Ann Johnson
ajohnson@esri.com
Environmental Systems Research Institute (ESRI)
Objective:
Introduce Raster Data Formats in a concrete way including resolution,
generalization of data when imputing a grid data set, problems in conversion from raster
to vector and back to raster formats, data accuracy limitations in raster formats.
Time:
The
“Introduction” and completion of all except the tables should take approximately
½ hour. The Tables can be completed as a
homework assignment.
Materials Needed:
Paper copies of the two grid sizes and tables pages
(4 in all) for each student, Overhead transparencies of the 4 student pages and, 2 or 3
each transparencies of the blank Small and Large Grids with no features on them and felt
tip pen to be used with the blank pages.
Grade Level: High
school to university intro class
Description of Activity: Introduce what a Raster Data Structure is and
talk about it being a grid with pixels containing a value in each pixel. Put the Transparency of the Large Grid on the
overhead and then tell the students to pretend they are the computer program that will
encode what value should be in each pixel for the Large Grid page. Students should use their paper Large Grid page
and put a 0 if nothing is in a pixel, a 1 if it is part of the happy face, a two if it is
the “Stream” and a 3 if it is the rectangle.
The students will ask what they should do if any part of a line touches a pixel. Tell them they are the computer and must decide
what gets put in each grid. It takes about 10
minutes for them to fill out the page. After
they are finished, place a featureless Large Grid Transparency on top of the original
Large Grid Transparency except you should have already put a value for each pixel –
line them up with “tic” marks on each transparency. Discuss what the “recoded” grid looks
like. Removed the underlying Grid with
Features – can you see a resemblance to the original features? Next say you are now going to convert the new
recoded grid data into a vector data format. Put
a dot in each pixel that has a value other than zero.
Again students will ask where the dot should go and why. Then connect the appropriate dots. Changing between the two transparencies, compare
the two versions to the original version of the Large Grid.
Talk about how the data has been “generalized” and talk about what has
happened to positional accuracy (best case is ½ the diagonal distance of a cell). Ask students how it could be improved. They should say increase the number and reduce
the size of the pixels. Then change to
transparency for the Small Grid with features
on it and give the students their paper copy and ask them to repeat filling it out. Then you repeat what you did with the Large Grid
Transparencies. Compare the Large and Small
Grid “revectorized” versions. Discuss
the fact that accuracy is better, there is less generalization, but that this is at the
cost of more time – which is equivalent to memory used in a computer.
Optional Extensions: You can give them the Tables for the two Grid
sizes and ask them to fill it in for home work – this again reinforces the
time/memory use.. You can also ask them what
would be the difference in the tables if the data has decimals. Actually, ArcView would show them a
“Count” if data was integers, but not if it was floating point – too many
values.
Contact for More Information: Ann Johnson, ajohnson@esri.com