The goal of this exercise is to learn about some of the spatial analysis capabilities in ArcGIS.
Part 1: Nearest Neighbor Analysis
Do the exercise called "Nearest Neighbor Analysis" on the course Blackboard site under Assignments.
Part 2: Raster Analysis
You will need to know how to use ArcGIS Spatial Analyst to complete this exercise. Review the ESRI Virtual Campus exercise called "Learning ArcGIS Spatial Analyst." The goal of this exercise is to create new spatial layers that might be useful in predicting some health outcome.
Section 1: Interpolating a Surface
Download the following dataset to practice with these tools- http://www.unc.edu/~emch/gisph/CAdata.zip and unzip it. First use the IDW interpolation tool to interpolate a surface of the NastyWater variable. Use the mask tool and mask the result to the California boundary area. Print the resulting map. Where is the water nastiest in California based on this interpolation?
Section 2: Calculating Densities
Using the CAdata calculate the density using the population field and a 50,000 meter search radius. Print the map. Now do the same using the clinic field. Then using the raster calculator, calculate a map that is the ratio of clinic density to population density. Are there some areas that don't have a lower ratio than others?
Section 3: Reclassification and Vector Conversion
Reclassify the population density raster into 3 classes: high, medium, and low based on natural breaks in the data. Then convert the dataset into vector, make a map that labels the different density categories, and print the resulting map.
Part 3: Descriptive Spatial Statistics in ArcGIS
Go through the free ESRI Virtual Campus Training Seminar called "Introduction to Spatial Pattern Analysis." You won't need a code for this training seminar. After completing this seminar you should understand the following concepts and methods: Average Nearest Neighbor, Getis-Ord G, Ripley's K, Moran's I, Anselin's Local Moran's I, among others. Then, go through the course called "Exploring Spatial Patterns in Your Data Using ArcGIS 10." You will need a code for this course. After completing this exercises you should understand the following concepts and methods: mean center, median center, directional distribution tools (distributional ellipse), histograms, normal QQ plot, Voronoi map, semivariogram, spatial autocorrelation, interpolation.
Download the following dataset to practice with these tools- http://www.unc.edu/~emch/gisph/CAdata.zip.
Section 1: Measuring Geographic Distributions
In ArcGIS under the "Measuring Geographic Distributions" tools of the "Spatial Statistics" toolbox run the following programs using the towns point layer: 1. Central feature, 2. Mean Center (without weighting), 3. Mean Center (weighted by the population field), 4. Directional Distribution (Standard Deviational Ellipse) using 1 SD, 5. Standard Distance using 1 SD. Make a map or maps that labels all of the features that you created. Then describe what each one means.
Section 2: Analyzing Spatial Patterns
In ArcGIS under the "Analyzing Spatial Patterns" tools of the "Spatial Statistics" toolbox run the global Moran's I program using the NastyWater variable and print the result. Is NastyWater spatially autocorrelated in California? If so, how much and what does that mean? Calculate the average nearest neighbor distance on towns. Are the towns clustered or dispersed? What is the average nearest neighborhood value and Z-score? Calculate the Getis-Ord General G Index on NastyWater. What is the value and Z-score? Is there high or low clustering?.
Section 3: Mapping Clusters
In ArcGIS under the "Mapping Clusters" tools of the "Spatial Statistics" toolbox run the program called "Hot Spot Analysis: Getis-Ord Gi*" for the NastyWater variable Use a distance band of 10,000. The result is a table that has a new field in it called Gi10000. Map that out with the small numbers (cool spots) in blue points and the large numbers (hot spots) in red. Does there seem to be spatial clustering of those hot spots? Then calculate the Anselin's Local Moran's I values for the NastyWater variable using the Town layer. Map the local Moran's I value using red and blue? Where is there high local spatial autocorrelation of NastyWater and what does that mean?
Part 4: Inferential Statistics in ArcGIS
Go through the free ESRI Virtual Campus Training Seminar called "Regression Analysis Basics in ArcGIS 9.3." I suppose they'll update this at some point to version 10 but the concepts are all there in this seminar. You should know the difference between OLS and GWR regression when you've completed this seminar.
Part 5: Network Analysis in ArcGIS
Go through the free ESRI Virtual Campus Training Seminar called "Introduction to ArcGIS Network Analyst." I suppose they'll update this at some point to version 10 but the concepts are all there in this seminar. You should know what network distance is after you've completed this exercise. You should also be able to calculate service areas and closest facilities.
Lab Deliverable Summary: Print out all of the outputs, put your name on them, and give them to the instructor. They include: Part 1: printout of results of nearest neighbor analysis and paragraph describing distribution. Part 2: a map of the interpolated nasty water surface and a short description of the spatial pattern, a raster map of clinic density/ population density and short description of the spatial distribution, a reclassified vector map that labels the different density categories. In Part 2, organize the deliverables by the three sections that are listed. Part 3: map or maps showing the 3 geographic distribution summary statistics and a description of what each means, Moran's I result printout and description of what the result means, average nearest neighbor calculation result and description of what it means, Getis-Ord General G Index result and description of what it means. In Part 3, also organize the deliverables by the three sections that are listed. In Part 4, write a paragraph describing a hypothetical GWR application and what you will get out of using GWR versus OLS regression. In Part 5, write a paragraph describing a hypothetical public health network analysis application.