In review of:
Wing, M. G., & Tynon, J. (2006). Crime Mapping and Spatial Analysis in National Forests. Journal of Forestry, 104(6), 293–298. doi: 10.1093/jof/104.6.293
Concurrent to the pillar of crime analysis within GIS, the primary objective of this paper is, “to examine whether patterns exist in the spatial distribution of crime and to explore the relationship of patterns to other landscape features,” and, “how crime violation types were distributed at various administrative levels,” within the context of forestry management (Wing & Tynon , 2006, p. 294). At the time this was written, crime analysis had been carried out only within urban spaces, despite the fact that crimes happen everywhere, so the urban crime analysis model was applied to national forests within Washington state and Oregon to analyze crime patterns and start building a preliminary model for forest crime management. As forestry law enforcement officers are often geographically isolated, they rely on support from a variety of other types of national and regional law enforcement and security protection agencies, who are often located an hour or more away. Because of the many different agencies tracking crime in forested regions, confusion, inefficiencies and inaccurately marked or geographically logged crimes are prevalent, and incidents may often go unnoticed, underreported or reported elsewhere. Thus, a geographical model identifying hotspots of crimes could improve geographic profiling and crime predication to enhance the spatial displacement of officers to mitigate crime.
This study used ArcMap and CrimeStats, much like lab 3 did, to calculate nearest neighbour indices and generate a quartic kernel density estimation to identify and visualize hot spots. This was done for three regions: Forest Service Region 6, which includes all national forests within Washington and Oregon; Siuslaw National Forest, located on the coast of Northwestern Oregon; and the Oregon Dunes National Recreation Area (ODNRA). The data for this study came from the Law Enforcement and Investigations Attainment Reporting System (LEIMARS), which is the largest digital repository for reporting crime incidents and the only source for crime statistics for nationally protected forested areas in the United States which also contains spatial features for geographic analysis. So, crime incident shapefiles from the years 2003 and 2004 were derived from LEIMARS, and other reference features like roads, boundaries, hydrology, and elevation were also used but their source was not specified. A digital orthophoto of Region 6 was used to better visualize land use around hot spots, but this source and tile were not specified either. All the regions were found to have a significant nearest neighbour indices, and ODNRA was found to have some of the greatest concentration of hot spots, despite being the smallest in land mass per square kilometer. Ultimately, crime hot spots were found to be adjacent to major population centers and main transportation networks, while popular recreational spots and trails were found to have a high proportion of crimes as well, and all regions had the highest proportion of misdemeanour type crimes by far.
I think that the analytical procedures carried out were very appropriate for this analysis. The only critique I would have of their analysis is to use a triangular instead of a quartic kernel density estimation because the hot sports for Siuslaw National Forest and ODNRA, which were further examined at multiple scales, were found to have a linear distribution and this function might’ve provided better insight in the visualization. The evidence to support their claims that these crimes happened near cities and transportation networks were highly supported through their orthophoto, providing a more realistic visualization. Also, they adjusted the hot spots from point data to a vector quadrat grid for larger scale visualization of hot spots, so they were more finely defined, which I think is good practice. Finally, I liked how in the discussion they noted that hot spots might be near these areas because they are more accessible for patrol, meaning that fewer crime points may not mean fewer crimes but few patrols. They also stated that the data, though was reportedly the best possible, was very flawed and that it was not consistent in similar violation codes were reported at different violation levels, so analysis like this could homogenize severity of punishment for similar crimes.
Overall, I thought this paper was really good and I would give it a 9.5/10. After doing the lab I could imagine carrying out this analysis and there are very few things I would adjust, and I think their reasoning and validity is strong.