Professional Documents
Culture Documents
PII: S0379-0738(17)30290-6
DOI: http://dx.doi.org/doi:10.1016/j.forsciint.2017.07.031
Reference: FSI 8934
Please cite this article as: Alberto Giordano, M.Katherine Spradley, Migrant deaths
at the Arizona–Mexico border: Spatial trends of a mass disaster, Forensic Science
Internationalhttp://dx.doi.org/10.1016/j.forsciint.2017.07.031
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Migrant deaths at the Arizona-Mexico border: Spatial trends of a mass disaster
Highlights
Spatial data from migrant deaths at the Arizona-Mexico border are explored.
GIScience was used to document how migrant death locations have varied over time.
Uncertainty and positional inaccuracy is noted that impacts data interpretation.
Standardized protocol for recording spatial data and meta-data is recommended.
Abstract: Geographic Information Systems (GIS) technology has been used to document, investigate,
and predict patterns that may be of utility in both forensic academic research and applied practice. In
examining spatial and temporal trends of the mass disaster that is occurring along the US-Mexico
Border, other researchers have highlighted predictive patterns for search and recovery efforts as well as
water station placement. The purpose of this paper is to use previously collected spatial data of migrant
deaths from Arizona to address issues of data uncertainty and data accuracy that affect our
understanding of this phenomenon, including local and federal policies that impact the USMexico
border. The main objective of our study was to explore how the locations of migrant deaths have varied
over time. Our results confirm patterns such as a lack of relationship between Border Patrol
apprehensions and migrant deaths, as well as highlight new patterns such as the increased positional
accuracy of migrant deaths recorded closer to the border. This paper highlights the importance of using
positionally accurate data to detect spatio-temporal trends in forensic investigations of mass disasters:
without qualitative and quantitative information concerning the accuracy of the data collected, the
reliability of the results obtained remains questionable. We conclude by providing a set of guidelines for
standardizing the collection and documentation of migrant remains at the U.S.-Mexico border.
1
INTRODUCTION
Geographic Information Technologies (GIT), can help inform forensic practice through mapping
of crime scenes [1], resolution of commingled human remains [2], the documentation in human
rights violations [3] and humanitarian forensic action. In producing maps of human rights
violations against indigenous Guatemalans, Steinberg, Height, Mosher and Bampton [3:67] write
that “mapping the locations of political violence is an important step in more deeply
understanding where and why violence took place in Guatemala from the late 1970s to the mid-
1990s.” As an analytical tool, Geographic Information Systems (GIS) technology can help not
only document, but also investigate, understand, and predict patterns that may be of utility in
One area of application for GIS documentation and investigation in the forensics field
concerns the staggering number of migrant deaths along the U.S.-Mexico border, a phenomenon
that has raised grave concern on the part of human rights organizations and activists, the general
public (at least locally), the U.S. and Mexican federal and state governments, and internationally
[4, 5], and that has created a medical and humanitarian emergency in Arizona and Texas, two
states that have witnessed a dramatic increase in the number of migrants trying to cross the
border into the U.S. [6, 7]. In this article we focus on one of these two states—Arizona—
although we will return to Texas in the conclusions section. Medical examiners and law
enforcement agencies are increasingly overwhelmed with the number of migrant remains found,
many of which remain unidentified [8]. The “where” of the phenomenon—points of origin and
destination and the route taken to enter the United States—plays a crucial role in the fate of the
migrants. Migrants crossing the border take routes for reasons that include personal
2
considerations, such as their own physical ability or financial resources, as well as external
factors such as temperature and weather, terrain, or access to roads and water. Most often they
use a coyote, someone that is paid to take migrants across the border along paths about which
In this article, we highlight geospatial and forensics aspects of the mass disaster (see
below for a definition) that continues to take place along the Arizona-Mexico border, with focus
on issues related to data uncertainty (e.g., in the cause of death) and data accuracy (e.g., where
the human remains were found) that are rarely given proper consideration in the literature.
The Arizona-Mexico border covers approximately 372 miles of mostly harsh terrain and
environment spanning the Sonora Desert. The desert is extremely dry, lacks water resources, and
daily temperature variations are high; unsurprisingly, the hot summer months of June, July, and
August are especially deadly for migrants (Table 1). In the last fifteen years the Arizona portion
of the border has emerged as the most heavily trafficked pathway into the United States [10].
(This situation might be changing: more recent data—see U.S. Customs and Border Protection
Medical examiners in Arizona, and principally the Pima County Office of the Medical
Examiner (PCOME), have been especially active in not only autopsying the remains of migrants,
but also in trying to identify the dead and connecting with their families in Mexico and in Central
America, an effort we are participating in in our home state of Texas. Migrant rights associations
are also especially active in the state, not only in their humanitarian efforts to prevent deaths
along the border, but also in documenting the size of the problem through the construction,
maintenance, update, and dissemination of datasets relative to migrant deaths. One of these
datasets, from the Humane Borders Initiative [11], has been used quite extensively by human
3
rights organizations, policy makers, and academic researchers, and is the starting point for our
work.
BACKGROUND
In 1994, following the “prevention through deterrence” principle [12, 13], the United
States Border Patrol implemented a Strategic Plan to dissuade undocumented immigrants from
gaining access into the United States through popular border crossings in Arizona, California and
Texas. The plan was to be implemented in three separate phases: Phase I in San Diego and El
Paso, known as “Operation Gatekeeper” and “Operation Hold the Line”, respectively; Phase II in
Tucson, known as “Operation Safeguard”; and Phase III, which is unnamed, for the rest of the
It has been argued that these policies have been largely ineffective in stopping illegal
immigration and, in the process, have cost many migrants their lives [12]. In fact, migrant deaths
have become so common and widespread along the U.S.-Mexico border that the phenomenon
has been described as an example of the “structural violence” that results from specific structural
and political conditions [7:263], a type of violence that often goes unnoticed because it does not
necessarily involve a perpetrator. The phenomenon also fits the definition of “mass disaster,” due
to the number of casualties—over 100,000 of active cases of missing persons and over 40,000
unidentified remains—and their capacity for exhausting and overwhelming the resources of
What is clear is that entrance into the country was made more difficult and dangerous by
the “funneling” of migrants through especially treacherous terrain, and particularly Arizona’s
Sonora Desert, in an attempt to deter illegal entry [8, 12, 14, 17]. In fact, in the Tucson Border
4
Patrol Sector, which includes the Sonora Desert, the number of migrant’s recovered bodies has
increased dramatically from 1990 to 2005 [8], and between 1990 and 2012 the remains of 2,238
migrants were brought to the Pima County Office of the Medical Examiner for examination [6],
a likely underestimate of the actual number of deaths, as not all human remains are recovered.
More recently [11], 2007 and 2010 had the highest numbers of recovered remains with 208 and
226, respectively (Table 2). This is in contrast with a decline in Border Patrol apprehensions,
from a maximum of over 700,000 in 2000 to approximately 70,000 in fiscal year (FY) 2015 [18],
increases were largely due to reductions in migrant apprehensions in California, while decreases
in more recent years have been counterbalanced by an increase in Texas. In the Tucson Sector,
apprehensions amounted to 10% or less of the total number of cases for FY 1960-1993, increased
progressively in the following years, reaching 30% or more between FY 1998-2006, 40% or
more in FY 2004 and then again in FY 2007-2010, and decreased in the following FY to 29% in
2013, 18% in 2014, and 19% in 2015. These trends notwithstanding, in 2011-2013 there were
more than 100 migrant deaths per 100,000 apprehensions, compared with 40 in the mid-2000s
[10: 54].
Studies conducted along the Arizona-Mexico border and employing spatial analytical
methods and Digital Elevation Models (DEM) have shown that migrant deaths are more likely to
occur near the border, near roads, near major drainages, and away from water tanks [14, 19], and
that recovered migrant remains tend to be spatially clustered, showing that it is possible to
predict certain paths migrants are likely to take [9]. In addition to terrain characteristics, a
variable used to study migrant deaths is temperature [20, 21]. Focusing on Pima County in
Arizona and using medical examiner data from 1998 to 2003, Keim, Mays, Parks, Pytlak, Harris
5
and Kent [22] found that fatalities correlated to heat waves but did not correlate with the number
of Border Patrol apprehensions (as we have already noted). For the years 2002-2003, Sapkota,
Kohl III, Gilchrist, McAuliffe, Parks, England, Flood, Mack Sewell, Perrotta, Escobedo, Stern,
Zane and Nolte [23] showed that heat exposure was the number one cause of death, while
Ruttan, Stoltz, Jackson-Vance, Parks and Keim [24: 407] used 2002-2009 medical examiner
datasets from Pima, Arizona and Cochise, Arizona to show that at a “critical threshold daily high
temperature (DHT) of 40° C, the probability of at least one heat death was 50%.”
From a forensic perspective, Martinez [12] looked at migrants’ cause of death for the
Tucson Border Patrol Sector for fiscal years 1990 to 2007, showing that causes of death were
related to what part of Mexico the migrant was from, with victims from the northern part of the
country significantly less likely to die of exposure and motor vehicle accident when compared to
homicide than victims from central Mexico. The author attributes this fact to the increased
likelihood for migrants from regions close to the border to have friends or family connection in
the United States. Also in Arizona, Anderson [25] reviewed how the PCOME identifies migrant
remains and discussed problems associated with identifying remains, including rapid
decomposition, lack of health records (in particular, dental), and families not being able to travel
to the U.S. to identify loved ones. More recently (2013), another study conducted by the BMI
using 1990-2012 data from the PCOME, noted that identified migrants came from thirteen Latin
American countries, with 82.2% Mexican nationals, 7.1% from Guatemala, 2.3% from El
Salvador, and smaller percent from the remaining countries, noting a geographical shift in the
region of origin within Mexico, with a substantial decrease of migrants from the north part of the
increase of migrants originating from the central and southern regions (from 14% in 1990-1999
6
to about 45% in 2000-2012), and an increase in migrants coming from countries other than
One aspect that stands out in the forensic literature we examined is the lack of a uniform
protocol for recording and identifying migrant remains. Depending on policy and/or
circumstances, such task may fall with a medical examiner, a Border Patrol agent, other law
enforcement agency, or a county coroner [14, 23]. Many counties have no medical examiner, so
the task of recording and identifying human remains gets passed off to someone with less
experience and knowledge [26]. From a spatial analytical perspective, the Arizona-based studies
cited are generally limited to a short span of time and do not allow for a longer temporal
perspective. Furthermore, these studies use spatial variations in temperature and elevation as
analytical factors, while other elements—such as the presence of temporal variations in the
clustering of migrant deaths—are not investigated, if not cursorily. Another limitation concerns
the often uncritical acceptance of the dataset(s) used in the analysis, with scarce attention paid to
the accuracy and uncertainty of the data collected and of the datasets created.
The main objective of our study is to identify the “where” aspects referred to by
Steinberg, Height, Mosher and Bampton [3] and how these locations have varied over time. We
are also interested in discussing the reliability of analytical results, taking into consideration both
the degree of uncertainty in the identification of the time and cause of migrants’ death , and the
degree of positional accuracy in recording the location where human remains were found,
information that the Humane Borders OpenGIS dataset includes as an attribute for all recorded
locations. In addressing the “where” and how these locations have varied over time in addition to
data scrutinization, we hope that our research can contribute to understand the spatial and
7
forensics dimensions of a mass disaster that continues to take place along the U.S.-Mexico
border.
Data
Migrant remains have been found in seven counties along the Arizona-Mexico border,
including La Paz, Maricopa, Yuma, Pinal, Pima, Santa Cruz, and Cochise (Figure 1), with a vast
majority in Pima County. The location of these deaths is recorded in the Arizona OpenGIS
humanitarian assistance local organization—in partnership with the Pima County Office of the
Medical Examiner [27]. The dataset includes information on where migrant remains were
recovered (i.e., GPS points and/or relative location), the cause of death, date of reporting, and
The temporal scope of our analysis include the years 2002 through 2015, for a total of
2,280 cases (Figure 1). Due to only a few cases reported, we eliminated 2001 from the analysis.
An example of the records retrievable online is provided in Table 3. For our work, we focused on
location (using GPS coordinates), cause of death, report year, and location precision. Once
downloaded, the 2002-2015 dataset was edited to fix errors such as capitalization, spelling,
punctuation, etc. This ensured that queries on records and attributes returned accurate counts.
After editing, the dataset was exported into ESRI ArcMap version 10.1 to create a point map
layer of migrant deaths and for analysis (Figure 1). Figure 4 also includes county layers and the
8
location of the Tohono O’odham Indian reservation from the Arizona Land Resource
geographic scales Kernel Density was used to derive one or more clusters (the choice is made by
the analyst) that identify neighborhoods at large local scales, allowing the researcher to focus on
relatively small areas on the ground. Once clusters were generated, mean center, directional
distribution, and standard distance were used to measure the geographic center of the spatial
distribution of migrant death locations and the directionality and degree of dispersion of such
locations, and therefore summarize spatial patterns at a small scale for the entire study area. By
looking at these measures over time, it is possible to identify temporal variations in the location
More specifically, Kernel Density calculates the density of points within a study area—in
our case, the location of migrant deaths in southern Arizona—using a kernel function. In order to
compare results over time, we standardized the parameters (search radius around point and
output cell size for the cluster) used by the function. The reliability of the spatial patterns
identified was then measured by comparing the results obtained regardless of the stated
positional accuracy of the point data used with results obtained using only the three most
accurately-recorded migrant death locations. The Mean Center calculates the average x and y
coordinates of a set of points, in this case the location of migrant deaths. Mean centers were
calculated for all migrant deaths for every year and then by cause of death, also for each year.
The Directional Distribution measures the geographical orientation of the data points,
represented on a map as an ellipse of varying size. Like in the Kernel Density analysis, we used
9
the same parameters for all years to ensure comparability of results. Finally, the Standard
Distance measures the degree of dispersion, or concentration, of the data points around the mean
center. The output of the analysis is a circle: the larger the circle, the more dispersed the data.
RESULTS
Temporal trends
The vast majority of migrant deaths along the Arizona-Mexico border occur in Pima
County and, within the county, in the Tohono O’odham reservation. The most frequently
reported cause of death is “exposure” (905 of 2,280 cases, or 39.7%), followed by “skeletal
remains” (650, or 28.5%). The use of “skeletal remains” does not represent a cause of death but
simply indicates the finding of human bones. The third and fourth causes of death (Table 2)—
“blunt force injury” (126 cases, or 5.5%) and “gunshot wound” (44, or 1.9%)—are the likely
result of violent events. In 364 cases (16%), the cause of death is recorded as “undetermined.”
Taken together, the top five causes of death account for 2,089 of the 2,280 cases recorded
(91.6%).
Temporal patterns also emerge. In the case of “blunt force injury,” although numbers are
low (10 or less), they are relatively high in 2004, 2005, 2006, and especially in 2008 and 2009
(Table 4). The numbers for “exposure” remain high throughout the time period, but drop
considerably in the period 2011-2015, in absolute as well as in percent values. The opposite
trend can be detected for “skeletal remains,” with just a handful of cases in the earlier years,
followed by a dramatic increase after 2010, peaking in 2014 with 106 cases. In fact, “skeletal
remains” overtakes “exposure” as the listed cause of death in 2011-2015. “Undetermined” causes
of death continue to increase from 2002 to 2011 and decrease dramatically afterwards. Finally,
10
2007 and 2010 show the highest recorded numbers of migrant deaths, with 208 and 226
respectively.
Close to half (43.4%) of the migrant deaths recorded occurred within the boundaries of
the Tohono O’odham reservation (Figure 2), which extends along approximately 20% of the
Arizona-Mexico border [28]. The numbers fluctuate considerably, from a peak of 67.7% in 2002
to a minimum of 32.8% in 2005, with most values in the 30% to 40% range. As Table 4 shows,
annual increases or decreases in the percent of migrant deaths recorded in the reservation are in
line with totals for the study area in eight of the thirteen years under study, with increase in
2010-2011, and 2011-2012. The largest differences occurred, in sign as well as magnitude, in
2003-2004 and 2012-2013. These differences seem to indicate that local factors play an
Kernel Density
Kernel Density analysis for the years 2002-2015 confirms that migrant deaths tend to
cluster in the Tohono O’odham Indian reservation, and in areas within Pima County to the east
of the reservation (in red and blue in Figure 3). Other areas of concentration (in yellow in Figure
3)—although less dense—include north of Tucson in Pima County and pockets in the eastern and
western part of the county. Year-by-year comparisons (maps not shown here for brevity) reveal
dense clusters in the reservation in 2003, 2006, 2007, 2010, 2012, and 2015. Clusters are smaller,
or less dense, in other years, which indicate that the patterns of migrant deaths tend to be more
dispersed in those years (2015 is somewhat of an exception, with strong clusters dispersed across
much of the study area). Figure 3, on the right, shows Kernel Density analysis limited to the
three most accurate location designations [“GPS Coordinate (precise to within ca. 100ft/100m)”,
11
“Street Address (precise to within ca. 1,000ft/300m)”, and “Physical description with directions,
distances, and landmarks (precise to within 1mi/2km”)]. A comparison of the two maps in Figure
3 reveals similar spatial patterns, although in certain years (most notably, 2008) and at the local
scale (i.e., single cluster) variations might occur (maps not shown here for brevity). Overall, the
year-by-year analysis reveals that the highest concentration of migrant deaths tend to be near the
border and in the southeast corner of the Tohono O’odham Reservation, especially in 2002,
Zooming out to a smaller scale, we can explore patterns for the entire Arizona-Mexico
border. Figure 4 shows the Mean Center (MC) of migrant deaths for the years 2002-2015. All
MCs are located in Pima County, inside the Tohono O’odham reservation, and relatively near
each other: the distance between the MC for 2015 and the MC for 2004—respectively the
westernmost and the easternmost MC—is about 28 miles. This area corresponds approximately
to the densest cluster in Figure 3. Note the outliers of 2015 in the west and 2004 and 2008 in the
east; if we exclude these three years from the analysis, the distance between the easternmost
The distribution of the MCs follows an east-west directionality, with relatively little
north-south variation (about 6 miles). The picture that emerges is interesting: the mid-years
2005-2009 (with the exception of 2008) are fairly similar to each other and similar to the overall
pattern 2002-2015, while the early years (2002, 2003) and the late years (2010-2012 and 2014)
are similar in their western orientation; 2004, 2008, and 2015 are outliers and transition years. As
Table 5 shows, the largest differences in distance from successive MCs occurs between 2003 and
2004 and between 20013 and 2014, when the spatial distribution of migrant deaths moved 17.5
12
miles to the east and 15.3 miles to the west respectively, indicating substantially different
Similar to Kernel Density, the reliability of the spatial patterns identified were verified
with the MCs by including in the analysis only the three most accurate causes of death. As
Figure 5 shows, using these limiting factors has the effect of pulling the MC slightly to the
southwest. The implication is that migrant deaths recorded near the border tend to be more
accurately located and also tend to have been recorded as being caused by one of the three or five
top causes of death. Finally, Figure 6 shows the size and extent of the Standard Distance and the
Directional Distribution for all migrant deaths (top left map). The ellipse representing the DD is
in Pima County and centered in the Tohono O’odham reservation, oriented parallel to the border
and relatively large, thus confirming the findings from Kernel Density that migrant deaths are
relatively dispersed across Pima County. This orientation persists for the top causes of death and
the most accurate location designations (top right and bottom maps), while the size of the SD
circle gets slightly smaller, indicating relatively higher concentration as the number of cases
considered decrease—recall that the smaller the circle, the more spatially concentrated the data
points are. As concerns the Mean Centers, what jumps out visually is that the MC for the top
three causes of death (bottom map)—“exposure” (905 cases), “skeletal remains” (650), and
“undetermined” (364)—are located close to each other and close to the overall MC for all causes
of death (top left map), with “blunt force injury” (126) and “gunshot wounds” (44) off to the
northeast (top right map). It thus appears that violent causes of death follow patterns that are
slightly different from the top three causes of death. However, this finding needs to be taken with
caution: as already remarked, the cause of death for “skeletal remains” and “undetermined” are
13
As Table 5 shows, when compared to the data for all years (2002-2015), the five top
causes of death appear to be considerably more concentrated in 2010 (73.3% of the size of the
2002-2015 circle) and especially in 2002 (56%) and more dispersed in 2004, 2005, and
especially in 2013, 2014, and 2015. The last three years of the study period are especially
noteworthy, as the increased dispersion of migrant deaths might signal a greater variety of paths
taken by migrants as they cross the border, with the year 2015 recording the most dispersed
spatial patterns. When only the top three causes of death and the three most accurate locations
are examined, similar patterns are confirmed, with increased concentration or dispersion
depending on the years, although it is only in the three most recent years that an increase in
dispersion is confirmed.
Finally, Table 6 shows variations in the location of MC between all migrant deaths and
only the top three causes of death with the most accurate locations. In all cases but one (2005),
data reduction leads to a movement of the MC to the southwest, closer to the border. As has
already been noted, this indicates not only that the top three causes of death tend to be located
near the border, but also that migrant deaths tend to have been more accurately recorded in that
area. A possible explanation could be Border Patrol recording locations more accurately with
GPS technology.
DISCUSSION
Stretching along approximately 20% of the Arizona-Mexico border and entirely in Pima
County and within the boundaries of the Tucson Border Patrol sector, the Tohono O’odham
reservation recorded 43.4% of migrant deaths in the years 2002-2015. Factors that might explain
the disproportion between percent of migrant deaths and length of the border include policies
14
implanted by the reservation authorities [4] and an increase in Border Patrol staffing numbers
[29] along the Tucson sector over the last decade, with numbers around 4,000 or more in FY
years 2011-2015. This might have forced migrants—or, more likely, the coyotes who lead them
into the U.S.—to take more arduous and more isolated routes, including going through the
reservation, where water stations are not allowed [30]. On the other hand, adding more agents
might by itself explain why more remains are found, even in the most remote areas [31]: some of
these remains might have been in the region for years, only to be found when more agents are on
the ground. However, these explanations are not especially convincing. Compare, for example,
the years 2002 and 2012, when approximately the same number of migrant remains were found
(130 and 122), in spite of a two-and-a-half-time increase (1,626 to 4,176) in the number of
The increase in border surveillance was spurred by federal and state policies, including
the Secure Fence Act of 2006, which called for building barriers, roads, efficient lighting, and
cameras along the U.S.-Mexico border from California to Texas [32], and by Arizona SB 1070
(later modified to HB 2162) in April 2010. The hotly debated Arizona SB 1070, also known as
the “Support Our Law Enforcement and Safe Neighborhoods Act”, included provisions that
“added state penalties relating to immigration law enforcement including trespassing, harboring
and transporting illegal immigrants, alien registration documents, employer sanctions, and
human smuggling” [33]. However, as Table 2 and Figure 2 show, the number of migrant deaths
actually went down after 2010 relatively to earlier years, although 2010 was itself a record year
for migrant deaths, both overall and in the reservation. A more likely explanation for the
presence of drug cartels traveling through the area, an event almost guaranteed to bring violence
15
[28, 34, 35]. The main cartels operating on the reservation—the Sinaloa and the Tijuana—have
expanded their services beyond smuggling drugs to include human trafficking, often abandoning
migrants in the desert or sacrificing them to the Border Patrol in an effort to get their drug cargo
across the border [36]. In the end, it is more likely that the variation in the number of migrant
deaths in the period under study were caused by local factors and by Mexico’s “drug wars,”
together with the state of the economy in Mexico and the U.S.
Zooming in on spatial patterns, Kernel Density reveals that the higher concentration of
migrant deaths tend to be in the southwest corner of the Tohono O’odham Reservation and in
general in areas of Pima County closer to the border, especially in 2002, 2003, 2006, 2007, 2010,
and 2012 (maps not shown here for brevity). The Mean Centers showed three main clusters: a
central one that includes the years 2005-2008 and 2013; one to the west for the years 2002-2003,
2010-2012, and 2014, with the extreme outlier of 2015; one to the east that include two outliers
year, 2004 and 2008. Thus, at both small (MCs) and large (Kernel Density) scales the early years
(2002-2003) appear somewhat similar in spatial pattern to the later years 2010-2014, with 2004,
2008, and 2015 as outliers. As Table 7 shows, between 2002 and 2015, 61.5% of migrant death s
were recorded with GPS technology and at the highest level of accuracy (within 100 meters),
with 81.1% of the data achieving an accuracy of 1 mile or better. Positional accuracy standards
increased consistently over the years: “vaguely” recorded locations stayed in the 30% range for
2002-2008 and decreased dramatically after 2009. In fact, in 2011-2012 and 2014-2015 over
90% of the location recorded had an accuracy of 100 m or higher, with values over 80% in 2010
and 2013. As for the Directional Distribution and the Standard Distance, the orientation of
migrant death is—not unexpectedly—parallel to the border in virtually all years (maps not
shown here for brevity), while concentration and dispersion patterns tend to be similar in all
16
years, with the very notable exceptions of 2013, 2014, and 2015, in which patterns tend to be
The three most frequently recorded causes of death (“exposure,” “skeletal remains,” and
“undetermined”) tend to stay closer to each other than the fourth and fifth causes (“blunt force
injury” and “gunshot wounds”), due probably to the relative infrequency of the latter (maps not
shown here for brevity). However, apparently random variations occur over the years. When
only the three most accurate locations are considered, the overall year-by-year variability of the
spatial patterns is reduced, especially in the early years, while in later years the patterns remain
stable. Of the top three causes of death, “skeletal remains” appears to be most sensitive to
variations in the positional accuracy of the data recorded, as Table 8 shows. Finally, it has been
noted that the positional accuracy of recorded migrant deaths tends to increase near the border.
The same could be said for the top three causes of death: they, too, seem to concentrate along the
border.
One additional point should be noted as concerns recorded causes of death: their number
tend to vary considerably year by year, and especially in the period 2011-2015 (Table 4). While
the reasons for such variations are difficult to pinpoint with certainty, it is likely they have to do
with how data were collected and recorded. When skeletal remains are found, the forensic
anthropologist provides an analysis, and then the medical examiner will take the anthropological
analysis and provide cause and manner of death, if possible. However, manner of death in
skeletal remains are typically only determined in the presence of obvious trauma; otherwise, the
cause and manner of death will be recorded as “unknown” or “undetermined.” Many skeletal
remains brought to the Pima County Office of the Medical Examiner are incomplete skeletons
and only when the complete skeleton is recovered can an assessment be made regarding the
17
presence or lack thereof of trauma. This could be the reason for reporting “skeletal remains” in
lieu of “undetermined.” Additionally, in early 2011 the chief medical examiner retired and this
human factor might explain why the designation “skeletal remains” overtook “exposure”
beginning in 2012.
The limitations of this study and of the dataset should be noted. The first concerns the
positional accuracy of migrant deaths locations, as just discussed. The second limitation is
qualitative in nature, and relates to the individual(s) who recorded the location and cause of death
of the migrant remains found, with Border Patrol agents using methods that often differs from
those of other agencies and of local law enforcement. Also to be considered, a body found near
the border is not necessarily the body of a migrant entering the U.S. illegally. In fact, due to the
or forensic anthropologist might accidentally classify a migrant’s body as a legal citizen and vice
versa [23]. However, the Pima County Office of the Medical Examiner—with which one of us
has collaborated for years—has a methodology in practice for assessing migrant vs. non-migrant
deaths, including the use of a cultural profile based on factors such as where the body was found,
what personal belongings were found at the scene, dental characteristics of the victim, and
various cultural items [37]. Maybe the most important limitation of this study—and of studies
similar to this one—is that an unknown proportion of human bodies are never recovered [8, 12].
The creation, update, and maintenance of datasets documenting and recording the
location of migrant deaths at the U.S.-Mexico border gives policy makers, law enforcement,
civic groups, and researchers an instrument to track the number and location of migrant deaths,
18
predict where migrants will cross the border, and provide humanitarian assistance [14] . This
information can help to better understand the where in relation to the why, to return to Steinberg
et al.’s statement in the Introduction. Analyzing the dataset at different spatial scales and with
different techniques, we have been able to strengthen the validity of some of the results: for
example, the presence of robust and consistent directional and concentration patterns in the years
under study. And, by looking at migrant deaths both spatially and temporally, we have confirmed
patterns that had been noted by others—such as the lack of relationship between Border Patrol
apprehensions and migrant deaths—as well as patterns that others had not observed, such as the
increased positional accuracy of migrant deaths recorded closer to the border. Finally, and
perhaps most importantly, we have stressed the importance of using positionally accurate data to
detect spatio-temporal trends: without qualitative and quantitative information concerning the
accuracy of the data collected, the reliability of the results obtained remains questionable. On this
point, it is interesting to note that the year 2008 seems to be an outlier for many of the spatial
in data collection techniques on the ground, the accuracy of reporting, the creation of the
differences that set 2008 apart from the other study years. Perhaps more interesting—because
more recent—is to note that the spatial patterns for the last three years, and especially for 2014
and 2015, show increasing spatial dispersion of migrant deaths as well as general westward
In the end, we call for the creation of a standardized protocol that would allow law
enforcement and others who come across human remains along the border to document their
findings according to a set of simple and clear guidelines [8]. Such guidelines should be
19
formulated and agreed upon though a collaborative process that involves medical examiners,
forensic anthropologists, public health professionals, GIScience experts, and law enforcement
agencies at the local, state, and federal level. The following guidelines are suggested:
Collect contextual data from the death scene that could also better inform cause and
manner of death;
anthropological analysis;
Assess and document the positional accuracy of the recorded location where the body or
Note the elevation at which the body was found (all GPS units display this information)
and the general environment characteristics of the surrounding area (e.g., the presence of
trees, the dryness of the area, etc.), which might be especially useful for predictive
modelling;
These recommendations are especially timely, including in these authors’ own state of
Texas, as Texas has seen a dramatic increase of migrant deaths along its 1,241 miles of border
with Mexico in the last two years, surpassing the totals for Arizona. Unfortunately, Texas is
unprepared to deal with the phenomenon, and especially with issues such as the documentation,
20
identification, and repatriation of the bodies found. The majority of the South Texas counties do
not have a medical examiner’s office, and therefore it is the Justice of the Peace who is
responsible for recording and maintaining records related to unidentified human remains. This
information is not centralized, it is not available to the public, and therefore it cannot be analyzed
in ways similar to what we and other researchers have done with the Arizona dataset. Our own
field work along the Texas-Mexico border, in conjunction with the Forensic Border Coalition—
an organization whose mission is to support the families of missing migrants searching for their
loved ones and to address issues related to the identification of human remains along the
border—indicates that in Texas in many cases migrant deaths have not been recorded at all.
To conclude, the Arizona model, in which data collected by the PCOME is made
available through the Humane Borders OpenGIS Initiative, is a novel and powerful way to start
understanding where migrant deaths take place and why and to better understand the patterns of
death in relation to federal policies [14]. Designing a standardized spatial data collection protocol
and the use of standard nomenclature, e.g., cause/manner of death, condition of remains, by
medical examiner’s and justices of the peace or coroners, can provide national data than can be
utilized by local, state, and federal government agencies to better document, understand, and
ultimately prevent migrant deaths. The use of this investigative tool for large-scale, search and
recovery efforts in HFA deserves to be researched further and may prove to be a valuable
Acknowledgments
21
We would like to thank Jamie L. Henkhaus (Master of Applied Geography, Texas State
University, 2014) for help in collecting and editing some of the data used in this article, the
Humane Borders OpenGIS Initiative, and the Pima County Office of the Medical Examiner.
22
REFERENCES
[2] N.P. Herrmann, J.B. Devlin, Assessment of Commingled Human Remains Using a GIS-
Based Approach, in: B.J. Adams, J.E. Byrd (Eds.), Recovery, Analysis, and Identification of
Commingled Human Remains, Humama Press2008.
[3] M.K. Steinberg, C. Height, R. Mosher, M. Bampton, Mapping massacres: GIS and state
terror in Guatemala, Geoforum 37(1) (2006) 62-68.
[4] M. Jimenez, Humanitarian Crisis: Migrant Deaths at the U.S.-Mexico Border, ACLU of San
Diego & Imperial Counties and Mexico's National Commission of Human Rights, San
Diego, 2009.
[5] T. Brian, F. Laczko, Fatal Journeys: Tracking Lives Lost during Migration, International
Organization for Migration, Geneva, Switzerland, 2014.
[8] R. Rubio-Goldsmith, M.M. McCormick, D. Martinez, I.M. Duarte, The “Funnel Effect” and
Recovered Bodies of Unauthorized Migrants Processed by the Pima County Office of the
Medical Examiner, 1990-2006, Binational Migration Institute, Tuscon, 2006.
[9] W.D. Grannis, Detecting and Assessing Aggregate Human Pedestrian Migration Patterns
Using High Resolution Multispectral Imagery adn Geographic Information Systems,
Department of Geography, Texas State University, San Marcos, 2012.
[10] R. Reineke, D. Martinez, Migrant Deaths in the Americas (United States and Mexico), in: T.
Brian, F. Laczko (Eds.), Fatal Journeys: Tracking Lives Lost during Migration, International
Organization for Migration, Geneva, Switzerland, 2014, pp. 45-83.
23
[11] I. Humane Borders, Arizona Open GIS Initiative for Deceased Migrants. <
http://humaneborders.info/, 2013 (accessed February/May 25.2016).
[12] D. Martinez, Unauthorized Border Crosser Deaths in the Southern Arizona Desert:
Examining the Relationship between Demographic Characteristics and Cause of Death,
Annual Meeting of the American Sociological Association, Boston, MA, July 31, 2008,
2008.
[13] U.S. Customs and Border Protection, Border Patrol Strategic Plan: 1994 and Beyond.
<http://cw.routledge.com/textbooks/9780415996945/gov-docs/1994.pdf>, 1994 (accessed
January 8, 2016.2016).
[15] G.R. Ciottone, Introduction to Disaster Medicine, in: G.R. Ciottone, P.D. Anderson, E.A.D.
Heide, I. Jacoby, E. Noji, S. Suner (Eds.), Disaster Medicine, Mosby Elseiver, Philadelphia,
2006, pp. 3-6.
[16] N. Ritter, Missing and Unidentified Remains: The Nation’s Silent Mass Disaster, National
Institute of Justice Journal 256 (2007) 2-7.
[18] U.S. Customs Border Protection, Southwest Border Sectors, Total Illegal Aliens
Apprehensions by Fiscal Year.
<https://www.cbp.gov/sites/default/files/documents/BP%20Southwest%20Border%20Secto
r%20Apps%20FY1960%20-%20FY2015.pdf>, 2015 (accessed May 25, 2016.2016).
[19] S. Lawrence, J. Wildgen, Manifold Destiny: Migrant Deaths and Destinations in the
Arizona Desert, Growth and Change 43(3) (2012) 482-504.
24
[21] M.J. Hinkes, Migrant Deaths along the California-Mexico Border: An Anthropological
Perspective, J Forensic Sci 53(1) (2008) 16-20.
[22] S.M. Keim, M.Z. Mays, B. Parks, E. Pytlak, R.M. Harris, M.A. Kent, Estimating the
Number of Heat-Related Deaths among Immigrants in Pima County, Arizona, J Immigr
Minor Health 8(2) (2006) 185-91.
[23] S. Sapkota, H.W. Kohl III, J. Gilchrist, J. McAuliffe, B. Parks, B. England, T. Flood, C.
Mack Sewell, D. Perrotta, M. Escobedo, C.E. Stern, D. Zane, K.B. Nolte, Unauthorized
Border Crossings and Migrant Deaths: Arizona, New Mexico, and El Paso, Texas, 2002-
2003, Am J Public Health 96(7) (2006) 1282-1287.
[25] B.E. Anderson, Identifying the Dead: Methods Utilized by the Pima County (Arizona)
Office of the Medical Examiner for Undocumented Border Crossers: 2001-2006., J Forensic
Sci 53(1) (2008) 8-15.
[26] C. Kovic, Searching fo the Living, the Dead, and the New Disappeared on the Migrant Trail
in Texas: Preliminary Report on Migrant Deaths in South Texas, Texas Civil Rights Project,
2013, Houston, 2013.
[27] I. Humane Borders, Arizona Open GIS Initiative for Deceased Migrants. <
http://humaneborders.info/, 2014 (accessed February/May 25.2016).
[28] E. Eckholm, War without Borders: In Drug War, Tribe Feels Invaded by Both Sides, The
New York Times, The New York Times 2010.
[29] U.S. Customs and Border Protection, Border Patrol Agent Staffing by Fiscal Year (as of
September 19, 2015).
<https://www.cbp.gov/sites/default/files/documents/BP%20Staffing%20FY1992-
FY2015.pdf>, 2015 (accessed May 25, 2016.2016).
[30] J. Burgess, H. Park, Migrant Deaths in Southern Arizona, The New York Times, 2013.
[31] M. Mellow, Immigrant Deaths Rose with Border Security Boost, Study Finds, Huffington
Posot, 2014.
25
[32] U.S. Congress, Public Law 09–367—OCT. 26, 2006: Secure Fence Act of 2006, in: U.S.
Congress (Ed.) Government Printing Office, Washington D.C., 2006.
[35] B. Pitts, D. Lieberman, In Efforts to Secure U.S.-Mexico Border, Ariz. Native Americans
Feel Caught in the Middle. <http://abcnews.go.com/US/efforts-secure-us-mexico-border-
ariz-native-americans/story?id=19496394>, 2013 (accessed December 9, 2015.2015).
[36] F. Diez, B. Vourvoulias, Cartels Exploit Immigration Crisis to Smuggle More Drugs across
Border. <http://latino.foxnews.com/latino/news/2014/07/03/cartels-exploit-immigration-
crisis-to-smuggle-more-drugs-across-border/>, 2014 (accessed July 28, 2014.2014).
[37] W.H. Birkby, T.W. Fenton, B.E. Anderson, Identifying Southwest Hispanics Using
Nonmetric Traits and the Cultural Profile, J Forensic Sci 53(1) (2008) 29-33.
26
Figure 1. Dot map of southern Arizona where migrant deaths have been recorded, 2002-
2015. Total in the dataset: 2280.
27
Figure 2: Reservation deaths by year in comparison to the total number of deaths, 2002-
2015. Total in the dataset: 990.
28
Figure 3: Kernel Density for all causes of death (left) and for the three most accurate
location designations (right), 2002-2015.
29
Figure 4: Mean centers of all deaths, 2002-2015.
30
Figure 5: Mean centers for: a) all migrant deaths; b) top five causes of death; c) top five
causes of deaths and three most accurate location designations; d) top three causes of
death; e) top three causes of deaths and three most accurate location designations.
31
Figure 6: Comparison of directional distribution, standard distance, and MCs for all
migrant deaths (top left), the top five causes of death (top right), and top three causes of
death (bottom).
32
Table 1: Migrant Deaths by month, 2002-2015.
Month Count
January 111
February 130
March 127
April 153
May 198
June 343
July 423
August 258
September 170
October 122
November 129
December 116
TOTAL 2,280
33
Table 2: Migrant Deaths by year, 2002-2015.
Year Count
2002 130
2003 137
2004 156
2005 186
2006 163
2007 208
2008 162
2009 187
2010 226
2011 172
2012 122
2013 150
2014 137
2015 144
TOTAL 2,280
34
Table 3. Sample of database attribute fields used in this study.
Report Cause of
Name Location Precision County Latitude Longitude UTM X UTM Y
Year Death
Abarca Hidalgo, GPS Coordinate (precise to Skeletal Santa
2011 31.42483 -110.826 516,537 3,476,699
Audenago within ca. 300ft./100m) Remains Cruz
Vague physical description
Abrego-Paz, Blunt Force Santa
2009 (precise to within 31.69694 -110.59884 538,016 3,506,914
Marlon Injury Cruz
15mi/10km)
Vague physical description
Aburto Aguilar,
2006 (precise to within Exposure Pima 32.31873 -111.71523 432,676 3,575,990
Juan
15mi/10km)
35
Table 4: Cause of death count by year. The top five causes of death are highlighted in bold.
CAUSE
OF 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 TOTAL
DEATH
1
Exposur 8 61. 8 62. 7 44. 71. 6 39. 8 42. 36. 39. 41. 28. 18. 28. 10. 19. 39.
3 59 74 94 49 23 43 14 28 905
e 0 5% 5 0% 0 9% 0% 5 9% 9 8% 4% 6% 6% 5% 9% 7% 2% 4% 7%
2
Skeletal 4.6 5.8 1.9 2.2 1 6.7 2 12. 21. 22. 28. 54. 63. 53. 77. 66. 28.
6 8 3 4 34 42 64 93 77 80 106 96 650
Remains % % % % 1 % 6 5% 0% 5% 3% 1% 1% 3% 4% 7% 5%
Undeter 1 14. 2 18. 4 25. 2 12. 5 32. 4 22. 21. 17. 20. 8.1 9.0 4.7 4.4 4.2 16.
35 32 46 14 11 7 6 6 364
mined 9 6% 5 2% 0 6% 3 4% 3 5% 7 6% 6% 1% 4% % % % % % 0%
Blunt
6.9 5.8 2 14. 1 5.9 1 8.6 3.8 11. 9.1 2.7 1.2 4.7 2.2 2.8 5.5
Force 9 8 8 19 17 6 2 - - 7 3 4 126
% % 2 1% 1 % 4 % % 7% % % % % % % %
Injury
2.3 1.9 0.5 3.7 1 8.7 4.7 3.6 2.8 2.1
Pending 3 - - 3 1 6 - - - - - - - - - - 7 5 4 47
% % % % 8 % % % % %
Gunshot 2.3 3.6 1.3 2.2 3.1 3.2 1.3 3.5 5.7 1.5 0.7 1.9
3 5 2 4 5 - - - - 6 3 6 7 - - 2 1 44
Wound % % % % % % % % % % % %
Drownin 1.9 1.6 1.2 0.5 3.7 2.7 1.8 0.6 2.5 0.7 2.1 1.4
- - - - 3 3 2 1 6 5 4 1 3 1 - - 3 32
g % % % % % % % % % % % %
Motor
3.1 0.7 1 7.1 1.1 0.6 1.9 0.6 1.1
Vehicle 4 1 2 1 4 1 - - - - - - - - - - - - - - 24
% % 1 % % % % % %
Accident
Other
Injury / 2.3 0.7 1.1 0.6 1.0 1.9 1.1 0.4 1.2 0.7 0.8
3 1 - - 2 1 2 3 2 1 2 - - 1 - - - - 18
Homicid % % % % % % % % % % %
e
Heart 1.5 1.5 1.3 0.5 1.8 0.9 1.2 0.7 0.7
2 2 2 1 3 - - - - - - 2 2 - - 1 - - - - 15
Disease % % % % % % % % %
Not
1 5.3 1.2 1.3 0.7
Reporte - - - - - - - - - - 2 - - - - - - - - 2 - - - - 15
1 % % % %
d
Other 0.8 0.7 0.5 0.6 2.1 1.3 1.2 0.8 0.6
1 1 - - - - - - 1 1 4 3 2 1 - - - - - - 14
Disease % % % % % % % % %
Other 2.1 0.4 0.7 1.4 0.4
- - - - - - - - - - - - - - 4 1 - - - - - - 1 2 8
Injury % % % % %
0.5 0.6 0.5 0.4 0.2
Asphyxia - - - - - - 1 - - - - 1 1 1 - - - - - - - - - - 4
% % % % %
0.5 0.6 0.5 0.6 0.2
Diabetes - - - - - - 1 1 1 1 - - - - - - - - - - - - - - 4
% % % % %
Drug
0.6 0.6 0.7 0.1
Overdos - - - - - - - - 1 - - - - - - - - 1 - - 1 - - - - 3
% % % %
e
Exsangui 0.4 0.0
- - - - - - - - - - - - - - - - 1 - - - - - - - - - - 1
nation % 4%
Lightning 0.7 0.0
- - 1 - - - - - - - - - - - - - - - - - - - - - - - - 1
Strike % 4%
Non- 0.5 0.0
- - - - - - 1 - - - - - - - - - - - - - - - - - - - - 1
viable % 4%
36
fetus
1 1 1 1 1 2
TOTAL 100 100 100 100 100 100 10 10 10 10 10 10 10 10 228 100
3 3 5 8 6 0 162 187 226 172 122 150 137 144
COUNT % % % % % % 0% 0% 0% 0% 0% 0% 0% 0% 0 %
0 7 6 6 3 8
37
Table 5: Size of Standard Distance circles in square miles and relative to each other for the top five causes of death.
38
Table 6: Distances between Mean Centers for all migrant deaths recorded and Mean Centers for the top three causes of death and the
three most accurate locations.
Year Distance
2002-2015 7,919 meters to the SW
2002 4,412 meters to the SW
2003 10,114 meters to the SW
2004 11,573 meters to the SW
2005 1,983 meters to the NE
2006 13,513 meters to the SW
2007 6,498 meters to the SW
2008 12,500 meters to the SW
2009 10,038 meters to the SW
2010 5,404 meters to the SW
2011 5,886 meters to the SW
2012 5,074 meters to the SW
2013 4,930 meters to the SW
2014 3,306 meters to the SW
2015 3,675 meters to the SW
39
Table 7: Location accuracy levels by year, 2002-2015.
Year totals
Location precision 2002-
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
2015
GPS Coordinate
(precise to within ca. 34 16 40 54 78 109 94 127 186 158 118 126 127 136 1403
300ft./100m)
Street Address (precise
to within ca. 2 2 11 8 12 5 7 3 8 6 - 5 4 3 76
1000ft/300m)
Physical description
with directions,
distances, and 53 64 53 66 29 30 14 24 15 3 2 13 4 1 371
landmarks (precise to
within 1mi/2km)
Vague physical
description (precise to 41 55 52 58 44 64 47 33 17 4 2 5 2 3 427
within 15mi/10km)
Individual moved from
location where they
- - - - - - - - - 1 - - - - 1
originally collapsed or
died
No location
- - - - - - - - - - - 1 - 1 2
description
TOTAL 130 137 156 186 163 208 162 187 226 172 122 150 137 144 2280
(Table 7 continued)
Year
Location precision
% 2002-
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
2015
GPS Coordinate
(precise to within ca. 2.4% 1.1% 2.9% 3.8% 5.6% 7.8% 6.7% 9.1% 13.3% 11.3% 8.4% 9% 9.1% 9.7% 61.5%
300ft./100m)
Street Address
(precise to within ca. 2.6% 2.6% 14.5% 10.5% 15.8% 6.6% 9.2% 3.9% 10.5% 7.9% - 6.6% 5.3% 3.9% 3.3%
1000ft/300m)
Physical description
with directions,
distances, and 14.3% 17.3% 14.3% 17.8% 7.8% 8.1% 3.8% 6.5% 4% 0.8% 0.5% 3.5% 1.1% 0.3% 16.3%
landmarks (precise
to within 1mi/2km)
40
Vague physical
description (precise
9.6% 12.9% 12.2% 13.6% 10.3% 15% 11% 7.7% 4% 0.9% 0.5% 1.2% 0.5% 0.7% 18.7%
to within
15mi/10km)
Individual moved
from location where
- - - - - - - - - 100% - - - - 0.04%
they originally
collapsed or died
No location
- - - - - - - - - - - 50% - 50% 0.1%
description
(Table 7 continued)
Year %
Location precision
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
41
TOTAL 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100%
42
Table 8: Location accuracy by top three causes of death, 2002-2015.
Year totals
Cause of death and
Location precision 2002-
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
2015
Exposure 80 85 70 132 65 89 59 74 94 49 23 43 14 28 905
GPS Coordinate (precise to
22 8 24 41 39 55 42 60 72 48 23 40 11 28 513
within ca. 300ft./100m)
Street Address (precise to
- 2 1 5 1 2 2 1 5 - - 2 1 - 22
within ca. 1000ft/300m)
Physical description with
directions, distances, and
38 39 24 47 9 13 2 3 6 1 - - 1 - 183
landmarks (precise to within
1mi/2km)
Vague physical description
20 36 21 39 16 19 13 10 11 - - 1 1 - 187
(precise to within 15mi/10km)
Skeletal Remains 6 8 3 4 11 26 34 42 64 93 77 80 106 96 650
GPS Coordinate (precise to
3 1 - 2 5 13 19 33 61 85 75 71 102 93 563
within ca. 300ft./100m)
Physical description with
directions, distances, and
1 6 2 - 2 4 4 1 2 1 - 4 1 - 28
landmarks (precise to within
1mi/2km)
Street Address (precise to
- - - - - - 1 1 - 2 - 1 2 1 8
within ca. 1000ft/300m)
Vague physical description
2 1 1 2 4 9 10 7 1 4 2 3 1 2 49
(precise to within 15mi/10km)
Individual moved from location
where they originally collapsed - - - - - - - - - 1 - - - - 1
or died
No location description - - - - - - - - - - - 1 - - 1
Undetermined 19 25 40 23 53 47 35 32 46 14 11 7 6 6 364
GPS Coordinate (precise to
7 3 12 7 24 29 27 21 39 12 10 6 5 6 208
within ca. 300ft./100m)
Street Address (precise to
2 - 7 1 4 - 1 - 1 1 - - 1 - 18
within ca. 1000ft/300m)
Physical description with
directions, distances, and
6 13 8 6 8 5 3 4 5 1 1 - - - 60
landmarks (precise to within
1mi/2km)
Vague physical description
4 9 13 9 17 13 4 7 1 - - 1 - - 78
(precise to within 15mi/10km)
TOTAL 105 118 113 159 129 162 128 148 204 156 111 130 126 130 1919
43
(Table 8 continued)
Year %
Cause of death and Location
precision 2002-
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
2015
Exposure 8.8% 9.4% 7.7% 14.6% 7.2% 9.8% 6.5% 8.2% 10.4% 5.4% 2.5% 4.8% 1.5% 3.1% 100%
GPS Coordinate (precise to within
4.3% 1.6% 4.7% 8.0% 7.6% 10.7% 8.2% 11.7% 14.0% 9.4% 4.5% 7.8% 2.1% 5.5% 100%
ca. 300ft./100m)
Street Address (precise to within ca.
- 9.1% 4.5% 22.7% 4.5% 9.1% 9.1% 4.5% 22.7% - - 9.1% 4.5% - 100%
1000ft/300m)
Physical description with directions,
distances, and landmarks (precise to 20.8% 21.3% 13.1% 25.7% 4.9% 7.1% 1.1% 1.6% 3.3% 0.5% - - 0.5% - 100%
within 1mi/2km)
Vague physical description (precise
10.7% 19.3% 11.2% 20.9% 8.6% 10.2% 7.0% 5.3% 5.9% - - 0.5% 0.5% - 100%
to within 15mi/10km)
Skeletal Remains 0.9% 1.2% 0.5% 0.6% 1.7% 4.0% 5.2% 6.5% 9.8% 14.3% 11.8% 12.3% 16.3% 14.8% 100%
GPS Coordinate (precise to within
0.5% 0.2% - 0.4% 0.9% 2.3% 3.4% 5.9% 10.8% 15.1% 13.3% 12.6% 18.1% 16.5% 100%
ca. 300ft./100m)
Physical description with directions,
distances, and landmarks (precise to 3.6% 21.4% 7.1% - 7.1% 14.3% 14.3% 3.6% 7.1% 3.6% - 14.3% 3.6% - 100%
within 1mi/2km)
Street Address (precise to within ca.
- - - - - - 12.5% 12.5% - 25% - 12.5% 25% 12.5% 100%
1000ft/300m)
Vague physical description (precise
4.1% 2.0% 2% 4.1% 8.2% 18.4% 20.4% 14.3% 2% 8.2% 4.1% 6.1% 2.0% 4.1% 100%
to within 15mi/10km)
Individual moved from location
where they originally collapsed or - - - - - - - - - 100.0% - - - - 100%
died
No location description - - - - - - - - - - - 100% - - 100%
Undetermined 5.2% 6.9% 11.0% 6.3% 14.6% 12.9% 9.6% 8.8% 12.6% 3.8% 3.0% 1.9% 1.6% 1.6% 100%
GPS Coordinate (precise to within
3.4% 1.4% 5.8% 3.4% 11.5% 13.9% 13% 10.1% 18.8% 5.8% 4.8% 2.9% 2.4% 2.9% 100%
ca. 300ft./100m)
Street Address (precise to within ca.
11.1% - 38.9% 5.6% 22.2% - 5.6% - 5.6% 5.6% - - 5.6% - 100%
1000ft/300m)
Physical description with directions,
distances, and landmarks (precise to 10% 21.7% 13.3% 10% 13.3% 8.3% 5.0% 6.7% 8.3% 1.7% 1.7% - - - 100%
within 1mi/2km)
Vague physical description (precise
5.1% 11.5% 16.7% 11.5% 21.8% 16.7% 5.1% 9.0% 1.3% - - 1.3% - - 100%
to within 15mi/10km)
44