G 424/524 GIS for the Natural Sciences
D. Percy
e-mail: percyd@pdx.edu

Assignment 5

Due March 3, 2009

• Statistical Analysis of Point Data

You have a choice, either do this exercise using the existing landslide data that you have already been using all term, or analyze earthquake distributions in Oregon. Your choice, either way zoom in on an area that has some point density to it, then draw the grid and do the Poisson analysis.

Optionally, you can also see how the choice of grid size for analysis can lead to different results, known as the MAUP.

• Download the earthquake database from http://nwdata.geol.pdx.edu/NW-Quake/
• Load it into ArcMap (this will require saving it in a format that can be added as an X-Y data set, a task that should be familiar by now). Play with it. Do you see any patterns? Try different ways of looking at it with graduated symbols, based on the different fields, like magnitude or depth or date. Get a feel for what your data set is like!
• Put in some context... Add the States from the ESRI data (another familiar task by now). Is everything plotting correctly? :-) Is Percy tricking you again? Look at longitutude and pay attention to sign (positive or negative). I'll show you a quick way to deal with this in class! (or just use the field calculator and multiply by -1)...
• Once you have your earthquakes sitting in the correct part of the world, save them as a shapefile so that you can geoprocess them. Event themes have a sort of "second-class" status in Arcmap, and certain operations don't work on them. I consider this an undocumented bug! Use Data->Export Data
• Clip the eathquakes theme so that you only have the Oregon set (just like in assignment 3!). If you still have Oregon from assignment 1 you can use that, otherwise add States, select Oregon, and choose Use selected features only.
• Use the GenerateGrid scripts (arcscripts.esri.com) to overlay a grid (quadrats). You can try a couple of different sizes. In the VB editor, use control-H (replace) to replace the word "esricore" with "esriGeometry", this updates the script to run in ArcMap9 (http://forums.esri.com/Thread.asp?c=93&f=992&t=60413&mc=5)
• Spatial Join the earthqaukes and the Quadrats. Think about the One-to-Many problem. Which theme gets to receive the attributes? Note: the script that makes the grids fails to define a projection for the shapefile. Trick: If you export the selected grids that intersect the state, you'll define one on-the-fly...
• Count how many points there are in each grid by opening the Join_Output table, right-clicking on Name, and choosing Summarize. Note that you can calculate statistics in this screen, like average magnitude or depth per grid...
• Now do a more formal quadrat analysis, detailed instructions are here: quadrat analysis.
• Optionally, try at least 2 different grid sizes, like 20 x 20, and 30x30. What differences do you see?
• Optionally, try different subsets of data, divided out by time, intensity or depth for example. (Use Selection-> Select by Attribute to define subsets). Are the spatial patterns the same?
• Extra points for finding a subset of the data, like maybe Klamath, that FAILS the randomness test.
• There is a lot going on in this assignment! Just pay attention to the big picture, and the details should fall into place.

Turn in the following: a map or maps of your grid (s)) with earthquakes overlaid, the poisson spreadsheet results, and a description of what you did, be sure to mention why you did it (and not just because "Percy said").