DASYMETRIC POPULATION MODELING TO ESTIMATE RESIDENT EXPOSURE TO HAZARDS: CLACKAMAS COUNTY, OR

Urban development in areas prone to natural hazards amplifies the potential for losses due to future catastrophic events.  In communities facing multiple hazards, detailing the extent of exposure provides vital information for emergency management planning.  In Clackamas County, Oregon, population increased approximately 21% from 1990 to 2000.  Multiple natural hazards (high volume snowfall/avalanche, advancing wildfires, volcanic unrest, and river flooding) also exist with the potential to disrupt local business and community livelihood. While numerous studies have characterized the hazards, there is significant lack in understanding the societal vulnerability among communities to such events.  Such an evaluation requires better estimates of the spatial distribution of population and what changes have occurred over time. 

In the United States, emergency managers have access to decadal census data that summarizes residential populations into aggregated areal units, but these units do not reflect population distributions relative to individual residences. More precise estimates of population distribution are needed so emergency managers can better address issues related to preparedness and response planning, such as for evacuations and delivery of emergency services.  Dasymetric population modeling refines coarse areal population data into more spatially relevant map units by using land-use information in the population estimation and, therefore, provides more functional data for planning purposes.

A dasymetric population modeling methodology (Figure 1 a-c) is discussed and a time series of Clackamas County residential development and habitation from 1990 to 2007 is presented.  In this analysis, U.S. Census block group records from 1990, 2000, and estimated values for 2007 were used as population input totals.  Select land-use categories were derived from the 1992 and 2001 30-meter pixel National Land Cover Dataset (NLCD).  Prior to dasymetric processing, the land-use categories were filtered to remove uninhabited regions using a rasterized county dataset of specific residential structure locations.  This extra step significantly improved resident estimates from large forested and agricultural land-use classes.  Final processing used our online dasymetric population modeling tool (available at http://geography.wr.usgs.gov/science/dasymetric/index.htm).  Population change in the county observed over the 17 year study period was calculated through raster subtraction of the 1990 population from that of 2007, and shows a general increase in urban development.  Loss of population in certain areas may reflect the movement of people to new suburban developments located in a different census block group (for instance, to “Bedroom Communities”) or may reflect dilution of population numbers due to faster rates of residential structure development as compared to block group population totals.

Overall, dasymetric population modeling helps emergency managers better understand societal vulnerability to natural hazards by more precisely mapping the spatial distribution of residents.  One of the U.S. Geological Survey (USGS) science strategies is to develop models to support emergency managers with hazard mitigation decision-making.  Improved population modeling coupled with community hazard vulnerability analysis further assists implementation of adaptation strategies to minimize social and economic disruptions from threat events. 

Acronyms Used:

USGS             United States Geological Survey

NLCD             National Land Cover Dataset