GEOG 566

         Advanced spatial statistics and GIScience

April 28, 2017

Tutorial 1: Calculating Global Moran’s Index on Point Data

Filed under: 2017,Tutorial 1 2017 @ 2:19 pm

A quick summary of my project and objectives: I am using a subset of the FIA (Forest Inventory and Analysis program run by the USFS) dataset, which is contained in a .csv file. I want to assess the understory vegetation response to ponderosa pine canopy cover and other environmental factors, and I am using USFS plot data to simulate the process I will be using on my own dataset in the future. The dataset currently looks like this:

Obviously there are multiple variables associated with each plot, but for the sake of this exercise, I will only be assessing canopy cover and total vegetation cover. Because my dataset is not a raster, nor is it regularly gridded, I decided not to attempt a semi-variogram. I was also unable to kick my ArcMap habit for this exercise. I decided to run a Global Moran’s Index assessment on the points in ArcMap.

The first step of this process was to load my .csv into Arc. This can be accomplished by adding the file as X,Y data. However, you can’t do much with X,Y data for our purposes, so I exported the X,Y data as a shapefile. This created an attribute table holding all the same information as the .csv file. I also added a boundary of Deschutes County, which is where my subset of data is located. I used the Project tool to make sure both the points and county boundary were properly oriented on the landscape, using the WGS_1984_UTM_Zone_10N coordinate system.

From here, the process was very simple. I used the search tool to seek out a spatial autocorrelation tool, which brought up the Global Moran’s Index. I selected my points layer as the input, and selected canopy cover and total vegetation as my input fields (in separate trials).

The reports showed the levels of clustering for each variable. The reports are shown below.

Canopy cover spatial autocorrelation:

Vegetation percent cover spatial autocorrelation:

I found that the variables had very different levels of clustering. The canopy cover variable had essentially no clustering. The Moran’s Index value was -0.034, with a p-value of 0.57, meaning the level of clustering was non-significant. However, the vegetation cover was highly clustered, with a Moran’s Index value of 0.0496 and a p-value of 0.00028.

This method was useful in giving me a different perspective on the data. It is difficult to simply review a spreadsheet and try to make any meaningful interpretation of the numbers. While the tool did not give any sort of visualization, it allowed me to imagine how the vegetation might be distributed across the landscape. The results mean that canopy cover varies across the focus area randomly, so stands with thin or dense canopies can be found just about anywhere. However, stands with dense understory vegetation are probably located in similar regions, as are stands with thin understories. This might imply that other environmental factors have a strong influence on how densely vegetation grows over a wide area.

Print Friendly, PDF & Email

No Comments

No comments yet.

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.

© 2019 GEOG 566   Powered by WordPress MU    Hosted by