Journal of Archaeological Science 78 (2017) 40e56

Contents lists available at ScienceDirect

Journal of Archaeological Science

journal homepage: http://www.elsevier.com/locate/jas

Least cost path analysis of early maritime movement on the Pacific

Northwest Coast
Robert Gustas a, *, Kisha Supernant b
a
Archaeological Survey, Historic Resources Management Branch, Alberta Culture and Tourism, Old St. Stephen's College, 8820e112 Street, Edmonton, AB
T6G 2P8, Canada
b
Department of Anthropology, University of Alberta, 13-15 HM Tory, Edmonton, AB T6G 2H4, Canada

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, we present a new method for modeling past maritime movement events using least cost
Received 14 August 2016 path analysis. Nontraditional measures of movement cost, including cultural, environmental, and
Received in revised form physiological variables, were calculated. Using multiple cost-weighting scenarios, spatial resolutions, and
9 November 2016
different considerations of overland travel, movement routes were predicted for five Pacific Northwest
Accepted 12 November 2016
Coast study areas. This work uses a new application of least cost path analysis to seascapes and marine
movement and the results have led to a better understanding of migration during the Late Pleistocene
and Holocene. The resulting routes were systematically analyzed and compared to determine which
Keywords:
GIS
produced the results most likely to predict high-use coastal movement corridors. We found that
Least Cost Path analysis modeling scenarios where culturally derived costs of movement were highly weighted and in which
Northwest Coast overland travel was very costly produced the best predictions of possible past movement events. These
Seascape models show that predicted routes cluster in distinct patterns which are influenced by the geography of
Landscape analysis the seascape through which the movement event is taking place and that areas of high traffic are most
Predictive modeling likely to be located immediately offshore and to the south of islands as well as in the spaces between
Peopling of the New World landmasses. This knowledge increases our ability to predict the location of drowned sites on the
Northwest Coast and is important in contemporary archaeology because it can help locate new sites in a
landscape that has radically changed over the last 20,000 years. GIS analysis can reveal new sites hidden
by changing sea levels, which may not be easily located using traditional forms of site prospection.
Accurate modeling of maritime movement opens many coastal areas to increased archaeological
exploration and has the potential for the discovery of new sites in drowned locations.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction been paid to how people moved through marine environments

Archaeologists have extensively studied how past peoples
moved through and interacted with terrestrial landscapes (Llobera,
2011; Rockman and Steele, 2003; Surface-Evans and White, 2012;
Wheatley and Gillings, 2012:133). Much of this work has made use
of Geographic Information System (GIS) technology, which has
become the preeminent tool for modeling and displaying move-
ment and migration (Anderson and Gillam, 2000; Hu, 2011; Judge
and Sebastian, 1988; Llobera, 2001; Lock and Harris, 1992;
Verhagen, 2007; Wheatley and Gillings, 2012). Less attention has
* Corresponding author.
E-mail addresses: gustas@ualberta.ca (R. Gustas), kisha.supernant@ualberta.ca
(K. Supernant).
(Howey, 2007; Leidwanger, 2013), despite the fact that maritime
technology has been used extensively by ancient humans across the
world (Ames, 2002; Bjerck, 2008; Montenegro et al., 2006; Van de
Noort, 2003). Least Cost Path analysis (LCP) in the past has been
applied to overland travel along lakeshores, on rivers, and through
ocean coastal environments (Bell and Lock, 2000; Carballo and
Pluckhahn, 2007; Hare, 2004; Sakaguchi et al., 2010; Sherman
et al., 2010; Siart et al., 2008; Taliaferro et al., 2010). Unlike
Leidwanger's (2013) previous work, which estimated vessel travel
speed, wind direction, and sailing conditions to predict how far a
Classical Mediterranean vessel could travel in a given time, we use a
multivariate weighted methodology to model areas of high travel
from discrete coastal points and predict possible new site locations.
Here we present a set of initial findings indicating that LCP analysis
can successfully yield new and insightful information on past

http://dx.doi.org/10.1016/j.jas.2016.11.006
0305-4403/© 2016 Elsevier Ltd. All rights reserved.
 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56 41

Fig. 1. Study areas on the Northwest Coast.

maritime movement events.
We applied our maritime LCP analysis to the question of the
timing and method of the peopling of the New World. For much of
the twentieth century, the archaeological community thought that
it understood the process and timing of initial human migrations
into the New World. These explanations took the form of the Clovis
First and Ice-Free Corridor theories (Adovasio and Pedler, 2013:512;
Arnold, 2002:437; Beck and Jones, 2013:273; Mandryk et al.,
2001:301). These hypotheses stated that as the Cordilleran and
Laurentide ice sheets retracted, a narrow, ecologically viable
42 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56

Table 1
Project spatial data sources.

Dataset Source Citation

Canadian Elevation DataBC Terrain Resource Management Program Province of British Columbia, 1997
Canadian Bathymetry CHS Hydrography Data Center Canadian Hydrographic Service, 2014
Alaskan Elevation USGS National Elevation Dataset U.S. Geological Survey, 2014
Alaskan Bathymetry NOAA National Geophysical Data Center National Ocean and Atmospheric Administration, 2014
Alaska Hydrography Alaska Department of Natural Resources Alaska Department of Natural Resources (2007)
Canadian Hydrography Natural Resources Canada e CanVecþ Natural Resources Canada, 2013
Prince Rupert Harbour LiDAR Prince Rupert Harbour Archaeology Project Nexen Energy ULC, 2014

Table 2
Sea level change by project area.
Area 10,000 cal yr BP 13,000 cal yr BP 16,000 cal yr BP
Prince Rupert Harbour þ3 m þ20 m þ50 m
Stephens Island þ9 m þ12 m þ14 m
Dundas Archipelago þ9 m þ12 m þ14 m
Haida Gwaii þ15 m 140 m 150 m
Alexander Archipelago þ10 m 125 m 150 m
corridor into North America was made available, through which
ancient peoples were able to travel from Siberia through Alaska and
Canada into the continental interior of the Americas (Dixon,
1999:30; Ives et al., 2013:150).
Over the last 40 years, the academic supremacy of the Clovis
First theory has gradually eroded. Paleoecological research has
determined that the Corridor could have physically opened as early
as 13,500 cal yr BP, and yet sites have been discovered that are older
than this, most notably in Chile, Pennsylvania, and Oregon, among
other locations (Adovasio and Pedler, 2013:513; Dillehay, 1989;
Jenkins et al., 2013). The discovery of these sites has caused ar-
chaeologists to look for alternative explanations for initial peopling
events, including a possible route along the Pacific Northwest Coast
(Erlandson , 2013; Fladmark, 1979; Pedersen et al., 2016). Better site
location modeling using LCP analysis should allow for the discovery
of sites from Paleoamerican1 coastal migration events.
The overarching question we ask is can LCP analysis be applied
to modeling early human maritime movement events? We use the
northern Northwest Coast during the Late Pleistocene and Early
Holocene as a case study. Within the umbrella of our larger research
question, we examine several more focused issues, including the
role of weighting scenarios in projecting movement routes and the
appropriate resolution for analysis. Additionally, we explore how
allowing or disallowing portions of overland travel impacts coastal
movement paths. We conclude by noting the limitations of the
study and suggesting future directions and refinements of the
model.
2. Background
Five areas of the northern Northwest Coast formed our case
study, including the northern Alexander Archipelago, the southern
part of Haida Gwaii, the Dundas Islands, Stephens Island, and Prince
Rupert Harbour (Fig. 1). Highly accurate elevation and bathymetric
data was available for all of these areas, as well as reasonably
precise sea level curves, all of which are essential elements for
conducting LCP analysis in this part of the world (Shugar et al.,
2014). Table 1 details all datasets and data sources for the project.
The extensive spatial coverage and high quality of this data allowed
1
Here we use the term Paleoamerican in the same context as it is used in
Paleoamerican Odyssey (2013) edited by Graf, Ketron, and Waters.
us to conduct geospatial analyses for time periods dating to the Late
Pleistocene.
Our study area has been home to the Coast Tsimshian, Haida,
and Tlingit peoples for thousands of years (Blackman, 1990; de
Laguna, 1990; Halpin and Seguin, 1990; Moss, 2011). These
hunter-gatherer-fisher cultures are most likely descended from
Paleoamerican peoples and provide valuable insight into early
lifeways (Brace et al., 2014:463; Chatters et al., 2014:753;
Rasmussen et al., 2014:225; Rasmussen et al., 2015:3). Throughout
the history of these peoples, boats have played a central role in their
cultures and were used for transportation and subsistence, among
other activities (Blackman, 1990; de Laguna, 1990; Halpin and
Seguin, 1990). Using evidence from the archaeological record,
ethnographic materials, and oral histories, we can derive elements
of past lifeways that inform our modeling of early maritime
movement. Factors that influenced site selection include the choice
of semi-protected harbours with high sinuosity, the presence of a
flat beach where boats could be drawn onto the shore, and the need
to keep the shoreline in sight when entering an unknown area
(Blackman, 1990:240; Cookson, 2013:100; de Laguna, 1990:206;
Mackie and Sumpter, 2005:350e351). These behaviors and char-
acteristics are important because they may be analogues to ancient
ways of life and provide variables which we use to inform how we
calculate movement costs in our analysis.
The Northwest Coast is a spatially and temporally dynamic
environment that has changed significantly over the last 20,000
years due to sea level changes caused by shifts in global climate
(Shugar et al., 2014:6). Sea levels change is spatially discontinuous,
as levels may rapidly rise at one location only to drastically plunge
at another (Shugar et al., 2014). To account for local variation, we
ran our analysis at 3000 year intervals between 10,000 and
16,000 cal yr BP in each region. Table 2 summarizes the sea level
changes in our study areas by time period.
3. Theory of maritime least cost path (LCP) analysis
LCP modeling is a geospatial analysis technique closely related
to the more traditional site prospection modeling methods often
used in archaeology. The central assumptions of LCP analysis are
that areas which are easy to access have a higher chance of con-
taining evidence of human activity than areas that are difficult or
undesirable to access and that humans will move from point A to
point B over the path with the lowest accumulated cost (Surface-
Evans and White, 2012:2; Wheatley and Gillings, 2012:142).
Movement cost values are calculated for all cells in a raster dataset
and produce a friction surface or cost surface layer (Mitchell,
2012:225). Least cost paths are generated by placing origin and
destination points of a movement event on top of the friction sur-
face and calculating the path of least resistance between these
points (Mitchell, 2012:214).
Most previous applications of LCP analysis to archaeological
questions have focused on movement through terrestrial environ-
ments (Anderson and Gillam, 2000; White and Barber, 2012). These
 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56 43

Fig. 2. Comparison of path by cell resolution.

studies have largely focused on movement costs that incorporate

Table 3 slope, Tobbler's Backpacking Formula, and other land-based, envi-
Movement cost reclassification values.
ronmentally derived factors. Ocean waters are often seen as a
Category Variable Movement Movement Movement barrier to travel rather than pathways between locations (Surface-
cost 1 cost 2 cost 3
Evans and White, 2012:136; Livingood, 2012:178). However,
Environmental Beach Slope 2% >2% NA
Environmental Beach Aspect 1 e180 181 e360 NA
Environmental Freshwater 200 m >200 m NA
Proximity Table 4
Physiological Travel Distance 5760 m >5760 m NA Weighting scenario influence breakdown.
Cultural Sinuosity (>0.5) 1000 m 1000 m-2000 >2000 m
Variable Physiological Cultural Environmental Unweighted
m
a scenario scenario scenario scenario
Cultural Visibility Most Visible 2nd Most 3rd Most
Visible Visible Beach Slope 0.125 0.08 0.25 0.2
Cultural Protected Waters Highly Medium Least Beach Aspect 0.125 0.08 0.25 0.2
Protected Protected Protected Freshwater 0.125 0.09 0.25 0.2
Cultural Inland Waters 181 e360 0 e180 NA Proximity
a Maximum 0.5 0.25 0.125 0.2
Break points were determined by Jenk's Method of classification. Break values
Distance
were not recorded due to the iterative nature of the work flow which overwrote
Near Shore 0.125 0.5 0.125 0.2
these values each time the model was run. Visibility reclassification never exceeded
Corridor
five classes.
44 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56

Fig. 3. PRH average path orientation through time. Darker coloured arrows represent earlier time periods.

because maritime transportation is a major method of travel on the
Northwest Coast, existing methodologies must be adapted to suc-
cessfully model movement in this environment (Mackie et al.,
2011:79).
Applying LCP analysis to maritime environments requires a
substantial shift of perspective from traditional terrestrial LCP
analysis. When working in a maritime context, it becomes impor-
tant to treat the study areas as a seascape with three different
categories of movement cost: time, distance, and energetic
expenditure (Mitchell, 2012:177; Surface-Evans and White,
2012:4). When using traditional movement cost metrics, all
points in a seascape are identical, i.e. flat, since ocean waters cannot
have slope. Cost must therefore be calculated in different ways,
such as a measure of the distance from a terrestrial resource,
alternative measures of physical ease of access, or using the cultural
logic that people use to choose where they travel. Capturing the
cultural agency of maritime movement is a key component of
determining the easiest travel routes through a seascape (Howey,
2007:1831e1832; Leidwanger, 2013).
 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56
Fig. 4. Haida Gwaii line density results - 10,000 cal yr BP.
4. Methods
We gathered elevation and bathymetric point data and merged
these measurements into one combined point cloud for each in-
dividual study area. This information was then used to generate
new Digital Elevation Models (DEM) using the Inverse Distance
Weighted (IDW) surface interpolation technique using ESRI ArcGIS
10.3 and ArcGIS Pro 1.1 (Monteleone, 2013:111). These surfaces
represent the landscape as it would exist if completely drained of
water. The DEMs were then flooded to represent the appropriate
sea level for the time period in question and passed through a map
algebra equation to reset all cells relative to the sea level height
being modeled. The shoreline and the study area perimeter were
used to define the boundaries of the landmass for each time period,
from which a terrestrial and marine dataset were generated for
analysis.
DEMs were calculated at the 5 m, 10 m, and 30 m resolutions for
Prince Rupert Harbour to test what the appropriate cell size was for
45
calculating maritime LCPs. We discovered that the results gener-
ated from the DEMs with 5 m and 30 m cells had significant path
overlap, while those generated from 10 m cells were substantially
different. The 5 m and 30 m DEMs produced results that were
better predictions of maritime movement as they more closely
followed the coastline and avoided moving through terrestrial cells
(Fig. 2). The 30 m DEM analysis was easily calculated using a
standard desktop computer and we therefore used the 30 m DEMs
for the rest of our analysis.
We used shoreline slope, aspect, and proximity to fresh water
sources to determine the environmental cost of movement
(Table 3). We identified beaches that had very low slopes (<2%) and
assigned them low movement costs because they are the most
suitable for landing a boat. All other areas were assigned a high
movement cost because the beaching and storage of boats would
have been very difficult at these locations (Blackman, 1990:240).
We assigned low cost to parts of the coastline that had an east
facing aspect (1 e180 ), representing the preferred aspect for site
46 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56
Fig. 5. Unweighted and culturally weighted paths - Stephens Island 16,000 cal yr BP.
selection. East-facing coastlines within our study areas were sub-
ject to milder weather and wave action, making them more desir-
able for habitation (Mackie and Sumpter, 2005:350). Lastly we
assigned low cost values to locations within 200 m of contempo-
rary freshwater sources (following Maschner, 1995:181) as recor-
ded in CanVecþ and Alaskan Department of Natural Resource
hydrology data. Fresh water is a necessity for human occupancy of
any length, and across our study areas the 200 m threshold pro-
vided a reasonable number of sources that would have been easily
accessible for travelers. For locations that were formerly exposed
but which are now inundated by ocean water, hydrology modeling
was undertaken to locate possible past watercourses (ESRI, 2015).
The Euclidean distance was then calculated from water features
and reclassified into two classes using the same 200 m threshold.
Areas with values under the threshold were assigned low move-
ment cost values, while those with values greater were given high
movement costs. Table 3 shows all the criteria used for movement
cost reclassification.
The physiological cost of moving by boat was determined by
looking at the maximum distance that can be travelled by boat in a
12 h period. Experimental archaeology using canoes to travel from
Skokomish, Washington to Dungeness at the eastern end of the
Straits of Juan de Fuca has determined that the maximum average
distance that could have been traveled by a fully loaded canoe is
57.6 km/day based on 12 h of paddling at a speed of 4.5 km/h
(Ames, 2002:30e31). In our preliminary analysis, using the 57.6 km
travel distance gave every cell in the smallest study areas the same
value, so after a process of trial and error, we used 10% of the
maximum distance value, or 5760 m, as our reclassification
threshold. This threshold value identified significant areas of low
movement cost without oversaturating the study area and making
movement unrealistically easy to pass through. Areas within
5760 m of a landmass were assigned a low movement cost and
areas outside this range a high movement cost.
Several variables were used to represent possible cultural
movement logics used by Paleoamerican coastal mariners. Early
travellers would most likely have tried to keep the shoreline in
sight when entering a new environment as scouting for suitable
landing stopping points. This would have allowed for waypoints
and other visual indicators of location to be established for future or
return migration events. Using a viewshed analysis of Prince Rupert
Harbour, which is representative of a typical Northwest Coast
environment, we determined that for individuals sitting in a boat at
distances greater than five kilometres from the shore, visual iden-
tification of important shore features became very difficult. Cu-
mulative visibility was run at 1000 m intervals from all shorelines.
Areas with low visibility were given high movement costs and vice
versa.
The next input for cultural movement cost was the location of
protected waters, which were identified as those areas that fell on
the inland side of islands, peninsulas, and other landmasses
(Monteleone, 2013: 69). These locations would be sheltered from
the weather of the open ocean and would be easier to travel
through, making them preferable even if passage through them
lengthened the total distance of a trip. Protected waters tended to
be eastward facing on the Northwest Coast and were identified by
calculating Euclidean distance and direction layers using landmass
polygons as inputs (Cookson, 2013:48). The direction layer was
reclassified into two classes: high and low, with low-value cells
being those facing between 0 and 180 . All other cells were
assigned high movement cost values.
Locations inside straits and passages were also favoured travel
routes due to their similarly protected nature. Using a visual anal-
ysis of our study area, we determined that no strait was wider than
2000 m and assigned areas within that distance of the coast low
movement costs.2 This was done by calculating Euclidean distance
from landmasses and then reclassifying these results into two
2
This technique only identifies areas that cannot possibly lie inside a strait and
we are attempting to come up with a better method of determining the location of
these geographic features.
 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56 47

Fig. 6. Clustered Grid Pattern e Dundas Islands 13,000 cal yr BP.

classes, with low movement costs assigned to cells within 2000 m
of the shoreline. The inland waters and straits/passage movement
cost values were then summed together for every cell. The resulting
values were classified into areas of low, medium, and high pro-
tection and assigned decreasing movement cost values.
Lastly we calculated the sinuosity of the coastline, a variable that
correlates with biodiversity and habitation suitability. Subsistence
resources, notably shellfish and other marine foodstuffs, are more
plentiful in these locations (Mackie and Sumpter, 2005:350e351;
Monteleone, 2013:15). Sinuosity is an index that measures the
cragginess of a coastline, with a value of 0 representing a perfectly
straight line and 1 an infinitely complex line. Coastlines were
divided into 1 km segments by placing points on the modeled
shorelines at 1000 m intervals. These points were used to cut lines
into smaller segments for which sinuosity was determined. Seg-
ments with values over 0.5 were selected and Euclidean distance
calculated using these lines as the input. These distance results
were then reclassified into three classes using the same approach as
the protected waters.
Once all the input movement cost layers were generated for a
study area these datasets were summed together, creating friction
surfaces. We used weighted summation to generate three friction
surfaces for each study area, each emphasising a different category
of cost values (Table 4). In these scenarios, half of the weighting was
equally split between each of the variables in the category being
emphasized with the rest of the weighting split equally between
the variables from the other two categories. Additionally, an un-
weighted surface was created in which all the input variables
48 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56

Fig. 7. Parallel to Perimeter Path Pattern e Prince Rupert Harbour 13,000 cal yr BP.

carried equal influence. This produced four friction surfaces of a
study area for each time period. Origin and destination points were
placed at twenty-degree intervals around the perimeter of the
study area to model entry and movement through a study area in
any direction.
For each origin point, cost distance and backlink layers were
calculated, which were used to create least cost paths to the other
points along the study areas perimeter. After repeating this process
for all study areas, a total of 1080 least cost paths were calculated.
A second analysis was run for the three geographically smallest
study areas, the Dundas Islands, Stephens Island, and Prince Rupert
Harbour. For these areas, landmasses were given an extremely high
movement cost value in order to force movement paths into marine
portions of study areas. The reasons that overland travel was not
disallowed in the first set of analyses was that it is possible to
portage a lightweight boat a considerable distance and we did not
want to eliminate this form of movement from inclusion in our
models (Moss, 2008:38: Stewart, 1984:52). Haida Gwaii and the
Alexander Archipelago were not included in the re-analysis
because these environments are composed predominantly of
terrestrial cells at and before 13,000 cal yr BP.
The results from all analyses were run through an average path
orientation analysis to determine the average cardinal orientation
of all the segments from specific weighting criteria from each
scenario. This allowed us to determine the average axis of move-
ment for each set of paths (Fig. 3) and test if people were in fact
moving through the environment in the expected north-to-south
direction.
All the paths from a given weighting scenario were also passed
through a linear density tool to determine the number of times
 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56 49

Fig. 8. East to West Average Directional Mean e Alexander Archipelago 16,000 cal yr BP.

Table 5
Results summary by location e overland travel allowed.

Variable Prince Rupert Harbour Stephens Island Dundas Island Haida Gwaii Alexander Archipelago

Greatest Proximity to Coastline Environmental Unweighted/Cultural Unweighted Cultural Environmental/Cultural

Greatest Amount of Overland Travel Physiological Cultural/Physiological Environmental Environmental Environmental/Physiological
Clustering Pattern Parallel to Perimeter Clustered (Western) Grid Clustered (Western) Grid Clustered Grid Clustered Grid
Most Common Mean Direction East Northeast Northeast East Northeast

paths passed through each cell in a study area. This information was results were then reclassified into areas with high, medium, and
used to generate a heat map showing what areas were most heavily low site potential.
trafficked (Fig. 4). Areas which were heavily traveled should have
the highest probability of containing undiscovered sites. These
50 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56
Fig. 9. Unweighted and Culturally Weighted Paths, Overland Movement Prohibited - Stephens Island 16,000 cal yr.
5. Results
Our discussion will examine overall trends and patterns in the
data, evaluate several different qualities of the produced least cost
paths, and explore the average orientation and line density values
from each scenario. In order to compare results from different
weighting scenarios, we looked at the amount of overland travel,
proximity to the coastline, and clustering patterns. Early mariners
would likely have followed the coastline when traveling, especially
when moving into unexplored areas. Additionally, they would have
minimized their overland travel, only portaging boats or leaving the
ocean when absolutely necessary. Looking at path clustering pat-
terns is important because this information sheds light on where
human activity was taking place and complements line density and
directional mean measurements.
In the movement scenarios that allowed overland travel, we see
that the cultural and unweighted scenarios produced the paths
with the closest proximity to coastlines regardless of study area.
Both the physiologically and environmentally weighted scenarios
produced paths which had the largest amounts of overland travel.
This indicates that the culturally weighted and unweighted paths
are the most suitable for predicting routes of travel over water
(Fig. 5). All study areas except Prince Rupert Harbour produced
clustered grid-like movement path patterns, where paths intersect
at right angles and congregate in the western or central portions of
study areas (Fig. 6). In Prince Rupert Harbour the paths move along
the perimeter of the study area, avoiding travel through the center
(Fig. 7). The average path orientation in all of these scenarios is
along the east-to-west axis as demonstrated in Fig. 8. Table 5
provides a full breakdown of results by study area.
In the modeling scenario which made overland travel prohibi-
tively expensive, similar patterns and trends were revealed. The
environmentally and culturally weighted scenarios had the closest
proximity to coastlines (Fig. 9). Overland travel was not a factor due
to high movement costs assigned to terrestrial cells. Paths clustered
into grid patterns and the east-to-west axis path orientation was
the most common axis of movement (Figs. 10 & 11). Full results by
study area can be seen in Table 6.
These results show that across location, time, and type of
movement event, the culturally weighted and unweighted sce-
narios generally produced the best results for modeling maritime
movement events. The environmental and physiologically
weighted scenarios produced results that were less predictive of
coastline travel than the unweighted scenarios. This pattern em-
phasizes the importance of including cultural agency in the
modeling of maritime movement.
One of the most striking results to come out of this analysis was
the trends and patterns associated with different types of land-
scapes (Fig. 12). Stephens Island and the Dundas Islands performed
much differently than Haida Gwaii, the Alexander Archipelago, and
Prince Rupert Harbour, each of which had a higher proportion of
terrestrial cells. The two island seascapes show higher amounts of
path clustering, a more northerly movement orientation, and
greater unweighted path suitability. In these locations, there seems
to be a slight trend for paths to cluster along the western side of
landmasses and for decreased overland travel. Paths through study
areas that include significant landmasses are characterized by less
clustering, greater suitability of cultural paths, and increased
overland travel. These two different types of geographies produce
very different results that are controlled by sea level history, the
locations of landmasses, and the percentage of the study areas
composed of marine environments.
The different clustering trends observed at different study areas
make sense within the framework of how early explorer groups
may have navigated the Northwest Coast (Anderson and Gillam,
2000:56; Beaton, 1991:223). It is expected that these groups
would move in small bands fluidly across diverse landscapes,
making use of a wide variety of resource patches. The clustering of
paths in different study areas suggests that the input factors in our
analysis form groupings of low movement cost cells that Paleo-
american travelers may have been using in similar ways as resource
patches. Early travelers could have glided from patch to patch,
 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56 51

Fig. 10. Cultural Paths - Dundas Islands 10,000 cal yr BP.

using them as waypoints in longer journeys. The clustering of
routes in specific areas of very low movement cost is to be expected
and is encouraging for the successful application of maritime least
cost analysis to site prospection.
Maritime least cost analysis results are highly influenced by
whether they were computed for areas that are predominantly
terrestrial landscapes bordering bodies of water or are mainly
marine seascapes that include small landmasses. The former sce-
nario is predicated largely on travel by walking with short maritime
segments, while the latter assumes the majority of the distance that
is traveled will be covered by boat with short overland segments.
The very high movement cost results produced for Prince Rupert
Harbour, the Dundas Islands, and Stephens Island closely fit the
expected patterns for Paleoamerican travel. In these locations,
culturally based limitational knowledge was important in path
selection and routes were highly clustered with path directional
mean remaining to the east and northeast (Rockman and Steele,
2003:4). Overall, this modeling scenario is a better fit for predict-
ing movement through mostly marine landscapes. High overland
movement cost maritime LCP modeling should be considered to
predict travel through these locations when the researcher is only
interested in the maritime segments or there is a significant body of
water present. The methodology that eliminates overland travel
should be applied to landscapes where landmasses are mostly
contained inside the analysis extent and have large marine areas. In
areas where marine travel may have been one segment of a larger
overland journey, modeling scenarios should be used in which
overland travel is not prohibitively expensive.
52 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56

Fig. 11. East to West Average Directional Mean, Overland Movement Prohibited e PRH 10,000 cal yr BP.

Table 6
Results summary by location e overland disallowed.

Variable Prince Rupert Harbour Stephens Island Dundas Island

Greatest Proximity to Coastline Cultural Environmental Environmental/Cultural

Clustering Pattern Clustered (Various) Grid Dispersed Grid Clustered (Southwest) Grid
Most Common Mean Direction East East Northeast

Our line density models suggest that sites are likely to be located
in marine environments that are immediately offshore of large
landmasses or currently located in the open ocean space between
smaller islands (Fig. 13). Sites should also be prevalently located in
the southern half of study areas with the density of sites decreasing
dramatically moving northward. The falloff of line density is more
gradual to the east and west but the amount of lines moving
through these areas is still much smaller than in the southern
portions of study areas. In the modeling scenarios that allow
overland movement, possible site locations tend to cluster
together; however, this trend is not true of the scenario that pro-
hibits overland travel. The line density results from these models
show a more random distribution of high probability areas that
frequently do not border one another. The position of high
 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56 53

Fig. 12. Comparison of marine and terrestrial seascapes.

probability site locations shows that the movement corridor routes
fitting the patterns described in the previously discussed results
were traveled much more heavily than others. These areas almost
always stretch from the interior of the study area to a point on the
boundary.
6. Conclusion
In conclusion, we found that LCP analysis can be used to model
maritime movement events, specifically those that occurred on the
Northwest Coast over the last 16,000 years. Disallowing overland
travel had the effect of reducing the overland segments of predicted
routes, better modeling maritime migrations. Using a 30 m cell size
maintained the quality of results while reducing the amount of
time needed to run the analysis, providing an appropriate balance
between accuracy and computational requirements. Scenarios
which emphasise the culturally determined costs of movement
produced the best results, while scenarios which emphasised
traditional measures of movement cost, such as physiological and
environmental variables, provided less accurate models of possible
movement events. These findings emphasize how important it to
look at maritime movement within the context of seascape envi-
ronments; the importance of cultural aspects of maritime move-
ment cannot be overlooked and their inclusion greatly increases
the power of this technique.
This paper demonstrates that LCP analysis can be applied to
maritime movement in the deep past and has the potential to
significantly change the way in which we search for new sites along
54 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56
Fig. 13. Comparison of Reclassified Path Density - Dundas Islands 10,000 cal yr BP.
temporally dynamic coastlines. Within the context of the North-
west Coast, the application of maritime Least Cost Path analysis
could lead to the discovery of new very old sites, which could shed
light on specific aspects of the Peopling of the New World. Poten-
tially these discoveries could help to answer questions about the
exact timing of these events within the Late Pleistocene. Addi-
tionally, new early sites could indicate whether a coastal route was
used and what its path may have been.
This work is still in its infancy and is by no means complete.
There are tools, such as regression analysis and geospatial statistics,
which could be used to determine how strongly different inputs
influence the results of the analysis. Comparison between known
site locations and projected paths from more recent time periods
could be created to determine how well paths predict these loca-
tions (Gustas, 2016). Once models have reached an acceptable level
of predictiveness, ground truthing should be conducted to see if
this technique can reliably be used to find new sites.
The inclusion of several types of geographic features including
ocean currents, rocky headlands, and glacial areas through time
could also greatly influence the results that are produced and
should be considered. Additionally, if suitable re-creations of the
locations of past ecological resources such as kelp beds could be
produced, these features could be incorporated into the analysis
(Erlandson et al., 2007). The process that was used to calculate the
location of past streams in drowned landscapes could be improved
using advanced hydrological analysis. As more accurate bathymetry
of the intertidal and LiDAR of terrestrial areas are produced, this
information can be incorporated in our work to increase the ac-
curacy of the predicted paths. Lastly, we would like to run this
analysis on the entire Northwest Coast from Cape Mendocino in
California to the northern edge of the Alexander Archipelago in
Alaska.
Acknowledgments
Data used in the project was provided by the U.S. National
Oceanic and Atmospheric Administration, the Canadian Hydro-
graphic Service, and Nexen. We would like to thank Kenneth Ames,
Andrew Martindale, Bryn Letham, and Thomas Brown for their
assistance in preparing this project. Additionally, Jack Ives and
Robert Losey provided invaluable feedback which shaped much of
the most recent work on this project. We appreciate the thoughtful
reviews from our two anonymous reviewers of an earlier draft of
the manuscript. Aaron Coons must also be mentioned for his
assistance in editing and preparing this article. Any errors or
omissions are our own.
References
Adovasio, J.M., Pedler, D.R., 2013. The ones that still Won't go away: more biased
thoughts on the pre-clovis peopling of the new world. In: Graf, Kelly E.,
Ketron, Caroline V., Waters, Michael (Eds.), Paleoamerican Odyssey. Center for
the Study of the First Americans, College Station, pp. 511e520.
Ames, Kenneth M., 2002. Going by boat: the forager-collector continuum at sea. In:
Fitzhugh, Ben, Habu, Junko (Eds.), Beyond Foraging and Collecting Evolutionary
Change in Hunter-Gatherer Settlement Systems. Kluwer Academic/Plenum
Publisher, New York, pp. 19e52. Fundamental Issues in Archaeology.
Anderson, David G., Gillam, J.C., 2000. Paleoindian colonization of the Americas:
implications from an examination of physiography, demography, and artifact
distribution. Am. Antiq. 65 (1), 43e66.
Arnold, Thomas G., 2002. Radiocarbon dates from the ice-free corridor. Radiocarbon
44 (2), 437e454.
Beaton, John M., 1991. Colonizing continents: some problems from Australia and the
Americas. In: Tom, D. Dillehay, Meltzer, David J. (Eds.), The First Americans:
Search and Research. CRC Press, Boca Raton, pp. 209e231.
Beck, Charlotte, Jones, George T., 2013. Complexities of the colonization process: a
view from the north american west. In: Graf, Kelly E., Ketron, Caroline V.,
Waters, Michael (Eds.), Paleoamerican Odyssey. Center for the Study of the First
Americans, College Station, pp. 273e291.
Bell, Tyler, Lock, Gary, 2000. Topographic and cultural influences on walking the
ridgway in later prehistoric times. In: Lock, Gary (Ed.), Theorising the Practice
or Practicing the Theory: Archaeology and GIS. IOS Press, Washington DC,
pp. 82e100.
Bjerck, Hein B., 2008. Colonizing seascapes: comparative perspectives on the
development of maritime relations in scandinavia and patagonia. Arct.
 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56
Anthropol. 46 (1e2), 118e131.
Blackman, Margaret B., 1990. Haida: traditional culture. In: Shuttles, Wayne,
Sturtevant, William C. (Eds.), Handbook of North American Indians, 7. Smith-
sonian Institute Press, Washington, pp. 240e260. Northwest Coast.
Brace, C. Loring, Seguchi, Noriko, Russell Nelson, A., Qifeng, Pan, Umeda, Hideyuki,
Wilson, Margaret, Brace, Mary L., 2014. The ainu and jo mon connection. In:
Jantz, Richard L., Owsley, Douglas W. (Eds.), Kennewick Man: the Scientific
Investigation of an Ancient American Skeleton. Texas A&M University Press,
College Station, pp. 463e471.
Carballo, David M., Pluckhahn, Thomas, 2007. Transportation corridors and politcal
evolution in highland mesoamerica: settlement analysis incorporating GIS for
northern tlaxcala, Mexico. J. Anthropol. Archaeol. 26, 607e629.
Chatters, James C., Kennett, Douglas J., Asmerom, Yemane, Kemp, Brian M.,
Polyak, Victor, Blank, Alberto Nava, Beddows, Patricia A., Reinhardt, Eduard,
Arroyo-Cabrales, Joaquin, Bolnick, Deborah A., Malhi, Ripan S.,
Culleton, Brendan J., Erreguerena, Pilar Luna, Rissolo, Dominique, Morell-
Hart, Shanti, Stafford Jr., Thomas W., 2014. Late Pleistocene human skeleton and
mtDNA link paleoamericans and modern native americans. Science 344 (6185),
750e754.
Cookson, Corey, 2013. An Analysis of Site Selection Behaviors and Landscape Use in
the Prince Rupert Harbour Area. University of Alberta, Edmonton. Master’s
Thesis, Department of Anthropology.
de Laguna, Frederica, 1990. Tlingit. In: Shuttles, Wayne, Sturtevant, William C.
(Eds.), Handbook of North American Indians, 7. Smithsonian Institute Press,
Washington, pp. 203e228. Northwest Coast.
Dillehay, Tom D., 1989. Monte Verde: a Late Pleistocene Settlement in Chile, vol. 1.
Smithsonian Institution Press, Washington, DC.
Dixon, E.J., 1999. Bones, Boats & Bison: Archeology and the First Colonization of
Western North America. University of New Mexico, Albuquerque.
Erlandson, Jon M., 2013. After clovis-first collapsed: reimagining the peopling of the
Americas. In: Graf, Kelly E., Ketron, Caroline V., Waters, Michael (Eds.), Paleo-
american Odyssey. Center for the Study of the First Americans, College Station,
pp. 127e133.
Erlandson, Jon, M., Graham, Michael H., Bourque, Bruce J., Debra, Corbett,
Estes, James A., Steneck, Robert S., 2007. The Kelp highway hypothesis: marine
ecology, the coastal migration theory, and the peopling of the Americas. J. Isl.
Coast. Archaeol. 2 (2), 161e174.
ESRI, 2015. An Overview of the Hydrology Toolset accessed April 23, 2015. http://
desktop.arcgis.com/en/desktop/latest/tools/spatial-analyst-toolbox/an-
overview-of-the-hydrology-tools.htm.
Fladmark, K.R., 1979. Routes: alternate migration corridors for early man in north
America. Am. Antiq. 44 (1), 55e69.
Gustas, Robert, 2016. Least cost analysis of movement events during the Holocene
on the Northwest coast. In: Paper Presented at the 67th Annual Meeting of the
Society for American Archaeology (Orlando).
Halpin, Marjorie M., Seguin, Margaret, 1990. Tsimshian peoples: southern tsim-
shian, coast tsimshian, nishga, and gitksan. In: Shuttles, Wayne,
Sturtevant, William C. (Eds.), Handbook of North American Indians, 7. Smith-
sonian Institute Press, Washington, pp. 267e284. Northwest Coast.
Hare, Timothy S., 2004. Using measures of cost distance in the estimation of polity
boundaries in the postclassic yautepec valley, Mexico. J. Archaeol. Sci. 31,
799e812.
Howey, Meghan C.L., 2007. Using multi-criteria cost surface analysis to explore past
regional landscapes: a case study of ritual activity and social interaction in
Michigan, AD 1200-1600. J. Archaeol. Sci. 34, 1830e1846.
Hu, Di, 2011. 2011 advancing Theory? Landscape archaeology and geographic in-
formation systems. Pap. Inst. Archaeol. 21, 1e14.
Ives, John W., Frose, Duane, Supernant, Kisha, Yanicki, Gabriel, 2013. Vectors, ves-
tiges, and valhallas e rethinking the corridor. In: Graf, Kelly E., Ketron, Caroline
V., Waters, Michael (Eds.), Paleoamerican Odyssey. Center for the Study of the
First Americans, College Station, pp. 149e170.
Jenkins, Dennis L., Davis, Loren G., Stafford Jr., Thomas W., Campos, Paula F.,
Connolly, Thomas J., Scott Cummings, Linda, Hofreiter, Michael, Hockett, Bryan,
McDonough, Katelyn, Luthe, Ian, O'Grady, Patrick W., Reinhard, Karl J.,
Swisher, Mark E., White, Frances, Yates, Bonnie, Yohe II, Robert M., Yost, Chad,
Willerslev, Eske, 2013. Geochronology, archaeological context, and DNA at
paisley caves. In: Graf, Kelly E., Ketron, Caroline V., Waters, Michael (Eds.),
Paleoamerican Odyssey. Center for the Study of the First Americans, College
Station, pp. 485e510.
Judge, James W., Sebastian, Lynne, 1988. Quantifying the Present and Predicting the
Past: Theory, Method, and Application of Archaeological Predictive Modeling.
U.S. Department of the Interior, Bureau of Land Management Service Center,
Denver.
Leidwanger, Justin, 2013. Modeling distance with time in ancient mediterranean
seafaring: a GIS application for the interpretation of maritime connectivity.
J. Archaeol. Sci. 40 (8), 3302e3308.
Livingood, Patrick, 2012. No crows made mounds: do cost-distance calculations of
travel time improve our understanding of southern appalachian polity size? In:
Surface-Evans, Sarah L., White, Devin Alan (Eds.), Least Cost Analysis of Social
Landscapes: Archaeological Case Studies. University of Utah Press, Salt Lake
City, pp. 174e187.
Llobera, M., 2001. Building past landscape perception with GIS: understanding
topographic prominence. J. Archaeol. Sci. 28 (9), 1005e1014.
Llobera, M., 2011. Archaeological visualization: towards an archaeological infor-
mation science (AISc). J. Archaeol. Method Theory 193 (3).
55
Lock, Gary, Harris, Trevor, 1992. Visualizing spatial data: the importance of
geographic information systems. In: Reilly, Paul, Rahtz, Sebastian (Eds.),
Archaeology and the Information Age: a Global Perspective. Rutledge, London,
pp. 49e59.
Mackie, Alexander P., Sumpter, Ian D., 2005. Shoreline settlement patterns in Gwaii
haanas during the early and late Holocene. In: Fedje, Daryl W., Mathewes, Rolf
(Eds.), Haida Gwaii: Human History and Environment from the Time of Loon to
the Time of the Iron People. UBC Press, Vancouver, pp. 350e371.
Mackie, Quentin, Fedje, Daryl, McLaren, Duncan, Smith, Nicole, McKechnie, Iain,
2011. Early environments and archaeology of coastal british columbia. In:
Bicho, Nuno F., Haws, Jonathan A., Davis, Loren G. (Eds.), Trekking the Shore.
Springer, New York, pp. 51e103.
Mandryk, A.S. Carole, Josenhans, Heiner, Fedje, Daryl W., Mathewes, Rolf W., 2001.
2001 late quaternary paleoenvironments of northwestern north America:
implication for inland versus coastal migration routes. Quat. Sci. Rev. 20,
301e314.
Maschner, H.G., 1995. The politics of settlement choice on the Northwest coast:
cognition, GIS, and coastal landscapes. In: Sweeney, Del (Ed.), Agriculture in the
Middle Ages: Technology, Practice, and Representation. University of Pennsyl-
vania Press, Philadelphia, PA, pp. 177e189.
Mitchell, Andy, 2012. Modeling, Suitability, Movement, and Interaction. The ESRI
Guide to GIS Analysis Vol. 3, ESRI Press, Redlands.
Monteleone, Kelly Rose Bale, 2013. Lost Worlds: Locating Submerged Archaeolog-
ical Sites in Southeast Alaska. The University of New Mexico, Albuquerque. Ph.D.
Dissertation. Department of Anthropology.

Montenegro, Alvaro, Hetherington, Renee , Eby, Michael, Weaver, Andrew J., 2006.
Modelling pre-historic transoceanic crossings into the Americas. Quat. Sci. Rev.
25, 1323e1338.
Moss, M., 2008. Islands coming out of concealment: traveling to Haida Gwaii on the
Northwest coast of north America. J. Isl. Coast. Archaeol. 3, 35e53.
Moss, M., 2011. Northwest Coast: Archaeology as Deep History. Society for American
Archaeology, Washington.
Graf, Kelly E., Ketron, Caroline V., Waters, Michael (Eds.), 2013. Paleoamerican
Odyssey. Center for the Study of the First Americans, Texas.
Pedersen, Mikkel W., Ruter, Anthony, Schwager, Charles, Harvey Friebe,
Staff, Richard A., Kjeldsen, Kristian K., Mendoza, Marie L.Z., Beaudoin, Alwynne
B., Zutter, Cynthia, Larsen, Nicolaj K., Potter, Ben A., Nielsen, Rasmus,
Rainville, Rebecca A., Orlando, Ludovic, Meltzer, David J., Kjær, Kurt H.,
Willerslev, Eske, 2016. Postglacial visibility and colonization in north America's
ice-free corridor. Nature 537, 45e49.
Rasmussen, Morten, Sikora, Martin, Albrechtsen, Anders, Korneliussen, Thorfinn
Sand, Moreno-Mayar, J., Victor, Poznik, G. David, Zolikofer, Christopher P.E.,
Leo n, Marcia S. Ponce de, Allentoft, Morten E., Moltke, Ida, Jonsson, H akon,
Valdiosera, Cristina, Malhi, Ripan S., Orlando, Ludovic, Bustamante, Carlos D.,
Stafford Jr., Thomas W., Meltzer, David, Nielsen, Rasmus, Willerslev, Eske, 2015.
The ancestry and affiliation of kennewick man. Nat. Lett. 1038 (14625), 1e8.
Rasmussen, Morten, Anzick, Sarah L., Waters, Michael R., Skoglund, Pontus,
DeGiorgio, Michael, Stafford Jr., Thomas W., Rasmussen, Simon, Moltke, Ida,
Albrechtsen, Anders, Doyle, Shane M., Poznik, G.D., Gudmundsdottir, Valborg,
Yadav, Rachita, Anna-Sapfo, Malaspinas, Samuel 5th, S. White, Allentoft, Morten
E., Cornejo, Omar E., Tambets, Kristiina, Eriksson, Anders, Heintzman, Peter D.,
Karmin, Monika, Thorfinn, S. Korneliussen, David, J. Meltzer, Tracey, L. Pierre,
Stenderup, Jesper, Saag, Lauri, Vera, M. Warmuth, Margarida, C. Lopes,
Malhi, Ripan S., Brunak, Sø, Sicheritz-Ponten, Thomas, Barnes, Ian,
Collins, Matthew, Orlando, Ludovic, Balloux, Francois, Manica, Andrea,
Gupta, Ramneek, Metspalu, Mait, Bustamante, Carlos D., Jakobsson, Mattias,
Nielsen, Rasmus, Willerslev, Eske, 2014. The genome of a late Pleistocene hu-
man from a Clovis burial site in western Montana. Nature 506 (7487), 225e229.
Rockman, Marcy, Steele, James (Eds.), 2003. Colonization of Unfamiliar Landscapes:
the Archaeology of Adaptation. Routledge, London.
Sakaguchi, Takashi, Morin, Jess, Dickie, Ryan, 2010. 2010 defensibility of large
prehistoric sites in the mid-fraser region on the canadian plateau. J. Archaeol.
Sci. 37, 1171e1185.
Sherman, Jason, Balkansky, Andrew K., Spemcer, Charles S., Nicholls, Brian D., 2010.
Expansionary dynamics of the nascent monte alb an state. J. Anthropol.
Archaeol. 29 (3), 278e301.
Shugar, Dan H., Walker, Ian J., Lian, Olav B., Eamer, Jordon B.R., Neudorf, Christina,
McLaren, Duncan, Fedje, Daryl, 2014. Post-glacial sea-level change along the
pacific coast of north America. Quat. Sci. 2014 (1), 1e23.
Siart, Christoph, Eitel, Bernhard, Panagiotopoulos, Diamantis, 2008. 2008 investi-
gation of past archaeological landscapes using remote sensing and GIS: a multi-
method case study from mount ida, crete. J. Archaeol. Sci. 35, 2918e2926.
Stewart, Hillary, 1984. Cedar: Tree of Life to the Northwest Coast Indians. Douglas &
McIntyre, Vancouver.
Surface-Evans, Sarah L., White, Devin, 2012. An introduction to least cost analysis of
social landscapes. In: Surface-Evans, Sarah L., White, Devin Alan (Eds.), Least
Cost Analysis of Social Landscapes: Archaeological Case Studies. University of
Utah Press, Salt Lake City, pp. 1e10.
Taliaferro, Matthew S., Schriever, Bernard A., Steven Shackley, M., 2010. 2010
Obsidian procurement, least cost path analysis, and social interaction in the
mimbres area of southwestern New Mexico. J. Archaeol. Sci. 37, 536e548.
Van de Noort, Robert, 2003. An ancient seascape: the social context of seafaring in
the early bronze age. World Archaeol. 3 (35), 404e415.
Verhagen, Philip, 2007. Case Studies in Archaeological Predictive Modeling. Leiden
University Press, Leiden.
56 R. Gustas, K. Supernant / Journal of Archaeological Science 78 (2017) 40e56
Wheatley, David, Gillings, Mark, 2012. Spatial Technology and Archaeology: the
Archaeological Applications of GIS. Taylor and Francis, London.
White, D.A., Barber, S.B., 2012. Geospatial modeling of pedestrian transportation
networks: a case study from pre-columbian Oaxaca, Mexico. J. Archaeol. Sci. 39,
2684e2696.
Data sources:
Alaska Department of Natural Resources, 2007. Information Resource Management,
Alaska Geo-Spatial Clearinghouse. Alsk. Hydrogr. 1 (63), 360. Anchorage Alaska,
U.S. Geological Survey, Bureau of Land Management, Alaska Department of
Natural Resources.
Canadian Hydrographic Service, Hydrography Data Center e Pacific Region,
Bathymetric Gridded Data, 2014. Bathymetric Data for NTS Cell (Zone 8: 138W-
132W, Zone 9: 132W-126W). Created by Tony Dill Hydrographer Canadian
Hydrographic Service, Sidney British Columbia: Fisheries and Oceans Canada.
National Ocean and Atmospheric Administration, 2014. National Ocean Service,
Digital Sounding Surveys. Sounding Data from South East Alaska Surveys.
University of Alberta. Created by Robert Gustas, Edmonton Alberta.
Natural Resources Canada, 2013. Earth Sciences Sector; Canada centre for mapping
and earth observation. CanVecþ Single Line Watercourse Code 1470171. Gov-
ernment of Canada. Sherbrooke Quebec.
Nexen Energy ULC, 2014. Digby Island LiDAR. Prince Rupert, British Columbia.
NEXEN.
Province of British Columbia, 1997. DataBC. Terrain Resource Management Infor-
mation Program, second ed. British Columbia: Crown Registry and Geographic
Base Branch, Victoria.
U.S. Geological Survey, 2014. EROS Data Center. National Elevation Data Set. U.S.
Department of the Interior, Sioux Falls, SD.