Working with Raster Grids and Map Algebra

Working with Raster Grids and
Map Algebra
Katherine Paybins, USGS
Basic Ingredients
• Grid of data, such as the National Land Cover
dataset, from 2006, and a point, line, or polygon
feature dataset.
• Area of interest, such as the Coterminous US, or
the boundary of WV, or drainage areas in WV
• Topic of interest, such as coincidence of dam
locations with land cover that is related, like
water bodies, wetlands, and emergent woody
• Well-defined scale of interest.
Area and topic of interest, and scale
• NLCD data is available for the NE, NW, SE, and
SW, from 2006 imagery.
• The grid cells are 30X30 meters.
• The land use or cover is classified from
LANDSAT satellite images
• National Inventory of Dam locations is a point
coverage of all types of dams, ranging from
< 20 cubic feet of storage, up to >2.5 million
cubic feet of storage.
Step one
• Increase processing speed by zeroing in on
your area of interest.
• Select a processing area in the options
• open the toolbox, and select raster
• Set the new extent and the name of the new
Setting processing area
Next- select the type of land cover you
are interested in
• In the toolbox, open the Spatial Analyst tools,
and then the Extraction toolset- you want to
extract by attributes of the raster
Now, convert point coverage to a grid
of 1 = dam, nodata = nodam
Open the conversion tools in the ArcToolbox
Open the “to Raster” directory
Choose feature to Raster
Make sure to reset the cell size to equal the
cell size you wish to compare the data to, so
for this example, use 30X30 meters.
Now to add the grids together with
Raster Math
The result
Some data analysis
• The WV_dam_grd has 645 dam locations cells = 1
in the grid
• The wv_water_dam combination results in values
of 12 and 91. In other words, the addition of the
two grids found no dams coincident with
emergent woody wetlands
• Additionally, the raster combination of water and
dams has only 284 cells with data, so we can tell
from this analysis that 2/3 of the dam locations
are not coincident with the NLCD water or
wetlands designation
Other possible grid combinations for
• Extract urban areas from the NLCD or from
USGS DRGs in raster format using the values in
the raster.vat to produce an impervious
surface raster
• Calculate change in forest or farm areas using
multiple years of NLCD data
• Compare multiple years of land cover data,
and animate in ArcScene the change of of land
cover over time.
• Know your scale– a 30 meter grid is admittedly coarse for
the analysis within, for instance a county. It is well-suited
for state-wide or US-wide analyses.
• A 30 meter grid of land cover may be your only resource
other than ortho images, though Aaron Maxwell earlier
illustrated that efforts are ongoing to improve land cover
• Some ground-truthing will add certainty in your results,
• Use the Xtools Pro toolbar function to see a photo of the
area of concern (Go to Google Earth function)Slide 21
• The NLCD data are good for more than just an interesting
and pretty background

similar documents