped. The coverage frame was
created in Arc/Info using the Generate function. Two boxes were created in
geographic coordinates, one for the eastern part of the study area (east of 180?W)
and another for the western part. The frame layer was then projected to the
Lambert Azimuthal Equal Area projection. The Arc/Info command sequence
(AML) can be found in Appendix A.
The projection of all data into a projection different than geographic
projection results in a loss of (longitude, latitude) references. However
geographic coordinates are the most common way of position identification.
Therefore, a net of five-degree cells was created and attributed with the
geographic coordinates of the lower left corner of each cell. The layer, fivedeg,
33
was created in Arc/Info using the Generate function (details are given in
Appendix A). The coverage was then converted to a shapefile in ArcView and its
attribute table edited. The lowerleft field in the table has the geographic
coordinate of each cell?s lower left corner. Figure 4.1 shows the fivedeg layer
overlying the land/ocean within the study area.
Figure 4.1 The fivedeg Layer Overlying the Study Area
4.2.2 Layers from Digital Chart of the World
Layers from the Digital Chart of the World were downloaded from their
Internet site (PSU 2000). The layers selected were ponet and rivers,forthe
geographic regions of East Russia, Alaska, the Yukon Territory and the
Northwest Territories. The files were in Arc/Info export format (.e00) and were
imported with the Import71 utility. The layers were then projected to the Lambert
Azimuthal Equal Area projection and merged in ArcView with the
mergethemes.ave script (Appendix A).
34
The coastlines, ponet, from the DCW, define where the ocean ends and
land begins. The river shapefile from the DCW only shows major rivers like the
Yukon and Mackenzie rivers. The position of the mouth of those big rivers
indicates where influences of freshwater inflow may be significant. Therefore
this layer is included on most maps showing interpolated biomass surfaces.
4.2.3 NOAA Nautical Charts
The NOAA nautical charts provide valuable information in the form of
place names and other landmarks. They are scanned versions of the paper maps
and include all the details found in the paper maps. The files are in a BSB digital
format, which is generally not a file format supported by ArcView. However an
extension has been developed by NOAA Coastal Services Center, Maptech, Inc.,
and ESRI that allows the GIS users to read a BSB-formatted chart (NOAA, 2000).
The extension, chartviewer, consists of two files, chartview.dll and chartview.avx.
Those files have to be added to the ArcView?s extension folder on the user?s
computer, chartview.dll to the ArcView/Bin directory and chartview.avx to the
ArcView/Ext directory. Then the chartviewer extension becomes an available
extension in ArcView. When the chartviewer extension is selected, the ?Add a
Nautical Chart? button appears in the projects toolbar. The charts can be
added to the view like any other ArcView layer by filenames.
The filenames of the charts do not provide information on the extent of
each map. Therefore a more advanced method of adding the charts to the view
was developed. A shapefile, NOAAmesh, containing the outlines of the charts
was created. Along with the shapefile, the script addnoaa.ave was created. The
35
shapefile NOAAmesh (Figure 4.2) was created by using the Avenue program
get_extent.ave to extract the outlines of every chart. Then the shapefile along
with all the charts were projected to Lambert projection using the chart projection
utility chartreprojector (NOAA, 2000). Finally the addnoaa.ave was written and
used to activate the hotlink for NOAAmesh.
Figure 4.2 The NOAAmesh Shapefile
The products of the above procedure can be used in ArcView in the
following way. The addNOAA script is added to the project and the name
changed to addNOAA. The NOAAmesh is added to the view, and its attribute
table opened. Then hotlink properties of the NOAAmesh are edited (Theme->
Properties-> Hotlink). The hotlink box should look like the one in Figure 4.3.
36
Figure 4.3 The NOAAmesh Properties Box
When the hotlink has been activated, charts can be added to the view by
first selecting the hotlink button and then clicking on a polygon in the
NOAAmesh where a NOAA nautical chart is wanted.
4.3 DIGITAL BATHYMETRY/ELEVATION BASEMAP
Four sources of digital bathymetry/elevation were reviewed in this project.
They were in different formats so procedures were developed to import the grids
into ArcView. In ArcView the grids were compared and a decision was made
about how to construct one basemap from these sources.
4.3.1 Importing the Bathymetry Grids
The ETOPO5 bathymetry grid was downloaded from an Internet site
(NGDC, 2000) in an ASCII comma delimited text file format, (a XYZ list long,
lat, depth/elevation). Grid points west of 180?E were downloaded separately from
37
those east of 180?E since the site server could not handle the total area all at once.
The two files were merged in Notepad and added to the ArcView project as a
table. From there, the data were added to the view as an event theme (View ->
Add Event Theme) and converted to a shapefile. In Arc/Info the layer was
projected to the Lambert Azimuthal Equal Area projection and a grid was
produced using the topogrid function; the command sequence can be found in
Appendix A. The resulting grid, ETOPO5 is shown in Figure 3.11.
The gravity based bathymetric prediction grid was downloaded from an
Internet site (Sandwell, 2000). The file format is an ASCII comma delimited text
file format, a XYZ list. The same procedure was therefore used for this grid as
the ETOPO5 grid, resulting in the grid shown in Figure 3.12.
The IBCAO Arctic bathymetry grid is only available in its complete
version from an Internet site (Jacobson, 2000b). The whole file was downloaded
and when it was unzipped the file proved to be 150 megabytes in size. Therefore,
Notepad, Excel, Access and Wordpad were unable to open the file. However in
Matlab, fractions of the ASCII file could be read and stored separately. The
procedure conducted in Matlab is in Appendix A. The resulting six comma
delimited text files were then imported into ArcView, added to a view as event
themes, converted to shapefiles and merged with mergethemes.ave.Thelayer
was projected using the pol2lamb.txt text file and a grid was interpolated from the
layer using the topogrid function. The same command sequence was used as with
the ETOPO5 grid.
38
Three GTOPO30 zip files (E140N90, W180N90, and W140N90) were
downloaded from the Internet site (EROS, 2000). The files were unzipped and
their extension was changed from *.dem to *.bil. Then the files were processed in
the Arc/Info according to the GTOPO30 readme file. The procedure can be found
in Appendix A.
4.3.2 Assembling the Basemap
Comparison of the three bathymetric grids reveals a great difference
between the one assembled from satellite gravity information and the other two,
ETOPO5 and IBCAO. The difference between the gravity based grid and the
IBCAO grid in the Bering Strait can be seen in Figure 4.4.
Figure 4.4 Comparison of the a) IBCAO and the b) Gravity Grids (Bering Strait)
Different cell sizes or interpolation algorithms cannot explain this major
difference. The cause must be found in the data sources that lie behind the
bathymetry models. The IBCAO is created only for the Arctic Ocean from ship
soundings while the gravity-based grid is created from satellite information on
39
gravity and calibrated by sparse ship soundings. A choice of which bathymetry
grid to use had to be made; therefore measured depth in the benthic data sampling
was compared to the grid values of the three bathymetry maps. In the statistical
program Minitab, the deviations of the grids from the measured values were
analyzed. Figures 4.5, 4.6 and 4.7 show a dotplots of the deviations for each grid,
the ETOPO5 grid, the gravity based grid and the IBCAO grid
Figure 4.5 Difference Between Measured Depth and the ETOPO5 Grid Value
40
Figure 4.6 Difference Between Measured Depth and Gravity Based Grid
Figure 4.7 Difference Between Measured Depth and the IBCAO Grid Value
The three dotplots above show two outliers, one far to the right and another far to
the left. These outliers were left out in the evaluation of the grids, since if they
were included their values would dominate the calculated mean and standard
deviation. The three grids are compared in Table 4.2.
Table 4.2 Comparison of the Deviations of the Three Bathymetry Grid Values
from Measured Depth
The comparison in Table 4.2 shows the statistics of the difference between
the measured ocean depth and the depth according to the three bathymetry grids.
41
The mean of pure deviations reveals whether the deviations are biased, i.e.
whether the grid values are consistently higher or lower than the measured depth.
While statistics of the squared deviations indicate the overall deviations. The
IBCAO grid has the smallest deviations, both when pure deviations are compared
and in the case of squared deviations.
Since the IBCAO grid proved to be closer to the measured values than the
other two grids, the IBCAO grid was used as a base map for the study area. Since
the grid does not cover the entire area (Figure 3.4), data from another grid was
used to fill in the gap. The ETOPO5 grid has the second smallest deviations after
the IBCAO grid (Table 4.2) and was therefore used along with the IBCAO grid.
Therefore the final base map is composed of the IBCAO grid for the most part of
the ocean, the ETOPO5 when IBCAO does not exist and GTOPO30 for the land
areas. There is a smooth transition between all three grids, no abrupt changes
appear in elevation or depth.
The procedure of merging those grid sources, to create the basemap grid,
was conducted in Arc/Info and the details can be found in Appendix A.
4.4 PROCESSING THE ARCTIC ICE-OCEAN MODEL OUTPUT
The output of the Arctic Ice-Ocean Model for the year of 1993 was
downloaded. The 12 month average data files were in a NetCDF file format.
Many ways of processing the NetCDF files were considered, most of which came
from the Unidata reference Internet site (UCAR, 2000) that describes software for
manipulating or displaying NetCDF data. Most of those methods appear to be
42
used in conjunction with UNIX computers, and were therefore not a feasible
option for this research. However, the Noesys data visualizing software works in
the Windows environment (RS, 2000). The software is not free but a trial version
was used for the project.
The Noesys system allows you to import the Netcdf files into a new
project (File -> Import) and a multilayer file shows up. The ah1, atemp, asal, au
and av are 368x304 matrices that have information on sea surface height,
temperature, salinity and the velocity vectors, respectively. The matrices can then
be exported (File -> Export Object), with selecting the file type ?Formatted text?
and choosing a {space} for a column delimiter and a {CR} as a row delimiter.
The output is a comma delimited text files with a matrix of numbers. The file can
then be converted to a grid in Arc/Info and viewed in ArcView. Details are in
Appendix A.
The only available information on the projection of the ice-ocean model
grids was that they were in a rotated spherical coordinate system with the equator
along 50?W and 130?E. To decipher the projection of the grids, the coordinates of
the gridpoints were projected into several different projections. The only
projection that showed the extent of the gridpoint collection as a rectangular was
an Azimuthal Equidistant map projection. Therefore the Azimuthal Equidistant
projection is regarded as the projection of the Ice-Ocean Arctic model. The
longitude of the center of projection proved to be 140?W and the latitude of its
center is the North Pole. The specific northing and easting parameters of the
projection were defined by projecting a coastline shapefile into Azimuthal
43
Equidistant projection. This was done several times with different projection
parameters. The projection parameters of the coastline coverage that fitted best
were considered the right parameters. The grids were assigned the Azimuthal
Equidistant projection and projected to Lambert projection using the azi2lamb.txt
projection file (Appendix A).
The velocity vector grid image, u and v, does not give a representative
image of the currents. Therefore a method of viewing the velocity vectors as
arrows with the correct direction and a relative length was developed. Grid points
with velocity values were extracted in Arc/Info from both velocity components
for every month. One point layer, flux_lamb, was created from these grid points,
with the velocity values for every month in the attribute table (Appendix A).
Now all the velocity information for the entire year 1993, is in one attribute table.
The scripts findflux.ave and arrow.ave (originally written by Ye Zichuan), were
rewritten for these data. The scripts can now be executed from a Draw Ocean
Current Arrows button and the question boxes shown in Figure 4.8 appear. If
more than one shapefile is in the view, the program asks you for which shapefile
it is supposed to draw arrows. The program then lets you choose a month and in
what projection the arrows should be drawn, the Lambert Azimuthal Equal Area
projection or the original Azimuthal Equidistant projection. Finally, a relative
length of arrows is chosen and the program draws the arrows for all the selected
points in the layer.
44
Figure 4.8 Question Boxes that Appear During Execution of findflux.ave
Figure 4.9 shows the current arrows for July 1993 in Chukchi Sea,
overlying the temperature grid for the same month. A strong current appears to
lie around Wrangell Island (Ostrov Vrangelya), and down through the Bering
Strait as well as up by the Alaskan Coast.
Monthly averages of temperature, salinity, ocean height and velocity
vectors for the year 1993 are available on the attached CD. The temperature,
salinity and ocean height are in a grid format, while the velocity vectors can be
viewed as flux arrows. The temperature and salinity conditions of the ocean
influence which benthic species inhabit the ocean floor and their abundance.
Information about the ocean currents is also essential to understand how nutrients
are transported and where they may be in high concentrations. Regions with high
nutrient concentration are likely to have high benthic productivity.
45
Figure 4.9 Currents in the Chukchi Sea Overlying a Temperature Grid, July 1993
4.5 THE BENTHIC RELATIONAL DATABASE
All the benthic data that was gathered for this research was assembled in a
Microsoft Access database called Arctic. Data were set up in Excel worksheets
and imported into Access, and then appropriate relationships between the tables
were established as seen in Figure 4.10. In order to transfer the files into
ArcView, append queries were used to create a table, Benthos, with all the
information of interest from all the tables. A sample of the append queries can be
46
seen in Figure 4.11, while a few rows in the Benthos tableareshowninFigure
4.12.
Figure 4.10 The Tables in the Arctic Database and Their Relationships
Figure 4.11 The Format of an Append Query
47
Figure 4.12 The Benthos, a Table Containing All the Benthic Data
From this format, the table can be transferred to an ArcView project. Two
methods were used for the transfer, exportation and database connection. The
table was exported as a comma delimited text file and added to the project view in
ArcView as an event theme. This method is simpler than the database
connection. In order to establish a database connection between Access and
ArcView an ODBC driver has to be set up. To set the driver up, the control panel
of the computer is opened and the ODBC icon clicked. Under the User DSN tab,
the Add button is selected and the driver is configured in the way shown in Figure
4.13.
Figure 4.13 The Configuration of the ODBC drive
48
The Arctic database can now be connected to the Arc View project. With
the project window active, the SQL Connect option is selected under the Project
menu. The Arctic database is selected in the Connection box and the Connect
button is pressed, then tables and field of interest can be selected (Figure 4.14)
and the Query button pressed to create a table. The table created by the query can
then be added to the view as an event theme.
Figure 4.14 The SQL Connect Window
This database connection has the advantage that the table can always be
refreshed (Table -> Refresh) to update the values after changes done in Access.
The other option, exporting a text file, has to be done after every change in order
to update the values. The event theme can then be converted to a shapefile and
projected to Lambert projection using the geo2lamb.txt projection file.
49
Quantitative benthic data have been assembled in an Access Relational
Database, titled Arctic, where appropriate relationships have been defined
between data tables. Individual tables have been queried to incorporate all the
benthic biomass data in one table titled Benthos. Information in the original
tables on specific species found at stations is not included in the Benthos table.
The species information can however be incorporated at later time with further
queries. The Benthos table has been imported into ArcView and converted to the
shapefile samples. The shapefile was used for further spatial analysis of benthic
biomass in ArcView and Arc/Info.
50
Chapter 5: Results
This chapter discusses the results of the research. An orientation
geodatabase has been assembled. It has layers that give quantitative benthic data
a frame of reference, i.e. coastlines, rivers, nautical charts etc. A bathymetry
basemap has been assembled of the study area. The output of an Arctic Ice-
Ocean model has been adapted to the ArcView framework. Finally, quantitative
benthic data have been assembled in a relational database for further analysis and
interpolated surfaces of biomass have been produced. All final versions of
shapefiles and grids produced in this research have been written to an attached
CD.
5.1 DIGITAL BATHYMETRY MODEL
The basemap in Figure 5.1 is a bathymetry grid of 2.5 km cell size. It
covers the study area and is assembled from three bathymetry/elevation data
sources. A reliable bathymetry grid is essential to understand the relationship
between ocean depth and benthic productivity.
51
52
5.2 INTERPOLATION OF BENTHIC BIOMASS
The Arctic database has 641 records with measurements of benthic
biomass. A dotplot in Figure 5.2 shows the distribution of the benthic biomass
data while general statistics are shown in Table 5.1.
Figure 5.2 The Distribution of the Benthic Biomass Samples
Table 5.1 General Statistics of the Benthic Biomass Samples
Benthic biomass is sampled at stations, i.e. point locations, whereas
knowledge of the continuous benthic biomass pattern is desired. Therefore
interpolation of the point data is important. Using interpolation, biomass at
locations where biomass has not been sampled is predicted from neighboring
biomass measurements. Two approaches were used to interpolate the biomass
53
data, the Arc/Info Workstation function Topogrid and the Geostatistical Analyst
in ArcMap. The Topogrid function was developed with elevation data in mind,
while the Geostatistical Analyst gives a variety of interpolation methods and is a
great tool for the interpolation of all spatially continuous data.
The Geostatistical Analyst gives a variety of tools to analyze the data. The
kriging interpolation technique predicts a surface based on the statistical
characteristics of the data, and the reliability of the predicted surface can be
estimated spatially. While other interpolation techniques cannot determine the
reliability of the predicted surface on a spatial scale. Therefore kriging techniques
were used in this study.
5.2.1 Interpolation Using Topogrid
The Topogrid interpolation function in Arc/Info was designed to handle
elevation data. It uses a discretised thin plate spline technique that is intended to
show surface continuity without loosing the ability to follow abrupt changes in
terrain (Wahba, 1990). The function is not designed for data such as the biomass
data; however, this method of interpolation is simple and good for initial
interpolation of the data. The interpolation was done for both the raw biomass
data and for transformed data. The transformation involved taking the logarithm
of the biomass data before interpolation and thereby reducing the range of the
data. By reducing the variability less importance is given to very high biomass
values, whereas in the raw data the influence area of high values is relatively
larger than the influence area of a low value. The exact procedure is described in
Appendix A, while the two resulting grids can be compared in Figure 5.3.
54
Figure 5.3 Interpolation of Biomass with Topogrid,a)RawData,b)Transformed
Data
5.2.2 Interpolation Using the Geostatistical Analyst
The Geostatistical Analyst extension is a new package within the scope of
ArcMap. It is intended for the interpolation of any continuous variable from point
data. The user can analyze the statistical distribution of data before interpolation,
select the interpolation method, and assess the quality of the results. The interface
is well developed and a tutorial explains the procedures used by the program.
Therefore the user has a good opportunity to evaluate which interpolation is
appropriate and to understand the results.
The Geostatistical Analyst was used to analyze the benthic data by
interpolation of the biomass. The use of the kriging methods was emphasized in
the study; these methods use the statistical correlation between data points to
predict a surface. Prior to the work in ArcMap the table Benthos was analyzed in
the statistical program Minitab. By stepwise regression more correlation was
found between the biomass and salinity, than between biomass and depth or
temperature. An attempt was made to use the salinity output of the Arctic Ocean
55
model along with cokriging methods to predict biomass distribution. The
cokriging attempts gave, however, a higher prediction error than the kriging
models. Further study may reveal more valuable information on the relationship
between variables.
The Geostatistical Analyst gives the user an opportunity to explore the
dataset before interpolation; this can give an idea of what interpolation techniques
may be applicable to the specific data. The tools available for exploration are
shown in Figure 5.4 where a trend analysis of the biomass data shows a decrease
in benthic biomass with higher latitudes. The apparent trend in East-West
direction may be caused by the distribution of the sample points. Points in the
Chukchi and Bering Sea are both west and south of those in the Beaufort Sea.
Figure 5.4 A Trend Analysis of the Benthic Biomass
Some of the interpolation methods require a normal distribution of data.
The Normal QQPlot is a graph where quantiles from the distribution of biomass is
plotted relative to a normal distribution. The graph gives an indication of the
univariate normality while more general parameters of the distribution are
56
calculated by choosing Histogram. As seen in Figures 5.5 and 5.6, an assumption
of normal distribution of the benthic biomass could not be established. However,
the logarithmic transformation of the data gives a distribution fairly close to a
normal distribution. Therefore, if univariate normality is required for a
distribution, the logarithmic transformation may be a feasible option.
Figure 5.5 A Comparison of the Biomass Distribution to a Normal Distribution
Figure 5.6 Histograms of the Biomass Samples, Raw Data and Log-Transformed
57
Different interpolation methods and parameters can give a variety of
interpolated surfaces. For a selected dataset, trends can be removed and data can
be transformed before interpolation, as seen in Figure 5.7. The data will then be
untransformed and trends will be added before the final surface is produced. If a
trend is removed from the data, the next screen will show the trend surface
removed. Figure 5.8 shows a global first order detrending surface that is removed
from the untransformed biomass data.
Figure 5.7 A Geostatistical Method Selection
Figure 5.8 A Detrending Surface
58
After detrending the biomass data, the program calculates the
autocorrelation of the data and fits a model to the results. The Semivariogram in
Figure 5.9 shows the squared difference in biomass, ?, between points separated
by a certain distance, h. The yellow line presents the statistical model fitted to the
measurements.
Figure 5.9 A Semivariogram
Not every pair of points is shown; each point on the graph stands for a
group of pairs. The grouping is called binning; each bin spans a distance range
(lag size of 158.5 km in Figure 5.9), while anisotropy also divides pairs of point in
bins according to the angle between them. The model above is not anisotropic but
59
anisotropic models then create separate models for each direction interval, which
can be useful in accounting for the directional characteristics of the data.
However, if data are sparse, empty bins can exist and bins with few points for
averaging may create undesirable variations.
The nugget, range and sill along with a shape function define the
semivariogram model. The nugget describes spatial variations on the scale of
sampling or measurement error, i.e. the difference in results one should expect
from two samples at the same station. The range is the distance at which the
model flattens out, i.e. at what distance samples stop being spatially
autocorrelated. The sill gives the variations one should expect from samples
separated by a distance beyond the range.
For every point in the analysis area, the program assigns neighboring
points a weight based on the semivariogram model and takes a weighted average.
A calculation of standard error contours (Figure 5.10) gives an indication of
where the model may be trusted, while Figure 5.11 shows a statistical evaluation
of the general quality of the surface.
60
Figure 5.10 Biomass Interpolation and Standard Error Contours
Figure 5.11 Cross Validation of Model Output
61
The Geostatistical Analyst was used to interpolate the benthic biomass.
When the entire data set was analyzed, three ordinary kriging models appeared to
have the smallest prediction error. One of them had no trend removed, another
had a first order global trend removed and the third had the same trend removed
and an anisotropy accounted for. The range of biomass values is large (Table
5.1), higher values are generally in the Bering and Chukchi Sea while biomass in
the Beaufort Sea is significantly lower. Therefore, the benthic biomass dataset
was also analyzed in two parts; one model was created for the Chukchi and
Bering Sea and another for the Beaufort Sea.
5.2.3 Interpolation of the Entire Benthic Biomass Data Set
Three biomass surfaces were created for the entire benthic biomass data
set with an ordinary kriging technique. One model with default parameters,
another where a first order north-south trend was removed and a third one where
anisotropy was also accounted for. These models give similar results, as seen in
Figure 5.12, while a model where data where logarithmically transformed gave a
huge standard error. An attempt to model the biomass by connecting it to the
salinity concentration from the ice-ocean model, using cokriging, did not give
lower prediction error than with the biomass data alone. Therefore the choice of
models came down to the aforementioned three ordinary kriging models.
62
Figure 5.12 Prediction Error Comparison of Three Ordinary Kriging Models
If Z(s
i
) is the predicted value and z(s
i
) is the observed value and ?(s
i
)is
the prediction standard error for location s
i
, the mean prediction error is defined as
shown in Equation 5.1. The root mean-squared prediction error is defined in
Equation 5.2, the average standard error in Equation 5.3 and standardized
prediction errors in Equations 5.4 and 5.5.
n
szsZ
n
i
ii?
=
?
1
))()((
(5.1) Mean Prediction Error
n
szsZ
n
i
ii?
=
?
1
2
))()((
(5.2) Root Mean-Squared Prediction Error
n
s
n
i
i?
=1
)(?
(5.3) Average Standard Error
63
The comparison of the three interpolation models shows that the second
model, where first order trend was removed, has a smaller mean error and the
root-mean-square prediction error is also smaller than the other two models. The
average standard error in all three models is a lot smaller than the root mean
square error, which indicates that the models underestimate the variability of the
biomass. The second model does however account for more variability than the
others and seems to be the best model. The second model (Figure 5.13) is on the
attached CD; it is an ArcMap layer titled biomass. Prediction standard error
contours for the biomass model is also on the CD, titled bio_standerr.
n
sszsZ
n
i
iii?
=
?
1
)(/))()(( ?
(5.4) Mean Standardized Prediction Error
[]
n
sszsZ
n
i
iii?
=
?
1
2
)(/))()(( ?
(5.5) RMS Standardized Prediction Error
64
65
Areas of high benthic productivity can be identified by an evaluation of
where the benthic biomass is higher than a certain value. A probability map was
created with the same method as the biomass layer was constructed. The
probability layer shown in Figure 5.14, bio_prob50, shows the probability of
biomass exceeding a value of 50 g/m
2
. The probability layer is only visible in
areas where the mean prediction error is expected to be less than 100 g/m
2
.In
areas of a high prediction error the reliability of the probability layer is smaller.
Figure 5.14 Probability of Biomass Exceeding 50 g/m
2
5.2.4 Interpolation of the Benthic Biomass Data in Two Sets
The range of the benthic biomass data is large; the biomass in Chukchi
and Bering Sea is significantly higher than the biomass in the Beaufort Sea. The
interpolation of the entire dataset results in a good evaluation of the biomass in
regions where the biomass is high. However, the standard error of prediction is
very high relative to the biomass values in areas of low biomass. Therefore a
separate kriging model was produced for the Beaufort Sea alone. A separate
66
model for the Chukchi and Bering Sea did however give higher prediction error
than the model for the entire area as seen in Figure 5.15. The model for the entire
area is therefore more practical for the Chukchi and Bering Sea.
An ordinary kriging model for the Beaufort Sea alone, with a second order
global trend removed proved to have the lowest prediction error for the Beaufort
Sea. The default parameters in the Geostatistical Analyst were used, and a
statistical evaluation of the model is shown in Figure 5.15. The model for
Beaufort Sea, bio_beaufort (Figure 5.16) shows more details in the distribution
of benthic biomass in the Beaufort Sea than the model for the entire study area
(Figure 5.13). The standard error of prediction is also significantly lower in the
Beaufort Sea model and the results are therefore more dependable in the Beaufort
Sea than the results from the model for the entire area.
Figure 5.15 The Ordinary Kriging Model for the Beaufort Sea
67
68
A probability map, beaufort_prob30, was produced for the Beaufort Sea
to identify areas of high benthic productivity from the shapefile beaufort.The
map (Figure 5.17) shows the probability of benthic biomass to succeed 30 g/m
2
.
The standard error contours are saved as beaufort_standerr. The map only
shows the probability evaluation where prediction is very reliable. The predicted
standard error in the area shown in Figure 5.17 is lower than 35 g/m
2
.
Figure 5.17 Probability of Biomass in Beaufort Sea Exceeding 30 g/m
2
The Chukchi and Bering Sea have much more benthic biomass than the
Beaufort Sea. The probability map in Figure 5.18, chukchi_prob250,showsthe
probability of benthic biomass to exceed 250 g/m
2
. The probability evaluation is
only visible in areas where the predicted standard error is lower than 100 g/m
2
.
69
Figure 5.18 Probability of Biomass in Chukchi and Bering Sea Exceeding 250
g/m
2
5.2.5 Interpretation of the Interpolated Benthic Biomass Maps
The prediction standard error contours give important information of the
dependability of the benthic biomass interpolation. Examples of how well the
maps can predict the benthic biomass were taken at four locations in the study
area. The locations of the example points are shown in Figure 5.19. One of the
samples, A is in the middle of the Chukchi Sea. Point B is further out to the west,
where no benthic biomass samples were close by. Point C is in the middle of an
area of high benthic biomass in the Beaufort Sea and Point D is in a region of
very low biomass in the Beaufort Sea.
70
Figure 5.19 The Location of the Example Points, A, B, C and D
Table 5.2 shows the predicted values at the four locations and their
standard errors. Values from both the model for the entire area and the model for
the Beaufort Sea are shown for points C and D. The following figures show the
probability density of benthic biomass in the example locations. A normal
distribution is assumed at the point locations, the mean of the distribution is the
predicted value and the standard deviation is the evaluated standard error.
Table 5.2 Predicted Benthic Biomass and Standard Error for Selected Locations
A t-statistical test shows whether the biomass is significantly different
from zero. If t = ?/S
e
> 2 the biomass is significant, otherwise not. According to
71
the model for the entire dataset, the biomass is only significant for points A and
B. While according to the model for the Beaufort Sea, the biomass is also
significant for point C.
A normal distribution probability density function describes the prediction
dependability. If ? is the mean (predicted value) and ? is the standard deviation
(standard error) the probability density is describet by Equation 5.6.
Figures 5.20 through 5.23 show the probability density for points A, B, C
and D, according to the model for the entire study area, while Figure 5.24 shows
the probability functions for A, B, C and D on one graph for comparison.
Figure 5.20 The Benthic Biomass Prediction at Point A
?
?
?
?
?
?
?
? ??
=
2
2
2
)(
exp
2
1
)(
?
?
??
x
xf (5.6) The Probability Density
72
Figure 5.21 The Benthic Biomass Prediction at Point B
Figure 5.22 The Benthic Biomass Prediction at Point C
73
Figure 5.23 The Benthic Biomass Prediction at Point D
Figure 5.24 The Benthic Biomass Prediction at all Four Points
The first graph, Figure 5.20, shows the benthic biomass at point A, in the
Chukchi Sea. The predicted value at point A is 800 g/m
2
and the probability
density curve shows that the benthic biomass is well determined. The uncertainty
74
is in the range of ? 100 g/m
2
. The second graph, Figure 5.21 shows the benthic
biomass at a point B in the Chukchi Sea. Point B is fairly far from any benthic
measurements and the biomass is in the range of 320 ? 150 g/m
2
, the uncertainty
of the prediction is larger than for point A. The biomass at the points C and D
(Figure 5.24) is not well determined. The predicted value at C and D is low
compared to the standard error that results in large uncertainty in benthic biomass
evaluation. The model for Beaufort Sea does however predict the biomass in the
Beaufort Sea area better than the model for the entire area. The probability
density of benthic biomass at points C and D according to the Beaufort Sea model
is shown in Figure 5.25 and 5.26.
Figure 5.25 The Benthic Biomass Prediction at Point D, Beaufort Sea Model
75
Figure 5.26 The Benthic Biomass Prediction at Point D, Beaufort Sea Model
Figure 5.25 shows the probability density for point C according to the
Beaufort Sea model. The benthic biomass is predicted with more confidence than
in the model for the entire area (Figure 5.27). According to the t-statistics the
prediction for point C in the first model is not significantly different from zero.
While in the Beaufort Sea model, t = 2.5 > 2, shows that the benthic biomass at
point C is significant. According to the Beaufort Sea Model the benthic biomass
at point C is in the range of 75 ? 30 g/m
2
. The biomass at point D is very low and
cannot be predicted well since the standard error is high compared to the
predicted value (t = ?/S
e
= 0.1 < 2). The biomass at point D is 3 ? 30 g/m
2
.
76
Figure 5.27 Comparison of the Two Models at Point C
The model for the entire area appears to predict biomass well in areas with
high biomass. A comparison of the standard error contours to the predicted
biomass indicates how well the biomass is predicted. The Beaufort Sea model
appears to predict the low benthic biomass in the Beaufort Sea significantly better
than the model for the entire area. The model gives lower standard error values
and shows regions with high benthic biomass in more detail than the model for
theentirearea.
Both of the benthic biomass models (Figure 5.13 and 5.16) show areas of
high benthic biomass in the Southern Chukchi Sea and the Bering Sea, where the
predicted biomass exceeds 700 g/m
2
. Beaufort Sea has in general low biomass,
less than 10 g/m
2
except in the vicinity of the mouth of large rivers. In the
vicinity of Colville River the biomass reaches 50?100 g/m
2
and east of Mackenzie
River the predicted biomass reaches 30 g/m
2
.
77
Chapter 6: Conclusions
6.1 INTRODUCTION
This research focuses on a GIS Analysis of the Western Arctic Ocean.
The objective is to establish extensive knowledge of the spatial and temporal
characteristics of the benthic community in the Beaufort, Chukchi and Northern
Bering Sea. Available quantitative benthic data for the area were assembled in a
relational database and analyzed in the ArcView and Arc/Info geographic
information systems. Various other geographical data in digital format were also
assembled to be viewed and analyzed along with the benthic data. Benthic
productivity is strongly dependent on the existence of nutrients and physical
factor such as ocean depth, temperature and currents.
This study shows that GIS is an effective platform for the analysis of the
benthic community in the Western Arctic Ocean. The visualization and analysis
of spatially distributed data can reveal clear relationships between physical and
biological factors such as inflow of fresh water and areas of high benthic
productivity
6.2 PREPARATION OF GEOSPATIAL DATABASE
The objectives of the study were to assemble geographically referenced
data and develop methods for further analysis of the benthic community in the
Western Arctic Ocean. A variety of information has been assembled and
78
preliminary analysis of benthic biomass has been performed. A detailed
description of original data, methods and products can be found in this report.
The description can therefore be used as a guide through further analysis with
additional benthic data.
Various sources of digital geographical data were explored and data
extracted. Data such as NOAA nautical charts, coastlines and rivers from the
Digital Chart of the World, now serve as orientation files for physical and
biological variables. The coastlines and rivers give a general reference system,
while the detailed nautical charts include valuable information of landmarks and
other details.
Several digital bathymetry/elevation models were studied and a
bathymetry grid, basemap, with a cell size of 2.5 km was successfully assembled
for the study area. The basemap grid is based primarily on the International
Bathymetric Chart of the Arctic Ocean (IBCAO).
Methodology was developed to import the output of a grid based Ice-
Ocean Arctic Model into ArcView. The model has a cell size of 18 km and gives
information on ocean temperature, salinity and currents, as monthly averages for
the whole Arctic Ocean. Twelve monthly files for the year of 1993 have been
processed so temperature and salinity as grids, and velocity vectors as flux arrows
can be viewed in ArcView.
79
6.3 INTERPOLATION OF BENTHIC BIOMASS IN THE WESTERN ARCTIC OCEAN
Quantitative benthic data were assembled in a relational database, where
their distribution can be viewed in ArcView. A GIS layer with the benthic
biomass data was analyzed with a Geostatistical Analyst in ArcMap where areas
of high benthic biomass could be identified with a kriging interpolation technique.
An ordinary kriging model was developed for the entire study area. Along
with benthic biomass values the prediction standard error was evaluated. The
prediction standard error maps show that the model predicts benthic biomass well
in the vicinity of benthic measurements in the Chukchi and Bering Sea. The
range of benthic biomass measurements was high; the benthic biomass in the
Beaufort Sea is significantly lower than the biomass in Chukchi and Bering Sea.
The kriging model does not predict the biomass with as much certainty in areas
with low benthic biomass. Regions with high biomass had greater influence on
the structure of the kriging than regions with low biomass.
Since the model for the entire area does not predict biomass with much
certainty in the Beaufort Sea, a separate model was created from benthic biomass
measurements in the Beaufort Sea. Regions of relatively high biomass can be
well identified in the Beaufort Sea model. Prediction standard error is on the
order of 30 g/m
2
, therefore, the model does not give much information for very
low benthic biomass values. This model does not determine well whether benthic
biomassisontheorderof1g/m
2
or 10 g/m
2
. Further analysis of log-transformed
data may give more information of areas with low benthic biomass.
80
Regions of high benthic biomass (above 700 g/m
2
) are apparent in the
Southern Chukchi and Bering Sea where the warm waters of the Pacific Ocean
mix together with the colder Arctic Ocean. The biomass is in general lower in the
Beaufort Sea (below 10 g/m
2
) while fresh water inflow from major rivers appears
to result in increased biomass in the vicinity of the rivers mouths. The biomass
west of Colville River reaches 50-100 g/m2 and east of Mackenzie River the
biomass reaches 30 g/m
2
.
6.4 RECOMMENDATIONS FOR FUTURE WORK
This research was directed primarily towards preparation of a geospatial
database of the study area, Western Arctic Ocean, whereas a further study could
aim at a more intensive analysis. The next steps would then be to find more
quantitative benthic data, research the temporal variability in the benthic data,
analyze the distribution of specific species in the study area and research the
relationship between benthic biomass and physical or chemical variables.
Quantitative benthic data have been assembled from various sources,
however more data would give more reliable results. The search for other data
sourcesisimportant.
Benthos are often considered integrators of the productivity in the water
column above the ocean floor. They show little seasonal variation in abundance
or biomass, and are good indicators for long term variation. In this research no
attempt has been made towards quantifying long-term changes. A study of the
81
time trends of the data may reveal important information on long-term changes,
which may be connected to global climate changes.
The quantitative benthic data often include information on what specific
benthic species inhabit the ocean floor. In this research the cumulative biomass
has been studied while a visualization of spatial distribution of specific species
reveals interesting patterns.
Spatially distributed information of physical and chemical variables may
be used to model the benthic biomass. The relationship between different
variables may be explored either with further use of the Geostatistical Analyst or
other statistical programs. Knowledge of links between benthic biomass and
physical/chemical factors may then be used to establish a good understanding of
the effects of global climate change on the benthic community of the Western
Arctic Ocean.
82
Appendix A: Scripts/Programs and Procedures
AML Projection Files:
Geo2lamb.txt
Pol2lamb.txt
Azi2lamb.txt
Command sequences, AMLs etc.:
Arc/Info commands for the creation of the frame coverage
Arc/Info commands for the creation of the fivedeg coverage
Arc/Info commands for the creation of the ETOPO5 grid
Matlab commands for the creation of the IBCAO grid
Arc/Info commands for the creation of the basemap grid
Commands for converting the output of the ocean model data to grids
Procedure for creating findflux_lamb for the ocean model
Procedure for interpolation with Topogrid, creating the biomass grids
Scripts:
Mergethemes.ave
Get_Extent.ave
Addtopo.ave
Findflux.ave
Arrow.ave
83
AML Projection Files:
Geo2lamb.txt
- Was used to project various layers from geographic projection to Lambert
Azimuthal Equal-Area projection
Azi2lamb.txt
- Was used to project the Ice-Ocean Arctic model output to Lambert
Azimuthal Equal-Area projection
Pol2lamb.txt
- Was used to project IBCAO grid from Polar Stereographic projection to
Lambert Azimuthal Equal-Area projection
Geo2lamb.txt:
input
projection geographic
units dd
parameters
output
projection
lambert_azimuth
units meters
parameters
0
-165 00 00
90 00 00
2000000
4000000
end
Azi2lamb.txt
input
projection azimuthal
units meters
parameters
0
-140 00 00
90 00 00
2233051
2566645
output
projection
lambert_azimuth
units meters
parameters
0
-165 00 00
90 00 00
2000000
4000000
end
Pol2lamb.txt :
Input
projection polar
units meters
parameters
00 00 00
75 00 00
0
0
output
projection
lambert_azimuth
units meters
parameters
0
-165 00 00
90 00 00
2000000
4000000
end
84
Arc/Info commands for the creation of the frame coverage (AML):
Most of the layers and grids in the project are clipped to the extent of this frame;
Figure 3.1 shows layers clipped by the frame.
Arc: &run framegeo.txt
Framegeo.txt:
generate framegeo
polygons
1,170,70
160,62
160,75
180,75
180,62
160,62
end
2,-170,70
-180,62
-180,75
-128,75
-128,62
-180,62
end
end
quit
Arc: clean framegeo
Arc: project cover framegeo frame geo2lamb.txt
Arc: clean frame
85
Arc/Info commands for the creation of the fivedeg coverage (AML):
The fivedeg coverage is shown in Figure 4.1.
Arc: Generate fivedeg_geo
Generate: fishnet
Origin Coordinate (X,Y): 160, 60
Y-Axis Coordinate (X,Y): 160, 5
Cell Sixe (Width, Heigth): 5, 5
Number of Rows, Columns: 3, 15
Generate: quit
Arc: build fivedeg_geo
Arc: project cover fivedeg_geo fivedeg geo2lamb.txt
Arc: build fivedeg
Arc/Info commands for the creation of the ETOPO5 grid (AML):
The ETOPO5 grid is shown in Figure 3.11
Arc: Shapearc etopopt_geo etopopt_geo
Arc: Project etopopt_geo etopo_point geo2lamb.txt
Arc: topogrid etopo 5000 (5 km)
Topogrid: datatype spot
Topogrid: point etopo_point Z
Topogrid: enforce off
Topogrid: end
Arc: Grid
Grid: etopo2 = selectpolygon(etopo, frame, inside)
Grid: etopo_floor = floor(etopo2 + 0.5)
Grid: ETOPO5 = int(etopo_floor)
The topogrid commands are the same as for the other bathymetry/ elevation grids
discussed in Chapter 3.2. The except cell size does however vary, it is 5 km here
but 2.5 km otherwise. Also the name of the elevation fields may have been
different for some of the other grids.
86
Matlab commands for the creation of the IBCAO grid:
The IBCAO grid is shown in Figure 3.13.
DLM = ?/?
- The delimitor between numbers is tab
RNG = [7 0 1000005 0]
- Range set to start at row 7 column 0 to row 1000005 column 0
A1 = DLMREAD('IBCAO.asc',DLM,RNGA)
- Read the part of the ASCII file IBACAO.asc that is within the range of
RNG into A1
A2 = rot90(reshape(A1,3,333333))
- The numbers were read into one long column, here they are divided up
into three columns again, X, Y and Z and stored in the matrix A2
DLM2 = ?,?
- The delimitor of the output file is a comma (,)
dlmwrite('A3',A2,DLM2)
- A file A3 is written from the matrix A2, it has 333333 records.
This was done for the first 1.8 million records, which span the study area,
i.e. the area above 64 N and to some extent somewhat below, since the extent is a
square in Polar Stereographic coordinates, which does not translate to an exact
circle in Geographic coordinates.
87
Arc/Info commands for the creation of the GTOPO30 grid (AML):
The GTOPO30 elevation grid is shown in Figure 3.14.
Arc: imagegrid E140N90 e140
Arc: grid
Grid: ee140 = con(e140 = 32768, e140 - 65536, e140)
Grid: kill e140 all
Grid: e140 = setnull( ee140 == -9999, ee140)
Merge the three files:
Grid: gtopo30_geo = merge( e140, w180, w140)
Grid: quit
Arc: project gtopo30_geo gtopo30_lam geo2lamb.txt # 1000 232293.734
916932.875
Grid: setcell 1000
Grid: setwindow flamb165
Grid: gtopo30_1 = selectpolygon(gtopo30_lam, flamb165, inside)
Grid: gtopo_floor = floor(gtopo2 + 0.5)
Grid: GTOPO30 = int(gtopo_floor)
88
Arc/Info commands for the creation of the basemap grid (AML):
Creating a grid from the IBCAO grid and the part of the ETOPO5 grid that the
IBCAO grid does not cover. The point coverages are already in a Lamberts
projection.
Arc: topogrid ibcao+etopo 2500 (2.5 km)
Topogrid: datatype spot
Topogrid: point ibcao_point z
Topogrid: point etopo_selpoint z
Topogrid: enforce off
Topogrid: end
Arc: grid
Grid: ibcao_etopo = selectpolygon ( ibcao+etopo, frame, inside)
Grid: ibet_floor = floor(ibcfive + 0.5)
Grid: ibet_int = int(ibet_floor)
Merge the bathymetry grid with the GTOPO30 elevation grid.
Arc: grid
Grid: setcell ibet_int
Grid: setwindow ibet_int
Grid: basemap = merge(gtopo30, ibet_int)
The final basemap isshowninFigure5.1.
89
Commands for converting the output of the ocean model data to grids:
Add the following header to the text files and save as, for instance, salt_1.txt:
ncols 368
nrows 304
xllcorner 0
yllcorner 0
cellsize 18532
nodata_value 1e+030
Convert to an ArcInfo grid by typing the following:
Arc: Asciigrid salt_1.txt flipsalt_1 float
The resulting grid is a mirror image of what it should be; therefore flip it:
Arc: grid
Grid: salt_1 = flip(flipsalt_1)
Now the grid can be viewed in ArcInfo/ArcView as shown in Figure 3.17.
90
Procedure for creating findflux_lamb for the ocean model:
For every u component grid (u_1) and the v component (v_1) extract the point
coverages.
Arc: Gridpoint u_1 u1_pt U
Arc: Gridpoint v_1 v1_pt V
From one of the velocity component coverages extract the coordinates of the
model points.
Arc: Ungenerate point u1_pt modelpt.txt
The modelpt.txt file comes out as a ?tab? delimited file but by importing it into
Excel, adding header, ?Point_id, X, Y? , saving it as a comma delimeted file
?.csv? and in notepad resaving it as a ?.txt? file it is formatted for ArcView.
Now in ArcView add the text file modelpt.txt as a table and add it to the view
with ?View-> Add Event Theme? Convert to a shapefile with ?Theme->Convert to
shapefile? and save as modelpt.shp
Now add the u_pt and v_pt files to the view and open the attribute tables of all the
themes. Start editing the modelpt table by ?Table->Start Editing? Add fields for
the u and v components by ?Edit->Add Field? and name them u1_93 and v1_93
(according to the date of the files) Then join the attribute table of modelpt with
the velocity component tables through their ID fields by ?Table->Join?andwith
the u1_93 field active select the ?Field->Calculate? and set the field equal to the
U field of the u1_pt field, and the same for the v components. Then save the table
and remove all joins.
Add the scripts plotflux.ave, arrow.ave and findmod.ave to the project and
compile them. Execute the script plotflux.ave on either the whole set of point or a
selected set of points. And the arrows appear as shown in Figure 4.9.
91
Procedure for interpolation with Topogrid, creating the biomass grids:
Addafieldtothesamples layer, with the logarithm to base 10 of the biomass
data. Then create a 100 km buffer around the biomass sample data. In Arc View
set distance and map units to meters. Then select the samples shapefile and chose
Theme -> Create buffers. Set the specified distance at which the buffers should
be to 10,000 meters, click next and save the buffer in a new theme, buffer100km.
Create polygons, land, from the coastline layer, ponet, in ArcInfo (Arc: build
ponet). Intersect the outline polygon of the study area, frame, with the land layer
using the Geoprocessing wizard in Arc View. Select the ocean parts of the
resulting file and convert to shapefile, ocean.
InArcInfocreatethetwobiomassgrids:
Arc: shapearc buffer100km buffer100km
Arc: build buffer100km
Arc: shapearc ocean ocean
Arc: build ocean
Arc: topogrid biomass 2500
Topogrid: bio1 2500 (2.5km cell size)
Topogrid: datatype spot
Topogrid: point biomass_l biomass__g
Topogrid: enforce off
Topogrid: boundary ocean
Topogrid: xyzlimits####0#
Topogrid: end
Arc: topogrid biolog 2500
Topogrid: datatype spot
Topogrid: point biomass_l log_biomass
Topogrid: enforce off
Topogrid: boundary ocean
Topogrid: end
Arc: grid
Grid: setcell bio1
Grid: bio2 = selectpolygon(bio1,ocean,inside)
Grid: biomass = selectpolygon(bio2,buffer100km,inside)
Grid: biolog2 = selectpolygon(biolog, ocean,inside)
Grid: biolog3 = selectpolygon(biolog2,buffer100km,inside)
Grid: biolog10 = exp10(biolog3)
- Now the grids biomass and biolog10 are both interpolated benthic biomass as
showninFigure5.3.
92
Mergethemes.ave
This Avenue script merges two or more shapefiles in ArcView
'MergeThemes
theView=av.GetActiveDoc
theThemes=theView.GetThemes
if (theThemes.Count<2) then
MsgBox.Error("Must have at least two themes in a view to merge.","")
exit
end
themesToMerge=List.Make
while (true)
t=MsgBox.Choice(theThemes,"Choose themes in view to merge:"+NL+"(Click Cancel to
end):","Merge Themes")
if (t<>Nil) then
themesToMerge.Add(t)
else
break
end
end
if ((themestoMerge=Nil) or (themesToMerge.Count<2)) then
MsgBox.Error("Not enough themes to merge.","")
exit
end
checkType=themesToMerge.Get(0).GetFTab.FindField("Shape").GetType
for each i in 1..(themesToMerge.Count-1)
t=themesToMerge.Get(i)
if (checkType<>t.GetFtab.FindField("Shape").GetType) then
MsgBox.Error("Theme feature type mismatch -- unable to merge.","")
exit
end
end
outFName=av.GetProject.MakeFileName("theme","shp")
outFname=FileDialog.Put(outFname,"*.shp","Output Merged Shapefile")
if (outFName=Nil) then
exit
end
fieldList=List.Make
for each f in themesToMerge.Get(0).GetFTab.GetFields
if (f.GetName="Shape") then
continue
else
93
fCopy=f.Clone
fieldList.Add(fCopy)
end
end
shapeType=themesToMerge.Get(0).GetFTab.FindField("Shape").GetType
if (shapeType=#FIELD_SHAPELINE) then
outClass=POLYLINE
elseif (shapeType=#FIELD_SHAPEMULTIPOINT) then
outClass=MULTIPOINT
elseif (shapeType=#FIELD_SHAPEPOINT) then
outClass=POINT
elseif (shapeType=#FIELD_SHAPEPOLY) then
outClass=POLYGON
else
MsgBox.Error("Invalid shape field type.","Merge Themes")
exit
end
mergeFTab=FTab.MakeNew(outFName,outClass)
if (fieldList.Count>0) then
mergeFTab.AddFields(fieldList)
end
for each t in themesToMerge
av.ShowMsg("Merging"++t.GetName)
inFTab=t.GetFTab
if (inFTab.GetSelection.Count=0) then
theRecordsToMerge=inFTab
numRecs=inFTab.GetNumRecords
else
theRecordsToMerge=inFTab.GetSelection
numRecs=theRecordsToMerge.Count
end
for each rec in theRecordsToMerge
av.SetStatus((rec/numRecs)*100)
newRec=mergeFTab.Addrecord
inField=inFTab.FindField("Shape")
outField=mergeFTab.FindField("Shape")
mergeFTab.SetValue(outField,newrec,inFTab.ReturnValue(inField,rec))
if (fieldList.Count>0) then
for each f in fieldList
fName=f.GetName
inField=inFTab.FindField(fName)
if (inField<>Nil) then
outField=mergeFTab.FindField(fName)
aValue=inFTab.ReturnValue(inField,rec)
mergeFTab.SetValue(outField,newRec,aValue)
end
end
94
end
end
end
av.ClearMsg
av.ClearStatus
if (MsgBox.YesNo("Add shapefile as theme to view?","MergeThemes",true).Not)then
exit
end
viewList={}
for each d in av.GetProject.GetDocs
if (d.Is(View)) then
viewList.Add(d)
end
end
viewList.Add("")
addToView=MsgBox.ListAsString(viewList,"Add Theme to:","Merge Themes")
if (addToView<>Nil) then
if (addToView="") then
addToView=View.Make
addToView.GetWin.Open
end
mergeTheme=FTheme.Make(mergeFTab)
addToView.AddTheme(mergeTheme)
addToView.GetWin.Activate
end
95
Get_Extent.ave
This Avenue script extracts the extent of selected themes in ArcView and creates
a rectangle with the same extent. The program was used to build the NOAAmesh
showninFigure4.2.
?Get extent of selected layers
theView = av.GetActiveDoc
for each t in theView.GetActiveThemes
'theTheme2=theView.getactivethemes.get(0)
theval=t.getname
msgbox.warning(theval,"Warning")
' Change the extent of the DRG, i.e. cut the edges off.
r = Rect.MakeEmpty
r = r.UnionWith(t.ReturnExtent)
'r = theView.ReturnUserRect
theTheme = theView.GetEditableTheme
if (r.IsNull) then msgbox.warning("r does not work","Warning")
'returnnil
else
if (theTheme <> nil) then
p = r.AsPolygon
thePrj = theView.GetProjection
if (thePrj.IsNull.Not) then
p = p.ReturnUnprojected(thePrj)
end
theTheme.getFTab.BeginTransaction
theField = theTheme.GetFTab.FindField("Shape")
rec = theTheme.GetFTab.AddRecord
theTheme.GetFTab.SetValue(theField, rec, p)
theTheme.GetFTab.EndTransaction
theTheme.GetFTab.GetSelection.ClearAll
theTheme.GetFTab.GetSelection.Set(rec)
theTheme.GetFTab.UpdateSelection
else
gr = GraphicShape.Make(r)
theView.GetGraphics.UnselectAll
gr.SetSelected(TRUE)
theView.GetGraphics.Add(gr)
end
av.GetProject.SetModified(true)
end
end
96
AddNOAA.ave
This Avenue script is used with the Hotlink button in ArcView to select a NOAA
nautical chart and add it to the view as discussed in Chapter 4.2.3.
' Name: addnoaa
' Title: Adds a topographic map to a view (Noaa) and zooms in...
'
'Topics: Views
'
' Autor: Jona Finndis Jonsdottir jonafinndis@mail.utexas.edu, 12/18/1999
'
' Instruction: - replace the noaamesh.shp name with the name of your mesh
' - change f:/noaa/ to the path of where the NOAA Nautical Charts are
theVal = SELF
theView = av.GetActiveDoc
if (not (theVal.IsNull)) then
theVal2 = "f:/noaa/" + TheVal + ".kap".AsString
lookupname = "Attributes of noaamesh.shp"
lookuptab=av.getproject.finddoc(lookupname)
if (lookuptab=nil) then
msgbox.info("Can't find Attribute noaamesh.shp","")
exit
end
lookupvtab=lookuptab.getvtab
theField = lookupvtab.FindField("chart")
found = false
p = theView.GetDisplay.ReturnUserPoint
for each t in theView.GetActiveThemes
if ((t.HasAttributes) and (t.GetHotField <> nil)) then
recs = t.FindByPoint(p)
for each rec in recs
found = true
theVal3 = t.ReturnValueString(theField.GetName, rec)
end
end
end
if (not found) then
System.Beep
end
97
if (not (File.Exists(theVal2.AsFileName))) then
MsgBox.Info("Insert " +theVal3+ " CD","" )
end
if (File.Exists(theVal2.AsFileName)) then
' Create the SourceName...
theSrc = SrcName.Make(theVal2)
'UsetheSourceNametomakeatheme...
aTheme = Theme.Make(theSrc)
' Add the theme to the view...
theView.AddTheme(aTheme)
' Set a new name for the theme...
aTheme.SetName(theVal)
' Draw the theme...
aTheme.SetVisible(true)
' Make txmesh unactive
for each t in theView.GetActiveThemes.clone
t.SetActive( false )
end
aTheme.SetActive(true)
' Zoom
av.GetProject.SetModified(true)
theThemes = theView.GetActiveThemes
r = Rect.MakeEmpty
for each t in theThemes
r = r.UnionWith(t.ReturnExtent)
end
if (r.IsEmpty) then
return nil
elseif(r.ReturnSize=(0@0))then
theView.GetDisplay.PanTo(r.ReturnOrigin)
else
theView.GetDisplay.SetExtent(r.Scale(1.1))
end
else
MsgBox.Warning("File "+theVal2+" not found.","Hot Link" )
end
end
98
Findflux.ave
This Avenue script is used to create the current flux arrows shown in ArcView
from the Ice-Ocean Arctic Model data, as shown in Figure 4.9.
'----------------------------
'--- Creation information ---
'----------------------------
'Name: findflux.ave
'Date: 9/95
'Author: Ye Zichuan
' Center for Research in Water Resources
' The University of Texas at Austin
' ye@crwr.utexas.edu
'Revisions:
'1. 9/16/96 by Ferdi Hellweger
' - cosmetic changes
' - added comments in code
'2. 6/12/2000 by Jona Finndis Jonsdottir
' - Changes to fit Arctic Ocean Data
'---------------------------
'--- Purpose/Description ---
'---------------------------
'This program plots flux vectors for data computed from Arctic Ocean data
'-------------
'--- Input ---
'-------------
'1. Theme of the region boundary.
' - point shape file.
' - 'point_id' field storing the point's id.
' - 'X' and 'Y' fields storing the point's coordinates, x and y.
' - 'U' and 'V' fields storing velocity components
'--------------
'--- Output ---
'--------------
'1. Flux vectors as graphics in view frame.
'------------------------------
'--- Get general parameters ---
'------------------------------
'--- Get project ---
theproject=av.getproject
'--- Get view ---
theview = av.getactivedoc
'--- Get graphics ---
vgraphics=theview.getgraphics
orgn=vgraphics.count
'--- Get theme ---
thethemes=theview.getthemes
99
if (thethemes.count = 0) then
msgbox.error("No themes found", "Arctic Ocean")
exit
end
if (thethemes.count = 1) then
thetheme = thethemes.get(0)
msgbox.info("Using " + thetheme.asstring + " as theme for flux vector plotting", "Arctic
Ocean")
end
if (thethemes.count > 1) then
thetheme=msgbox.choiceasstring(thethemes,"Select the theme for flux vector plotting","Arctic
Ocean")
end
if(thetheme = nil)then
exit
end
monthlist = { "1_93", "2_93", "3_93", "4_93","5_93", "6_93","7_93", "8_93","9_93",
"10_93","11_93", "12_93"}
themonth=msgbox.choiceasstring(monthlist,"Select the month for flux vector plotting","Arctic
Ocean")
projectionlist = { "Lambert Azimuthal Projection", "Equidistant Azimuthal Projection" }
projection=msgbox.choiceasstring(projectionlist,"Select the projection for flux vector
plotting","Arctic Ocean")
if (projection = "Lambert Azimuthal Projection") then
proj = "lamb"
else
proj = "azi"
end
'--- Get display extent ---
ext = theview.getdisplay.returnextent
ox = ext.returnorigin.getx
oy = ext.returnorigin.gety
ex = ext.getwidth
ey = ext.getheight
pt = point.make((ox+(ex*0.66)),(oy+(ey*0.872)))
len = (theview.getdisplay.returnextent.getwidth)/80
'-----------------------
'--- Set date format ---
'-----------------------
date.setdefformat("M_yy")
'-------------------------
'--- Set up input data ---
'-------------------------
'--- point theme ---
lftab=thetheme.getftab
if (lftab = nil) then
msgbox.error("Can't open point theme","Arctic Ocean")
100
exit
end
lshapef = lftab.findfield("shape")
if (lshapef = nil) then
msgbox.error("Can't find point shape field","Arctic Ocean")
exit
end
lcx = lftab.findfield("X_" + proj)
if (lshapef = nil) then
msgbox.error("Can't find center x field","Arctic Ocean")
exit
end
lcy = lftab.findfield("Y_" + proj)
if (lshapef = nil) then
msgbox.error("Can't find center y field","Arctic Ocean")
exit
end
llidf = lftab.findfield("point_id")
if (llidf = nil) then
msgbox.error("Can't find point_id field in point theme","Arctic Ocean")
exit
end
lrecs = lftab.getnumrecords
lftabsel = lftab.GetNumSelRecords
if (lftabsel = 0) then
selectlist = { "No", "Yes"}
select=msgbox.choiceasstring(selectlist,"Do you want flux plotted for every point in the
layer","Arctic Ocean")
if (select = "No") then
exit
else
lftab.GetSelection.SetAll
lftab.UpdateSelection
end
end
'------------------------------------
'--- Set scale factor for vectors ---
'------------------------------------
relarrow=msgbox.input("Choose a Relative Length for Arrows","Arctic Ocean","10")
maxflux = 0
utime = lftab.findfield("U" + themonth)
if (utime = nil) then
msgbox.error("Can't find time field in attribute table","Arctic Ocean")
exit
end
for each lfrec in lftab.getselection
101
if (lftab.returnvalue(utime, lfrec) > maxflux) then
maxflux = lftab.returnvalue(utime, lfrec)
end
end
scalefactor = theview.getdisplay.returnextent.getwidth / (100 * maxflux / relarrow.asnumber)
'@msgbox.error("Scalefactor= "+ scalefactor.asstring,"Arctic Ocean")
'---------------------------------
'--- Flux vector plotting loop ---
'---------------------------------
mydate = date.make(themonth, "M_yy")
firsttime = true
plot = true
while(true)
if (firsttime.not) then
plot = false
end
firsttime = false
if (plot) then
'--- Cut old graphics --
mglist = theview.getgraphics
gfcnt = mglist.count-1
if (gfcnt > orgn) then
for each g in orgn..gfcnt
mglist.get(g).setselected(true)
end
mglist.clearselected
end
'--- Plot flux vectors ---
error = false
foreachlrecinlftab.getselection
av.setstatus(lrec / lrecs * 100)
lidfstr = lftab.returnvalue(llidf, lrec).asstring
if (lidfstr = nil) then
msgbox.error("Can't read from point_id field in point theme","Arctic Ocean")
exit
end
mydateu = lftab.findfield("u" + mydate.asstring)
mydatev = lftab.findfield("v" + mydate.asstring)
quq = lftab.returnvalue(mydateu, lrec)
if (quq = nil) then
msgbox.error("Can't read flux from u flux table","Arctic Ocean")
exit
end
qvq = lftab.returnvalue(mydatev, lrec)
if (qvq = nil) then
msgbox.error("Can't read flux from v flux table","Arctic Ocean")
exit
end
102
lshape = lftab.returnvalue(lshapef, lrec)
if (lshape = nil) then
msgbox.error("Can't read point shape","Arctic Ocean")
exit
end
partlist = lshape.asmultipoint.aslist
pcnt = partlist.count - 1
fpt = partlist.get(0)
tpt = partlist.get(pcnt)
cx = lftab.returnvalue(lcx, lrec)
cy = lftab.returnvalue(lcy, lrec)
fpt = point.make(cx,cy)
tpt = point.make((cx + (quq * scalefactor)),(cy + (qvq * scalefactor)))
appoint = polyline.make({{fpt,tpt}})
apshp = graphicshape.make(appoint)
apsym = apshp.getsymbol
apsym.setcolor(color.getred)
avpoint = av.run("arrow.ave",{fpt,tpt,(15.0).asradians,len})
vgraphics.add(apshp)
avsym=avpoint.getsymbol
avsym.setcolor(color.getred)
vgraphics.add(avpoint)
end
textstr = mydate.asstring
mytext = graphictext.make(textstr,pt)
vgraphics.add(mytext)
av.setstatus(100)
end
'
'--- No more plotting ---
'
if (plot.not) then
av.setstatus(100)
exit
end
end
'-----------
'--- End ---
'-----------
Arrow.ave
103
This Avenue Script is used along with the findflux.ave script to plot the flux
arrows in ArcView. The findflux.ave script will not work if this script is not
included in the ArcView project.
'----------------------------
'--- Creation information ---
'----------------------------
'Name: arrow.ave
'Date: 9/95
'Author: Ye Zichuan
' Center for Research in Water Resources
' The University of Texas at Austin
' ye@crwr.utexas.edu
'Revisions:
'1. 9/16/96 by Ferdi Hellweger
' - cosmetic changes
'---------------------------
'--- Purpose/Description ---
'---------------------------
'This program creates an arrow.
'-------------
'--- Input ---
'-------------
'1. From point.
'2. To point.
'3. Angle.
'4. Length.
'--------------
'--- Output ---
'--------------
'1. Arrow.
' -polygon graphic
'------------
'--- Code ---
'------------
fp = self.Get(0) ' from point
tp = self.Get(1) ' to point
theta = self.Get(2) ' angle
length = self.Get(3) ' length
x1 = tp.GetX
y1 = tp.GetY
x2 = fp.GetX
y2 = fp.GetY
xshift = length * (theta.cos)
yshift = length * (theta.sin)
totLength = (((y1-y2)^2)+((x1-x2)^2)).Sqrt
' point along the line
p1 = ((1.0 - (xshift/totLength)) * x1) + ((xshift/totLength) *x2)
p2 = ((1.0 - (xshift/totLength)) * y1) + ((xshift/totLength) *y2)
pi = Number.GetPI
104
if (x2<>x1) then
slope = (y2-y1)/(x2-x1)
slant = (slope/(((slope^2)+1).Sqrt)).ASin
else
slant = pi/2.0
end
a1 = p1 - (yshift * (slant.sin))
a2 = p2 + (yshift * (slant.cos))
b1 = p1 + (yshift * (slant.sin))
b2 = p2 - (yshift * (slant.cos))
lpnt = Point.Make(a1,a2)
rpnt = Point.Make(b1,b2)
aPolyLine = PolyLine.Make({{lpnt,tp},{tp,rpnt},{rpnt,lpnt}})
gr = GraphicShape.Make(aPolyLine)
return(gr)
'-----------
'--- End ---
'-----------
105
Appendix B: Data Dictionary
Unless otherwise noted, the data are in ArcView 3.2 shapefile format.
Data Description Class Attribute Units
*.kap NOAA nautical charts BSB images
addnoaa Avenue script that
works with noaamesh in
ArcView
Text file
Arctic.apr ArcView project with
extensions and data
produced in this
research
ArcView project
Arctic.mdb Benthic data relational
database
Microsoft Access
database
arrow.ave Avenue script that
works with findflux in
ArcView
Text file
azi2lamb Arc/Info projection file Text file
basemap Bathymetry / Elevation Grid Value Meters
basemap.avl ArcView legend for
basemap
Legend
Beaufort Benthic data sample
points in the Beaufort
Sea
Point Same as
samples.shp
Beaufort_prob30 Probability of benthic
biomass exceeding
30g/m2 in the Beaufort
Sea
ArcMap layer Probability Map
Beaufort_standerr Standard error contours
of the bio_beaufort
interpolation
ArcMap layer Prediction Standard
error
g/m
2
bio_beaufort Interpolated biomass
from beaufort.shp with
ordinary kriging
ArcMap layer Prediction g/m
2
bio_standerr Standard error contours
of the biomass
interpolation
ArcMap layer Prediction Standard
error
g/m
2
bio_prob50 Probability of benthic
biomass exceeding
50g/m
2
ArcMap layer Probability Map
106
Data Description Class Attribute Units
biomass Interpolated biomass
from samples.shp with
ordinary kriging
ArcMap layer Prediction g/m
2
buffer100km 100km buffer around
samples.shp
Polygon bufferdis Meters
chartreprojector Program that projects
the NOAA nautical
charts
EXE
chartview Chartviewer ArcView
extension for viewing
NOAA nautical charts
AVX, DLL
chukchi_prob250 Probability of benthic
biomass exceeding
50g/m2
ArcMap layer Probability Map
findflux Avenue script that
translates velocity
vectors to flux arrows,
works with flux_lamb
Text file
fivedeg Five degree boxes with
coordinates of their
lower left corner
Polygon lowerleft Degrees
flux_azi Ice-Ocean model,
average monthly ocean
surface velocity vectors
(1993) in original
projection
Point X_azi, Y_azi,
U*_1993,V*_1993
Meters,
cm/sec
flux_lamb Ice-Ocean model,
average monthly ocean
surface velocity vectors
(1993)
Point X_lamb, Y_lamb,
U*_1993,V*_1993
Meters,
cm/sec
frame Outlines of the study
area
Polygon
geo2lamb Arc/Info projection file Text file
Geostatistics ArcMap project, works
with a Geostatistical
Analyst extension
ArcMap project
h1_* Ice-Ocean model,
average monthly ocean
height (1993)
grid Value cm
107
Data Description Class Attribute Units
land Land coverage of study
area
Polygon
mergethemes Avenue script that
merges themes
Text file
noaamesh Outlines of NOAA
nautical charts
Polygon Chart
oc-buffer Ocean coverage of the
study area except for
100km buffer around
samples.shp
Polygon
ocean Ocean Coverage of the
study area
Polygon
pol2lamb Arc/Info projection file Text file
ponet Coastlines and political
boundaries
Polyline
rivers Major rivers Polyline Name System
salt.avl ArcView legend for
salt_*
Legend
salt_* Ice-Ocean model,
average monthly ocean
surface salinity (1993)
grid Value
samples Benthic data extracted
from Arctic.mdb
Point Station, Source,
Latitude, Longitude,
Date, Depth(m),
Temp(C), Sal(ppt),
Biomass(g/m
2
)
t.c6.93-*.cdf Ice-Ocean model,
original output
NetCDF
temp.avl ArcView legend for
temp_*
Legend
temp_* Ice-Ocean model,
average monthly ocean
surface temperature
(1993)
grid Value
?C
108
References
Arctic Systems Science (ARCSS). 2000. Ocean-Atmosphere-Ice Interaction
Internet site, http://arcss.colorado.edu/Projects/oaii.html
Broad,A.C.,H.Koch,D.T.Mason,G.M.Petrie,D.E.Schneider,andR.J.Taylor.
1978. Environmental assessment of selected habitats in the Beaufort and
Chukchi littoral system. Environmental Assessment of the Alaskan
Continental Shelf: Principal Investigators?. Boulder, CO.
Broad, A.C., A. Benedict, K.H. Dunton, H. Koch, D.T. Mason, D.E. Schneider,
and S.V. Schonberg. 1979. Environmental assessment of selected
habitats in the Beaufort and Chukchi littoral system. Environmental
Assessment of the Alaskan Continental Shelf: Principal Investigators?
Boulder, CO.
Broad, A.C., A. Benedict, K.H. Dunton, H. Koch, D.T. Mason, D.E. Schneider,
and S.V. Schonberg. 1980. Environmental assessment of selected
habitats in the Beaufort and Chukchi littoral system. Environmental
Assessment of the Alaskan Continental Shelf: Principal Investigators?
Boulder, CO: National Oceanic and Atmospheric Administration.
Broad,A.C.,M.Childers,K.H.Dunton,J.Hanes,H.Koch,D.E.Schneider,and
S.V. Schonberg, J. Zehr. 1981. Environmental assessment of selected
habitats in the Beaufort and Chukchi littoral system. Environmental
Assessment of the Alaskan Continental Shelf: Principal Investigators?
Reports for the ear Ending March 31, 1981. Boulder, CO.
Carey, A.G., M.A. Boudrias, J.C. Kern, R.E. Ruff. 1984. Selected Ecological
Studies on Continental Shelf Benthos and Sea Ice Fauna in the
Southwestern Beaufort Sea. Outer Continental Shelft Environmental
Assessment Program. Final Reports of Principal Investigators 23:1-164
Earth Resources Observation Systems (EROS) Data Center (EDC). 2000.
GTOPO30 Documentation Internet site, http://www1.gsi-mc.go.jp
/gtopo30/README.html#h1
Environmental Systems Research Institute (ESRI). 2000. Geostatistical Analyst
Tutorial.
109
Feder, HM, RH Day, SC Jewett, K McCumby, S McGee, SV Schonberg. 1982.
Infauna of the northeastern Bering and southeastern Chukchi Seas.Final
Report. Outer Continental Shelf Environmental Assessment Program
RU5. Fairbanks.
Grebmeier, J.M. 1987. The ecology of benthic carbon cycling in the northern
Bering and Chukchi Seas. Ph.D. dissertation, Inst. Mar. Sci., Univ. Alaska,
Fairbanks.
Grebmeier, J.M. and L.W. Cooper. 1995. Influence of the St. Lawrence Island
Polynya on the Bering Sea Benthos. J. Geophys. Res. 100: 4439-4460
Jakobsson,M, R.Macnab. 2000. International Bathymetric Chart of the Arctic
Ocean (IBCAO), Beta Version, Technical Reference and User?s Guide
Internet site, http://www.ngdc.noaa.gov/mgg/bathymetry/arctic/
IBCAO_TechnicalReference.PDF
Jakobsson,M.,N.Z.Cherkis,J.Woodward,R.Macnab,and B.Coakley. 2000. New
grid of Arctic bathymetry aids scientists and mapmakers;
Eos,Transactions,American Geophysical Union,v.81,no.9,p.89,93,96.
Jakobsson,M. 2000. IBCAO Beta release of Arctic Ocean bathymetry Internet
site, http://www.ngdc.noaa.gov/mgg/bathymetry/arctic/ibcaobetagrid.html
Maptech. 2000. NOAA Digital Charts Internet site, http://www.maptech.com
Maslowski, Wieslaw. 2000. ?Towards prediction of Arctic climate change?
Internet site, http://www.arcus.org/annual_meeting_00/fr_papers.html
National Science Foundation (NSF). 2000. Arctic Systems Science (ARCSS)
Internet site, http://www.nsf.gov/od/opp/arctic/system.htm
National Geophysical Data Center (NGDC). 1988. Data Announcement 88-
MGG-02, Digital relief of the Surface of the Earth. Boulder, Colorado
National Geophysical Data Center (NGDC). 2000. Information about the
ETOPO5 5-minute gridded elevation data Internet site,
http://www.ngdc.noaa.gov/mgg/global/etopo5.HTML
National Oceanic and Atmospheric Administration (NOAA). 2000, NOAA
Nautical Chart Extension for ArcView 3.x Internet site,
http://www.csc.noaa.gov/products/chartview/index.html
110
National Oceanographic Data Center (NODC). 1974-1979. Intertidal Organisms
and Habitats (File 030). National Oceanographic Data Center.
Washington, DC
Pennsylvania State University Libraries (PSU). 2000. About the Digital Chart of
the World Data Server Internet site, http://www.maproom.psu.edu/dcw/
Research Systems (RS). 2000. Noesys - The Easiest Way to Explore Data
Internet site, http://www.rsinc.com/noesys/index.cfm
Sandwell,DavidT,WalterH.F.Smith,StuartM.Smith,andChristopherSmall.
2000. Measured and Estimated Seafloor Topography Internet site,
http://topex.ucsd.edu/cgi-bin/get_data.cgi
Smith, W. H. F., and D. T. Sandwell. 1996. Global Bathymetric Prediction for
Ocean Modelling and Marine Geophysics Internet site,
http://topex.ucsd.edu/marine_topo/text/topo.html
Snyder, John P. 1987. Map Projections - A Working Manual. U.S. Geological
Survey Professional Paper 1395, United States Government Printing
Office, Washington
University Corporation for Atmospheric Research (UCAR). 2000. Software for
Manipulating or Displaying NetCDF Data, Unidata Internet site,
http://www.unidata.ucar.edu/packages/netcdf/utilities.html
Wahba, G. 1990. Spline models for Observational data. CBMS-NSF Regional
Conference Series in Applied Mathematics, Philadelphia: Soc. Ind. Appl.
Maths.
Wacasey, JW. 1974. Biological productivity of the southern Beaufort Sea:
Zoobenthic Studies. Beaufort Sea Technical Report #12b. Department of
the Environment, Victoria, B.C., 39 p.
Western Arctic Shelf-Basin Interactions (WASBI) Project. 2000. SBI Goals
Internet site, http://utk-biogw.bio.utk.edu/SBI.nsf
111
Vita
J?na Finnd?s J?nsd?ttir was born in Bl?ndu?s, Iceland on November 7,
1974 the daughter of ?rni J?nsson and Bj?rg Bjarnad?ttir. She is the fourth of
five daughters. After completing her work at Menntask?linn ? Akureyri in 1994,
J?na Finnd?s received a Bachelor?s of Geophysics degree from University of
Iceland, in 1998. During her studies at the University of Iceland she worked at
the National Energy Authority, Hydrological Service Division. In August 1998
she entered the Graduate school at the University of Texas at Austin and became a
graduate research assistant at the Center for Research in Water Resources.
Permanent address: S?lvabakki
541 Bl?ndu?s
Iceland
This report was typed by the author.