Dasymetric Mapping - NC GIS Conference

Presentation by
Calvin “Not an Expert” Harmin
MS Candidate (2015)
East Carolina University Dept. of Geography
[email protected]
linkedin.com/in/charmin
nccoastalatlas.org
Dasymetric Mapping
IMPROVING ESTIMATES OF VULNERABLE
COASTAL POPULATIONS
Disclaimer: I’m no “dasy expert”
But I hope you enjoy this intro to
dasymetric mapping!
How do we improve our understanding
of where people live?
How could this aid our efforts in
emergency management?
How do we know where people live?
 Field Work/Surveys
US Census
Population sampling
 Residential addresses
 HOW MANY people
 Demographics
 statistical estimation
within Census boundaries
Census Aggregation Issues
Residential
Census Block
Bizniss
Census Aggregation Issues
Modifiable Areal Unit Problem:
“…the areal units (zonal objects) used in many geographical
studies are arbitrary, modifiable, and subject to the whims
and fancies of whoever is doing, or did, the aggregating.“
-Dr. Stan Openshaw (1983)
The modifiable areal unit problem.
Norwick: Geo Books
How else can we know where people live?
 Property Information
 Tax assessors
 Parcels
 Building information
 Remote Sensing
 Land Cover
 Land use
 “Developed”
 Residential square feet
 Bedrooms
 Apartment units
 Building footprints
 Most on “developed” land
 NOT in water
 NOT in fields
 NOT in forests (mostly)
Dasymetric Mapping
http://eomag.eu/
http://eomag.eu/
Census Data
(or other stats)
+
Ancillary Data
(land use/property)
+ magic
=
Dasymetric
Map!
US Census Heirarchy
 North Carolina
 Counties: 100
 Tracts: 2,195
 Block groups: 6,155
Blocks: 297,238
US Census Variables to “Dasy-fy”
 A wealth of other socio-economic and demographic
variables can be used instead of just “total population”
 Disability/Health/Children/Other risk-associated factors
 Pets?
 However, fewer attributes may be available for blocks
compared to ‘higher’ Census districts.
CENSUS Example – Currituck County
http://www.nhgis.org is AWESOME!
Census Data Sources: www.census.gov;
Minnesota Population Center. National Historical Geographic Information System: Version
2.0. Minneapolis, MN: University of Minnesota 2011. http://www.nhgis.org
CENSUS Example – Currituck County
Total Population ~ 24,000
Census Tracts: 8
Block Groups: 15
Blocks: 741
CENSUS Example – Currituck County
 Coastal
 Rural
 Vulnerable to
storm surge and
riverine flooding
World Street Map Basemap
Total Population ~ 24,000
CENSUS Example – Currituck County
Census block population per acre
Census block population per acre
Currituck Census Blocks
Population
0 - 19
20 - 65
66 - 163
164 - 359
360 - 998
World Street Map Basemap
Total Population ~ 24,000
US CENSUS – Currituck County
Some ‘Empty’ Outer Banks Blocks
Currituck Census Blocks
Population
0 - 19
20 - 65
66 - 163
164 - 359
360 - 998
Blocks with zero population? Second homes/tourism?
ESRI Imagery Basemap
LAND COVER – Currituck County
Coastal Change Analysis Program (CCAP)
http://coast.noaa.gov/digitalcoast/data/ccapregional
Background
Unclassified
Developed, High Intensity
Developed, Medium Intensity
Developed, Low Intensity
Developed, Open Space
Cultivated Crops
Pasture/Hay
• 2010 data
• Landsat-derived
Grassland/Herbaceous
Deciduous Forest
Evergreen Forest
Mixed Forest
Scrub/Shrub
Palustrine Forested Wetland
Palustrine Scrub/Shrub Wetland
Palustrine Emergent Wetland
Estuarine Forested Wetland
Estuarine Scrub/Shrub Wetland
Estuarine Emergent Wetland
Unconsolidated Shore
Bare Land
Open Water
Palustrine Aquatic Bed
• 30m pixels
LAND COVER – Currituck County
Coastal Change Analysis Program (CCAP)
http://coast.noaa.gov/digitalcoast/data/ccapregional
Derived from Landsat, like
the National Land Cover
Dataset (NLCD) but with
extra processing for
coastal environments.
LAND COVER – Currituck County
Coastal Change Analysis Program (CCAP)
http://coast.noaa.gov/digitalcoast/data/ccapregional
Background
Unclassified
Developed, High Intensity
Developed, Medium Intensity
Developed, Low Intensity
Developed, Open Space
Cultivated Crops
Pasture/Hay
Grassland/Herbaceous
Deciduous Forest
Evergreen Forest
Mixed Forest
Scrub/Shrub
Palustrine Forested Wetland
Palustrine Scrub/Shrub Wetland
Palustrine Emergent Wetland
Estuarine Forested Wetland
Estuarine Scrub/Shrub Wetland
Estuarine Emergent Wetland
Unconsolidated Shore
Bare Land
Open Water
Palustrine Aquatic Bed
ESRI Imagery Basemap
Dasymetric Processing Tool
USGS Dasymetric Tool
http://geography.wr.usgs.gov/science/dasymetric/
 Can calculate 3 different
population weights for 3
“inhabited” land use
classifications:
 High/Low/Rural
 Some land cover classes
need to be combined
(subjective).
Rasterization of population polygons
Land Cover Reclassifying
Land
Cover Reclassifying
Background
Dasy Tool Classes
Unclassified
Developed, High Intensity
Developed, Medium Intensity
Developed, Low Intensity
Developed, Open Space
Cultivated Crops
Pasture/Hay
Grassland/Herbaceous
Deciduous Forest
1. High+Medium = High Intensity Urban
2. Low+Open+Bare = Low Intensity Urban
3. Crops+Pasture = Non-Urban
Evergreen Forest
Mixed Forest
0. All Others Excluded From Pop
Scrub/Shrub
Palustrine Forested Wetland
Palustrine Scrub/Shrub Wetland
Palustrine Emergent Wetland
Estuarine Forested Wetland
Estuarine Scrub/Shrub Wetland
Estuarine Emergent Wetland
Unconsolidated Shore
Bare Land
Open Water
Palustrine Aquatic Bed
Reclassification scheme decided from
visual inspection in Currituck County.
Chose most applicable classes -- those
which often contained buildings.
ArcGIS: Spatial Analyst > Reclass >
Reclassify Tool
DasyTool Notes
USGS Dasymetric Tool
http://geography.wr.usgs.gov/science/dasymetric/
NEW VERSION:
USGS Dasymetric Mapping Tool - ArcGIS 10+ Toolbox (Python)
• Make sure your input datasets are all using the same coordinate and
projection information
• Tool runs more efficiently with ArcGRID files
• Your Ancillary Raster Land Use file should be in a thematic format, NOT
continuous.
Beta version… sort of broken  (as of 2/2015)
Beta version requires you to rename your feature/raster files
names and field names to match the python script. Or edit
the script. Still workable, hopefully will be updated soon.
Dasymetric Processing
USGS Dasymetric Tool
http://geography.wr.usgs.gov/science/dasymetric/
Dasymetric Processing
USGS Dasymetric Tool
http://geography.wr.usgs.gov/science/dasymetric/
Census blocks
CCAP land use (reclassified)
Census block unique ID field
Census block population
value field
Magical empirical sampling
for land use “weighting”
See
http://geography.wr.usgs.gov/science/d
asymetric/data/methods.pdf
Dasy Output Comparison
“persons per acre” vs “persons per pixel”
Census blocks with land
Dasy raster ouput – stretch 2.5 std dev
Comparing 100-yr Flood Zone Intersect
Census blocks with land
Dasy raster ouput – stretch 2.5 std dev
Comparing 100-yr Flood Zone Intersect
Census blocks with land
Dasy raster ouput – stretch 2.5 std dev
Dasy in the Literature – Other Methods
Mapping urban risk: Flood hazards, race, & environmental justice in New York
Maantay, J., & Maroko, A. (2009). Applied Geography, 29(1), 111-124.
 No Land Cover
 Use parcel data with
building information
like residential ft2.
 Use improved
estimates of
flooded residences
to investigate E.J
issues.
Dasy in the Literature – A Small Sample
• Dasymetric methodology / uncertainty
• (Mennis, J. 2003)
• (Maantay, J. A., Maroko, A. R., & Herrmann, C. 2007)
• (Petrov, A. 2002)
• (Nagle, N. N., Buttenfield, B. P., Leyk, S., & Spielman, S. 2014)
• crime mapping
• (Bowers & Hirschfield, 1999; Poulsen & Kennedy, 2004)
• accessibility measures in health studies
• (Langford & Higgs, 2006)
• environmental justice and health research
• (Maantay, J., & Maroko, A. 2009)
• (Maantay, Maroko, & Porter-Morgan, 2013)
• Identifying socioeconomic/environmental risk patterns
• (Parrott et al., 2007)
• facilitate accessibility measures
• (Linard, Gilbert, Snow, Noor, & Tatem, 2012)
Major Issues with Dasy
 Road Networks are often “developed” land use classes
 Filtering/aggregating can remove many roads, but also lose ‘valid’ cells
 Pre-processing road networks out of land cover model can improve this
Unfiltered
Aggregated to 60m
Major Issues with Dasy
 No standardized methodology
 Every study does it differently?
 30 meter raster too coarse to capture rural homes
 Difficult to get salient property data
 Digital records not standardized between counties
Bottom Line
 Problems aside, dasy techniques readily increase
accuracy of estimated population within “areas of
interest” (e.g. hazard overlay).
 Not everyone will need such accurate population
density data, but the potential value and use of
should be investigated further.
Going Global?
 Once dasy methods mature more, higher resolution
global land use change data may be ubiquitous
 Enhance population estimation of remote areas?
 Disaster assessment?
 Enhance tracking and modeling of urban change
 Sprawl / Climate change refugees?
Please Comment / Question
 How do you think improved population data might be used,
and by whom?
 How do you decide “where people are at risk” for hazard
studies?
 Hospital populations?
 Night time vs. Day time?
Calvin “I’m Looking for a Job” Harmin
ECU - MS Geography (2015)
[email protected]
linkedin.com/in/charmin
Special thanks to NC GIS
Organizers and Workers
My advisor Dr. Tom Allen, Rob
Howard, and Herbert Stout
`