EXTENDED ABSTRACT
Submitted for presentation at the XXVI IUSSP International Population Conference
Climate change and population predictions: Spatial variability in populations at risk for
sea level rise
Katherine J. Curtis White
Department of Rural Sociology
Center for Demography and Ecology
Applied Population Lab
Annemarie Schneider
Center for Sustainability and the Global Environment
Nelson Institute for Environmental Studies
September 2008
Please direct all correspondence to White at 1450 Linden Drive, Madison, WI 53706 or
kwhite@ssc.wisc.edu. The authors wish to acknowledge Paul R. Voss and Jenn Huck for
technical assistance.
2
Considerable popular and scientific attention has been given to the potential impacts of climate
change. Chief among these concerns are the consequences for the human population. Indeed,
significant technical and conceptual advances have been made in recent years to understand
the interrelationship between human populations and the environment by several teams of
researchers (e.g., McGranahan et al. 2007; O’Neill et al. 2001). Despite this progress and the
compelling political and scientific motivations to understand the demographic implications of
climate change, the study of the two areas has not intersected to produce meaningful localized
estimates of the demographic implications of climate change. For example, research on climate
often makes a case for the likely impacts of global warming on human populations, yet the
resulting climate change scenarios are not related to current or future population estimates.
Extant research also has tended to focus at the national or regional scale, thus masking spatial
variability in climate impacts on populations at the sub-national scale. Further, demographicallyoriented
research on the environment tends to focus on the human contribution to climate
change; population estimates are used to improve, for example, pollution scenarios on
emissions. There is little to no work on the future populations in these areas, their composition,
migration patterns, or other population characteristics. This information is critical for
understanding the vulnerability of specific population groups, for planning mitigation and
adaptation strategies, and for informing policy. The current study makes an important
contribution to multiple fields by exploiting discipline-specific tools, placing climate and
population models on the same temporal scale, and by producing population projections at a
socially and politically meaningful spatial scale (i.e., the county level).
Our objective is to demonstrate the value of examining spatial variability in time-correlated
climate and population projections at the sub-national scale. We demonstrate the
methodological approach by focusing on sea level rise and total population size for a select
sample of counties in low-lying coastal zones within the United States. We restrict the analysis
to regions affected by one climate change outcome, sea level rise, and we limit our projections
to total population size. Our intention is to develop a larger research agenda to pursue the
impacts of a range of geophysical events related to climate change (e.g., land use degradation,
increased hazards) on current and future populations, and most critically, determine the
implications for specific population groups (e.g., age-, race-, and income-specific groups) within
the United States and across the globe. Our initial results show the potential of this type of
detailed demographic projection for local populations.
Data & Analysis
Climate-Change Scenarios
The balance of scientific evidence now shows that anthropogenic emissions of greenhouse
gases are having a discernible effect on the Earth’s climate. Global average air and ocean
temperatures have increased, with global average temperatures projected to climb between 1.4
and 5.8 degrees C by the end of this century (IPCC 2007). Widespread melting of ice and snow
has occurred as a result of global warming and is evidenced, in part, by the observed shrinking
of the Artic sea ice extent. When combined, these changes have resulted in sea level rise at an
average rate of 1.8mm/yr since 1961, and 3.1 mm/yr since 1993. Recent IPCC scenarios show
that the rise of global average sea level by 2100 will be in the range from 18-38 to 26-59 cm
depending on the emissions scenario (IPCC 2007).
The anticipated climate changes have important consequences for the human population given
settlement patterns. As temperatures increase and sea level rises at faster rates than previously
observed, a substantial number of persons currently live in coastal areas considered at high risk
3
for sea level rise, flooding and storm surges (Small et al. 2004). Recent studies show that more
than 10 percent of the world’s population live in the world’s low elevation coastal zones (a
contiguous zone along the coast less than 10 m above sea level), with a larger share of the
population (14 percent) in developing countries living in this area compared to more developed
regions (10 percent) (McGranahan et al. 2007). Although research has begun to bring together
climate change scenarios and population projections, investigations in the geophysical sciences
continue to use static estimates of current population, while the demography arena has focused
on coarse, brush-stroke models of population projections at the region- or country-level without
regard for local or spatial variability. The current approach uses this past research as a point of
departure to examine questions about localized impacts of climate change.
Case Selection
While it is often difficult to disentangle the impacts of climate change on human populations
from other driving forces (e.g., the impact of rising temperatures on human health), the potential
effects of sea level rise are unequivocal and will undoubtedly cause an immediate and important
impact on population (i.e., increased vulnerability, displacement, and migration). With this in
mind, we use sea level rise scenarios (1m and 4m rise) to define ‘at-risk’ locations within the
continental U.S. Areas of potential inundation are derived from Mulligan’s (2007) analysis of
90m remote sensing data from the Shuttle Radar Topography Mission (SRTM V3 data with
corrections applied by the Consortium for Spatial Information) coupled with the coastlines and
water body dataset derived from the NASA SRTM Water Body dataset. After compiling,
mosaicking and projecting the dataset to an equal-area projection, we intersect the maps of
predicted sea-level rise with county political boundaries within a GIS to determine the areas with
the greatest amount of inundated land (Figure 1). From this step, we produce a rank of the
counties with the highest degree of impact in terms of overall area inundated and percent of
county inundated (Table 1). Our study sample, therefore, represents five areas that consistently
appear at the top of the rankings as those most impacted by either 1m or 4m sea level rise.
Five areas, comprised of several contiguous counties, have been selected based on sea level
impacts in addition to population size and composition. The reader will quickly note the absence
of New Orleans and other southern areas that were impacted by Hurricane Katrina. While these
counties were estimated to experience significant damage from sea level rise, methodological
problems arise because of the timing of Katrina (2005) and the baseline population estimate
(2000). Although the 2000 population estimates for areas hit by Katrina are accurate for this
date, the areas experienced dramatic out-migration which makes forecasting area population
dubious at best and completely unreasonable at worst.
In total, 24 counties are analyzed. All selected areas are estimated to experience at least a 1-
meter rise in sea level, with some counties experiencing greater impact (in square kilometers
damaged). The selected areas are distributed across the United States and capture five distinct
place types: (1) the California cluster is an area rich in agricultural production and has a large
immigrant and Latino population; (2) the Florida cluster is a popular retirement destination and
an immigration destination for distinct Latino groups; (3) the counties within the New Jersey
cluster have a tradition of industrial production; (4) the South Carolina cluster has a relatively
large African American population and is within the southern region that has, in recent decades,
experienced population growth through internal migration; and (5) the Virginia cluster is a high
density area that is comprised of a largely professional population. Combined, the selected
areas represent various geographic and demographic profiles that characterize the nation.
Population Projections
4
Annual population forecasts are estimated through 2030 by projecting forward the 2000
population baseline estimate according to county migration, fertility and mortality rates reported
by the U.S. Census Bureau (2001) and the National Center for Health Statistics (2001a, 2001b).
We use migration rates that have been adjusted to address census undercounts among specific
age and race groups by a team of researchers headed by Dr. Paul Voss (Voss et al. 2004). In
the current study, county estimates available through the national organizations are compared
with estimates reported by state organizations as well as 2005 population estimates.
This strategy, like all forecasts, is imperfect. Weaknesses arise from error in the population
baseline estimates themselves and error in the assumptions underlying the forecasts. In terms
of the estimates, census data are reliable but not without error; certain populations are
undercounted. Regarding underlying assumptions, forecasts are based on trends believed to be
valid for the projection horizon. Future growth, however, may depart from historical patterns.
Despite the imperfections, population forecasts are critical for analysts and service providers
interested in the implications of climate change, like sea level rise. The projections are not
intended to be perfect predictions of what will come. Rather, population projections are
scenarios of what could happen given model assumptions. The employed model assumes that
current rates of natural increase and migration will generally persist through 2030. This
assumption is inaccurate given that factors affecting these sources of population growth can
change, yet it is reasonable given that we do not have a strong sense of precisely what
exogenous factors might arise and how they might alter trends in population growth.
Results
Study results include estimates of the population impacted by sea level rise for each of the
selected study areas. Preliminary results suggest that the magnitude of the estimated impact
ranges between the counties from 11,821 to about 3.3 million people. In addition, the dynamics
of migration are analyzed. The top destinations (out-migration) and sending counties (inmigration)
in 1990 and 2000 for a subsample of the selected counties with the largest
metropolitan area are examined. This analytical strategy illustrates that the effects of sea level
rise are not only experienced by the county that lost suitable land, but the impacts extend to
counties that will need to house the uprooted population and to counties that would have sent
migrants to the no longer inhabitable areas. Moreover, the population implications of sea level
rise are further compounded by the connectedness of places that are directly affected by sea
level rise; some of the top receiving and sending counties will also experience a loss of
inhabitable land due to sea level rise.
References
Intergovernmental Panel on Climate Change, 2007. Climate Change 2007, The Physical
Scientific Basis: Summary for Policymakers, 21 pp., Geneva, available at
http://www.ipcc.ch.
McGranahan, G., D. Balk, and B. Anderson. 2007. “The Rising Tide: Assessing the Risks of
Climate Change and Human Settlements in Low Elevation Coastal Zones.” Environment
and Urbanization 19(1):17-37.
Mulligan, M., 2007. Global sea level change analysis based on SRTM topography and coastline
and water bodies dataset (SWBD).Version 1.0. Database available at:
http://www.ambiotek.com/sealevel
O’Neill, B.C., F.L. MacKellar, and W. Lutz. 2001. Population and Climate Change. Cambridge,
UK: Cambridge University Press.
5
Small, C., and J. Cohen, 2004. “Continental physiography, climate, and the global distribution of
human population”, Current Anthropology, 45(2): 269-277.
Voss, P.R., S. McNiven, R.B. Hammer, K.M. Johnson, and G.V. Fuguitt. 2004. "County-Specific
Net Migration by Five-Year Age Groups, Hispanic Origin, Race and Sex 1990–2000."
Working Paper 2004–24. Center for Demography and Ecology, University of Wisconsin–
Madison.
U.S. Department of Commerce, Bureau of the Census. 1995. 1990 Census of Population and
Housing: County-to-County Migration Flow File on CD-ROM Special Project (SP) 312
[machine-readable data file]. Washington, DC: U.S. Bureau of the Census.
U.S. Department of Commerce, Bureau of the Census. 2001. 2000 Census of Population and
Housing: Summary File 1 United States. Washington, DC: U.S. Census Bureau.
U.S. Department of Commerce, Bureau of the Census. 2003. 2000 Census of Population and
Housing: County-to-County Migration Flow Files [Computer file]. Washington, DC: U.S.
Bureau of the Census.
U.S. Department of Health and Human Services, National Center for Health Statistics. 1993-
2001a. Natality Detail Data, 1990-2000. Hyattsville, MD: U.S. Department of Health and
Human Services, National Center for Health Statistics.
U.S. Department of Health and Human Services, National Center for Health Statistics. 1993-
2001b. Multiple Cause of Death of ICD- 9 Data, 1990-2000. Hyattsville, MD: U.S.
Department of Health and Human Services, National Center for Health Statistics.
6
Table 1: Rank of counties by extent of area inundated and proportion of county flooded for sea
level rise scenarios of 1m and 4m
Note: Counties in Louisiana have been grayed to denote that they were not considered during
sample selection.
7
Figure 1: A map of the South Carolina coastal zone, one of the five selected study areas. The
study area is shown in dark green, while inundation is shown in orange (1m sea level rise) and
red (4m sea level rise). For reference, urbanized areas are shown in yellow.