UW - Geog 460 - Autumn 2003
Due date: November 14 (yes, a Friday) in lecture
Learn how to perform neighborhood statistical analysis using
grid themes in ArcMap.
Examine the differences between neighborhood grid size analyses.
Completed answers to the questions and process diagrams of the steps performed. Blank diagrams are not provided for this. Please answer the questions (Q1-Q8) in your own words.
In this exercise you will develop your own ArcMap Project. This exercise will use two grid themes for Bainbridge Island.
The Neighborhood Statistics Tool:
The "Neighborhood Statistics" operation will create a result raster that carries out a given "statistic" on a specified "neighborhood" around each cell. The statistics supported in this version of ArcGIS Spatial Analyst include some dominance (maximum, mininum, etc.) rules and some contributory rules (mean (average), variety (diversity), sum, etc.). The neighborhood can be defined as a rectangle, a circle or a "doughnut" (nested circles), and dimensions of the neighborhoods can be given in cells or distance units. The values of the result grid belong to the center cell of each neighborhood, and reflect the function (statistics) applied to the neighborhood.
Establish a working folder on the "P" drive, open ArcMap, and create a new project [Follow the instructions on the "Saving your work" handout, if you don't remember them...]. Remember to enable the Spatial Analyst Extension and open the tool bar. Access the data set at "p:\geog460Aut03\Data".
Add the grid layer Cover1998.lyr from p:\geog460Aut03\Data
Select: Spatial Analyst -> Neighborhood Statistics.
Use layer Cover1998.lyr.
In the box, select Variety as the statistic, a rectangular neighborhood of 3 cells by 3 cells (the immediate neighbors of a cell), and hit the "OK" button. The result will be Nbr Variety of Cover1998. [Reset the Symbology (Properties>Symbology) to Unique Values and Add All Values. ]
Q1. What is the maximum value in the resulting grid? How did this value occur?
[There are very few cells with this maximum value. One can be found near the prominent western-most point of Bainbridge Island - "Battle Point".]
Q2. Explain how so many cells have a value of 1. (What can you tell about these cells?)
Do a second analysis of Cover1998 with a 5 by 5 cell neighborhood keeping the other parameters the same. Rename the result as nbrvariety5.
Q3. What is the maximum value, and how many cells have this
value?
Q4. What is the maximum value that you could obtain using this
Land Cover map as input?
Q5. Are the areas with high values in the first result also the
highest values in the second?
Add the layer bi_patch (found in p:\geog460Aut02\Data\ex5_data)
[It will run a lot faster if you copy this to your temporary working
space - use the procedure established in prior exercises...]
Select: Spatial Analyst -> Neighborhood Statistics.
Use the layer bi_patch you just added (This is a digital elevation
model derived from a LIDAR survey - aerial measurement using laser
ranging)
In the dialog window, select Mean as the statistic, a rectangular neighborhood of 3 by 3 cells, and OK. The result will be nbrmean3 (after you rename it to get rid of the embedded blanks).
Compare the maximum and minimum values of the result and the
source.
Q6. Would there be any difference? Why?
Select: Spatial Analyst -> Neighborhood Statistics.
Select the layer bi_patch again.
In the box, select Max as the statistic, a rectangular neighborhood
of 3 by 3 cells, and OK
The result will be nbrmax3.
Select: Spatial Analyst -> Raster Calculator.
Compose (you should do this by clicking on the list of layer names
and the operator buttons instead of typing it in...) the following
argument in the command box:
[nbrmax3] - [nbrmean3]
Click Evaluate.
Select and display the result of this operation.
Q7. How do you interpret the differences between the mean and
max operations?
Q8. What features in the landscape seem to have been detected
?
Version of 4 November 2003