[Openspace] GeoDa Question - Comparing Shapefiles

[Julia Koschinsky]

Dale,

If I understand your setup correctly, you are interested in a bivariate relationship between landform elongation and depth to bedrock. To assess this relationship in GeoDa, you would first have to decide on one base layer (landform polygon or well points) and then spatially join the attributes of the other layer(s) to the base.

So far, your question is non-spatial in the sense that you are not focusing on the relationship between elongation/depth to bedrock at location i and neighboring locations but only looking at location i. In this case, there is no need to separate the file into three subsets since you can use non-spatial tools in GeoDa (e.g., bivariate scatterplot with linking and brushing or the regular non-spatial regression). 

If you're also interested in elongation and/or depth to bedrock at location i and neighboring locations, then you would need to create a spatial weights matrix that defines your neighborhood structure. If the three sub-regions are not connected, you could subset the analysis. If they are connected, you could use, e.g. distance weights, to avoid islands. Once you have the weights, you could apply the spatial tools in GeoDa such as global and local Moran's I and spatial regression, e.g., by testing for local clusters of similar elongation values or depth to bedrock or by regressing y (elongation at location i) on Wy (elongation  at neighboring locations) and X (depth to bedrock at i) and possibly WX ((depth to bedrock at neighboring locations).

Julia

---- Original message ----
>Date: Tue, 30 Oct 2007 10:25:33 -0400
>From: "Dale Hess" <dalehess at eng.buffalo.edu>  
>Subject: [Openspace] GeoDa Question - Comparing Shapefiles  
>To: openspace at sal.uiuc.edu
>
>Hello,
>
>I am new to GeoDa (and spatial statistics for that matter...) and I'm hoping
>that someone in this community can point me in the right direction.  I
>appreciate any help or advice that stems from this request.
>
>I have a polygon-shapefile that consists of ~6500 digitized landforms spread
>over a (roughly...) 120km x 55km rectangle.  The length (L) and width (W) of
>each feature have been measured.  I have divided L by W to determine
>the elongation of each feature and have assigned it as an attribute.  I also
>have 3 separate point-shapefiles that consist of well locations, however
>each of these files represent a much smaller area (largest is 35km x 9km)
>within the region covered by the polygon shapefile.  I have identified and
>assigned depth to bedrock as an attribute for each point within these 3
>files.
>
>I would like to determine if there is a relationship between depth to
>bedrock and landform elongation.  For example - do the very elongate
>landforms form in areas of shallow depth to bedrock?  Is this possible to do
>in GeoDa?  If so, will I be required to subset the polygon-file to match the
>coverage area of each of the 3 point-files?  If not, do you have any
>suggestions for software that would be suitable for performing this task?
>
>Your help is very much appreciated.
>
>Cheers,
>
>Dale Hess
>University at Buffalo
>_______________________________________________
>Openspace mailing list
>Openspace at sal.uiuc.edu
>http://sal.uiuc.edu/mailman/listinfo/openspace