Uploaded by IPC CDM-Savar

Dasymetric Mapping Exercise

advertisement
http://web.pdx.edu/~jduh/courses/geog492w11/lab04.htm
Introduction
You will use a dasymetric mapping technique in ArcGIS to create a population distribution map. Imagine that
you want to create a map of population distribution for the Beaverton Creek subwatershed for your
hydrological research and the only population GIS layer you have is a 1990 census block group map. Such an
aggregate census map doesn't reflect the actual distribution of population in the watershed, so you decide to
disaggregate the census population using ancillary data. You take a 1992 land-cover map derived from
Landsat imagery (MRLC) and use dasymetric mapping to disaggregate the population. Please read the
Holloway et al. (1997) paper for more information.
Instructions
Click here to download the data you will need for this exercise. You need to unzip the file. Follow the
instructions below and answer all questions at the end of this exercise.
1. Set up the ArcMap document.
*
*
*
*
Start ArcMap and select a New Empty Map.
Add the landcover grid and BlockGroups feature class from the Beaverton Creek feature dataset
to ArcMap.
Enable Spatial Analyst and open its toolbox.
Set extent and cell size to the same as layer "landcover" in the Processing Extent and Raster
Analysis sections in Environments under the Geoprocessing menu. (Note: Every time you restart
ArcMap, be sure to set the analysis extent and cell size again.)
2. Calculate the population for each land cover cell.
The instructions below show you how to implement the dasymetric method in ArcGIS. You need to
understand the method so that the ArcGIS procedures would make sense. We will use a modified equation
based on the equation show in Figure 2 in Holloway et al.'s article to calculate the population for each
land-cover cell (pixel).
P = ((RA*(PA/ PA))*N/E)/AT
Where, P is the population of a cell,
RA is the relative density of a cell with land-cover type A,
PA is the proportion of cells of land-cover type A in the enumeration unit,
N is the actual population of enumeration unit (i.e., census block group)
E is the expected population of enumeration unit calculated using the relative densities. E equals the
1 of 5
08/03/2019, 13:35
http://web.pdx.edu/~jduh/courses/geog492w11/lab04.htm
sum of the products of relative density and the proportion of each land-cover type in each
enumeration unit.
AT is the total number of cells in the enumeration unit.
PA/ PA, which cancels each other out, is to calculate the unit area in the grid-based dasymetric
mapping method. The output population is estimated for each cell, not for each mapping unit (in the
vector approach). The cell size of the land-cover layer is 133 by 133 feet. You need to divide the area
of the enumeration units by 133 x 133 to get the value of AT. The values of RA for different landcover types are given in the table below.
Land-Cover Code
*
Descriptions
Relative Density (RA)
1
Low density residential
15
2
High density residential
65
3
Commercial/Industrial
5
4
Agricultural
10
5
Natural
5
6
Water/wetland
0
First, we generate a raster map of N using the convert, feature to raster tool in the Conversion
Toolbox > To Raster > Polygon to Raster.
*
Set BlockGroups as the input feature, and
*
POP90 as the field, and
*
popu90 as the output grid saved in your lab folder. Click OK.
3. Create a raster layer for RA, the relative density per cell.
*
Open the Reclassify tool from the Spatial Analyst toolbox.
*
In the reclassify dialog window, set landcover as the input raster
*
Type in the corresponding RA values in the new values column.
*
Set rdensity as the output raster. Click OK.
4. Next, we need to calculate E, the expected population per block group, using the population densities
per cell calculated in the previous step. The value of E is decided by the proportions of land-cover
types in each census block group. To calculate the proportions, first create a raster map of census block
groups' FIPS codes.
*
2 of 5
Open the polygon to raster tool again.
08/03/2019, 13:35
http://web.pdx.edu/~jduh/courses/geog492w11/lab04.htm
*
Set BlockGroups as the input feature..
*
Set FIPS as the field.
*
Set bgfips as the output raster. Click OK.
5. Create a table where you will calculate E and PA for each land cover type.
*
Open the Tabulate Area tool in the Zonal toolset in Spatial Analyst Toolbox.
*
Set bgfips as the input raster (zone data),
*
Set landcover as the input raster (feature class data), and
*
Set lctab as the output table. Put the output table in the geodatabase. Click OK.
6. Add fields to lctab for the proportion of the population expected for each land use in each block group.
ArcMap automatically adds the lctab to the data frame after the process is finished. Click on the List
by Source button at the top of the window to see the table if it doesn't take you there automatically.
The lctab table contains the areas of different land-cover types in each census unit. The areas of Low
Density Residential are stored in VALUE_1 field, High Density Residential in VALUE_2,
Commercial/Industrial in VALUE_3, etc. However, we need the proportion data not the area data.
*
*
Right-click on the lctab table and select open to open it.
Select add field from the options menu and add the following fields as double precision fields:
Total, P1, P2, P3, P4, P5, and E. Use the drop down menu to change the data type to double
precision.
7. Calculate values for the new columns you've added to lctab.
*
Right click on the Total heading and select Field Calculator.
*
In the window under "total =", enter the following formula:
VALUE_1 + VALUE_2 + VALUE_3 + VALUE_4 + VALUE_5
*
Then, click OK.
Because water/wetland has a relative population density of 0, we don't count the areas labeled as
water/wetland.
8. To calculate P1, use the formula (P1=) VALUE_1 / Total following the same procedures described
above. When done, continue to calculate P2, P3, P4, and P5 with their corresponding value fields.
9. Now, we have all the information needed to calculate E. Use the following formula to calculate E:
P1 * 15 + P2 * 65 + P3 * 5 + P4 * 10 + P5 * 5
10. Once we have the total area, proportion, and E data, we will use them to create raster maps of these
data.
*
3 of 5
Right-click on the BlockGroups polygon layer and select joins and relates, join.
08/03/2019, 13:35
http://web.pdx.edu/~jduh/courses/geog492w11/lab04.htm
*
Set OBJECTID as the field in BlockGroups that the join will be based on.
*
Select lctab as the table to join to and
*
Select VALUE as the field that the join will be based on in lctab. Click OK. (Say Yes if offered to
index table) When done, you will see the proportion data attached to the attribute table of
BlockGroups.
11. Follow the procedures described in step 2 to rasterize BlockGroups using lctab.Total and lctab.E to
create total and e raster maps.
12. Now, we can calculate the population per cell.
*
Open Raster Calculator and create the following equation, then, click OK. (Note: Don't type in
the equation, instead, use the Raster Calculator user interface to select and enter the terms of the
equation. Everything between a pair of square brackets in the equation refers to a GIS layer. The
layer names might be different depending on whether you use exactly the same names as listed in
the lab instructions. The layer names may also be doubled, such as [rdensity-rdensity])
[rdensity] * [popu90] * 133.058 * 133.058 / ([e] * [total])
*
Set the output raster to cellpopu.
*
You may also want to change the color ramp so that densely populated cells are dark and less
densely populated cells are light, a mapping convention.
13. Done! Before you begin to answer the questions below, it's probably a good idea to check if the
dasymetric mapping method works. This can be done by summarizing the population of census block
groups based on the dasymetric map.
*
*
*
Remove the join from the BlockGroups feature class
Select the Zonal Statistics as Table tool from the Spatial Analyst Tools, Zonal toolset in
ArcToolbox.
*
Set BlockGroups as the Input zone data,
*
Set FIPS as zone field,
*
Set cellpopu as input value raster, and
*
Create popucheck as the output table. Click OK.
*
4 of 5
You will compare the population of the northernmost census block group polygon (FIPS =
410670316034) in BlockGroups.
Compare the values of SUM in popucheck and POP90 in BlockGroups for the northernmost
census block group polygon.
08/03/2019, 13:35
http://web.pdx.edu/~jduh/courses/geog492w11/lab04.htm
Questions
1. The 1990 population of the selected census block group is 1547. What is the value you estimated based
on the dasymetric map?
2. What are the units of the values reported by the Tabulate Area tool in this exercise?
3. Print a choropleth map showing 1990 census block group population and a dasymetric map showing
1990 population. Include legends and a short text box for each map explaining your results.
4. The lab data set contains a polygon feature class called BeavertonCreek. Use it to estimate 1990
population of the subwatershed itself, and report your result.
5 of 5
08/03/2019, 13:35
Download