Week 9: Corridor Analysis

Corridor model

This week's assignment involved creating cost paths and corridors.  Ultimately, a model for black bear movement between two national park units was created.  Cost paths and corridors are used to predict the pathways that are least costly to construct (in the case of roads and utilities, etc.) or traverse (in the case of wildlife, utilities, etc).

The corridor output at right shows a predicted black bear corridor.  The parameters used to create this corridor were elevation, land use (not shown), and roads (not shown).  Bears prefer certain elevations, certain habitats, and avoid roads.  So a suitability analysis was performed to find the most bear-friendly areas.  Then the corridor analysis was use to create the best path between the two units.

Week 8: Network Analysis

This week's assignment dealt with network analysis.  At right is a service area analysis depicting the change of catchment coverage for Austin Community College after the closure of the Cypress Creek Campus.  The service area consists of several layered areas within a specified range.  It is clear that coverage changes, and the ultimate effect of this change was further assessed using demographic data.  For example, over 1,000 college student aged citizens who were closest the the closing campus must now attend a different campus, thus increasing the student-averaged travel time to any campus.

Week 7: Suitability Analysis

Final map product outputs

This week's assignment involved applying suitability analyses using several criteria and different methods.  Both vector and raster methods were applied for a straight-forward 'binary-type' suitability analysis, then a weighted overlay analysis was performed.  Ultimately a comparison was made between two different weight schemes using the overlay method, and the final outputs were compared (see map product at right).

First, a raster grid displaying a landscape of suitability for each criterion was made.  Generally, places on low slopes, far from rivers, close to roads, on particular soils, and on particular land cover types were considered more suitable.  Then, the five criterion themselves were weighted.  The map compares the equally weighted criteria output, and the unequally weighted criteria output.  In real applications, weight assignment is determined by accepted research, personal observation, or popular opinion within the decision-making community.  Clearly in the map above, slope has a large influence on the suitability of particular areas.

Week 5: Crime Hotspot Analysis

This week's lab involved creating several different Crime Hotspot Maps and comparing them in their predictive power.  We created hotspots using Local Moran's I, Kernel Density, and Grid Overlay from 2007 burglary data.  Then we compared the number of 2008 burglaries within these 2007 hotspots.

 I argue that the best crime hotspot predictive method in this scenario is using Kernel Density.  I base this on the observed highest crime density of 2008 burglaries within the 2007 hotspots. This is category (above) that I view as the best metric of predictive power because it takes into account hotspot total area, and thus would provide efficient preventative resource allocation.

It is clear that the Grid Overlay hotspot contains the most 2008 burglaries, therefore it is certainly a suitable predictive tool.  However, the total area is >65km² and may be too large to effectively patrol by police.  With unlimited resources (i.e. police officers/vehicles), it would be feasible to use this model as area to patrol. In a limited resource situation – which is what is most often the case – higher priority areas must receive preferential resource allocation.  Thus, the Kernel hotspots.

The Local Moran’s I hotspots are large, but contain the fewest amount of 2008 burglaries.  The large area appears to be a result of extreme density regions.  By visual assessment, it appears that areas between several clusters, which don’t look particularly dense themselves, are influenced by adjacency (see screenshot below).  It appears that the red area toward the top is in between two hotspots.

Week 4: Damage Assessment

Damage Assessment of structure on East New Jersey
coast after Hurricane Sandy

This weeks lab involved damage assessment of structures near the New Jersey Coast after Hurricane (super storm) Sandy in Fall 2012.  Damaged structures in this case were homes and businesses that were inundated or wind-ravaged.  We compared pre- and post-Sandy aerial imagery of our study area and visually determined the severity of damage (see image top-right; description below).  Then, we regressed the severity of structure damage on the distance from the coastline in intervals of 100 meters (e.g., 0-100m from coastline, 100-200m, etc.; see table below).

I first coarsely assessed the entire affected area using pre- an post-Sandy aerial imagery.  This visual assessment included areas outside the delineated study area that were highly affected, as well as areas that appeared almost entirely unaffected.  This was done to make my damage assessment more objective.  Based on this broad-scale approach, I determined that the study area was one in which structure damage was highly variable – some properties were destroyed completely, while some looked unharmed.  Further a majority of destroyed structures were located in the easternmost region of the study area.

The general process of identifying the structural damage for each parcel was a left to right sweep of each block.  This allowed sufficient detail without taking hours to process a small area.  I primarily used the ‘slide’ effect tool to compare the two aerial images of the pre- and post-Sandy study site. If a structure was clearly moved from its original foundation or was leveled, then I chose to label it as destroyed.  Some structures were absent all together, those were labeled as ‘destroyed’ as well.  If a structure had major collapse, which was noticeable from aerial imagery by debris or a change in shape, then it was labeled ‘major damage’.  'Minor damage' was subjectively decided to describe a home that was generally surrounded by other severely affected homes, but was not obviously damaged from imagery.  ‘Affected’ was given to any home that appeared to have debris in the yard, which I presumed to be material from the structure.  A structure was labeled ‘no damage’ if it appeared generally identical in shape to the pre-Sandy imagery, and was surrounded by other homes which appeared unharmed.  This was based on the assumption that adjacent homes protected those upwind and uphill of the storm.

Structural Damage Category	Count of structures within distance category
	0 – 100 m	101 – 200 m	201 – 300 m
No Damage	0	1	7
Affected	0	9	24
Minor Damage	0	16	6
Major Damage	0	8	4
Destroyed	12	6	4
Total	12	40	45

Week 3: Flood Zone Analysis

This weeks assignment was to perform coastal flood zone analyses.  Specifically, we created a map of the District of Honolulu after a modeled 3ft and 6ft sea level rise (SLR) scenario, then analyzed the social impacts of the flooded area using demographic data from the US Census Bureau.  The map above shows the area flooded by a 6 ft sea level rise and the general population density in Census Tracts.  It was created by first demarcating the area that would be flooded using the Less Than tool with the original Raster DEM.  Then the depths were calculated by using the Minus tool between the new raster and the the original DEM.  The high population density near the coastline is noticeable in this map.  Below is a table of demographic data extracted from Census blocks (finer scale than Tracts).

Variable	Entire District	6ft SLR
		Flooded	Not Flooded
Total Population	953207	60005	893202
% White	20.85%	29.58%	20.26%
% Owner Occupied	56.77%	38.13%	58.02%
% 65 and Older	14.53%	17.04%	14.36%

The table shows demographic data for the areas flooded and not flooded by the simulated sea level rise of 6ft in Honolulu District, HI.  For comparison, there is demographic data for the entire Honolulu District.  This data is based on the 2010 U.S. Census.  All homes affected are located within close proximity to the coastal in Honolulu District. Clearly the populations affected by the 6ft SLR flood zone are different than the populations not affected by them.  Specifically, there is a greater proportion of persons 65 years and older and persons who describe themselves as “white” in the 3ft and 6ft SLR flood zones compared to the non-flooded zones.  Conversely there is a smaller proportion of owner occupied housing. 

These data can be examined in the context of social vulnerability.  While it is unclear that there is any social or political marginalization in this area due to racial disparity, areas in the United States with a high proportion of persons who describe themselves as “non-white” may be more vulnerable to environmental catastrophes (e.g. Hurricane Katrina; Cutter & Emrich 2006 as cited in Shephard 2012).  While the proportions described here for “white” persons is low compared to the national average, this is to be expected on a Hawaiian island to which “white” persons are secondary colonists.  The lower proportion of Owner Occupied housing in the flooded zones may be a result of seasonal/vocational visits by the owners, or some sort of rental-tenant situation.  This is important for social vulnerability because renters may not know local evacuation routes, or may not have access to personal vehicles for an evacuation situation.  Further, non-owner occupied housing is disproportionately uninsured, which could cause issues during flooding situations.  Lastly, the higher proportion of individuals 65 and older is significant because they may require assistance in an evacuation scenario, increasing their vulnerability to harm.

It should be noted that the 6ft flood zone affects a more racially diverse group of people than the 3ft flood zone.  This is observed in the decrease in the proportion of “white” persons affected from the 3ft to 6ft flood zone scenarios (from 36.79% to 29.58%).  It could be speculated that the greatest density of “white” persons is near the coast of Honolulu.

Week 2: Watershed Analysis

This week we focused on flow and watershed analysis.  While I was familiar the concepts behind these analyses (e.g. physics and geology), the process of running the appropriate procedures in ArcGIS was new.  We were given a Digital Elevation Model of Kuauai, HI and lead through the process of creating a stream network.  This involved: processing the raw DEM to fill any sinks, generating a Flow Direction Raster, Flow Accumulation Raster, and ultimately analyzing the output against the actual streams and watersheds as designated by the USGS.

Above is my comparison of a single watershed on Kuauai, the Lumahai River Watershed, to the USGS designated watershed boundaries and stream network.  This is a large watershed on the northern portion of the island, which at its terminus is a single river emptying into the Pacific Ocean, just west of the Hanalei Bay proper.  The two large maps compare the modeled Lumahai watershed and streams to the actual features, using a 200 cell threshold for the stream delineation.  The modeled watershed is significantly different in appearance near the northeast coast and the modeled stream system is relatively truncated compared to the actual stream system.  This is likely an affect of the threshold cell size, which could be decreased to better match the true stream network.