Kirsten Vinyeta's GEOG 582 Blog

Lab Assignment 3: Search and Rescue

Q1 : What does the DEM layer show? What is the cell size and how do you know? Why is the cell size important?

The DEM provides elevation values over a topographic surface. The cell size in ~32.8×32.8, it states so in the layer properties. Cell size is important because it gives you a sense of how much detail and landscape differentiation is going to be recognizable in the DEM. The larger the cell size, the lesser the detail. The smaller the cell size the greater the detail, but also the larger the file size and processing time.

Q2 : Why reclassify water in this way? How will the value of the water affect the resulting cost surface?

We are assuming that water is unlikely to be traversed by the mushroom hunter and therefore we give all water a uniform, high value (100) representing high cost. The value of water bodies will dramatically increase the cost associated with traversing any water-covered area.

Q3 : What other data could be used to generate a cost surface? Do you think our slope/water cost surface is the most realistic representation of how difficult it is to travel from one grid cell to another? Why or why not?

A more detailed cost surface would also include vegetation density, which can greatly affect the speed and effort of travel, as well as visibility. I think slope definitely is a good measure of travel difficulty, but water can be a bit misleading, because it is more likely that a person will cross a shallow stream than ascend a steep slope, and yet we assign all water a high cost. I think a slope/vegetation density cost surface would more accurately represent travel difficulty, and while water also is a strong indicator, it would be better if we could filter out shallow streams and rivers that may be easily crossed.

Q4 : What does the attribute table of the Observer Points raster tell you? How would you determine which cells can be seen from which lookout?

The attribute table tells me the amount of cells that can be seen from Lookout 1, the amount of cells that can be seen from Lookout 2, the amount of cells that can be seen from both, and the amount of cells that can be seen from neither of the lookouts. The lookout shapefile attribute table numbers the lookouts 1 and 2, and the observer points attribute table has columns titled obs1 and obs2 which reflect the observation range possible from the respective lookouts.

Q5 : Based on the results of your observation points analysis and the least cost path you generated, which lookout would you recommend the Forest Service use to the mushroom hunter? Why?

I would pick lookout # 1 (the easternmost lookout) because it offers a much larger area of observation (2,699,816 cells, versus 504,912 cells for lookout 2), and because its areas of observation includes areas closer to the least cost route.

Q6 : With reference to each map, explain which areas you would prioritize in searching for the mushroom hunter. Make sure to also explain the steps you took to conduct your analysis, including any limitations or caveats of the techniques you used.

I would first prioritize along the least cost path (see Figure 2). If unsuccessful, I would follow some lesser cost routes that don’t necessarily lead to Nash Lake, but that make for easier walking. At one point the East-West canyon that comprises the beginning of the least-cost path curves Southward toward lookout 1 (see Figures 1 and 3), so the mushroom hunter may have just followed the canyon South. I would prioritize lookout 1 for viewsheds (see Figure 3), given that it allows for visibility surrounding the end of the least cost path approaching Nash Lake, and also given that its viewsheds cover a lot more ground than lookout 2.

The main limitation of the techniques used is that our analysis lacks vegetation data which limits the accuracy of our least cost path. It is also difficult to predict human behavior in such a situation, limiting our ability to make anticipate the mushroom hunter’s decisions when lost.

Lab Assignment 2: Mapping Campus Safety & Security

Q1: Why does it matter what datum your waypoints are recorded in?

If recorded in the incorrect datum, once we imported the data into QGIS, the location of the lights would likely be inaccurate in relation to other data.

Q2: What coordinate system are the campus data layers in? What coordinate system is the GPX file you added in?

Campus Data:

+proj=lcc +lat_1=42.33333333333334 +lat_2=44 +lat_0=41.66666666666666 +lon_0=-120.5 +x_0=1500000 +y_0=0 +ellps=GRS80 +towgs84=0,0,0,0,0,0,0 +units=ft +no_defs

GPX file:

+proj=longlat +datum=WGS84 +no_defs

Q3: What distance did you use to buffer the call boxes and why?

I used 20 ft. because I feel like outside of a 20 ft range a crime victim in a state of distress would not have the ability to notice and/or physically reach the emergency call box in time.

Q4: What operations would you perform if you wanted to identify areas of sidewalk that were outside both call box and streetlight buffers?

You would merge the Lights_Buffer and EmergencyCallBoxes_Buffer shapefiles and then use the difference tool to determine the difference between the original sidewalks shapefile and the newly merged shapefile combining the two buffers.

Q5: What areas seem adequately covered by both streetlights and emergency call boxes? Where would you recommend the University install additional lights and call boxes?

Our lights data was incomplete because we did not gather GPS points for lights East of Susan Campbell Hall. As such we have a large pocket in the lawn area surrounding the Pioneer Mother with missing lights data for which I cannot make an assessment. In regards to the rest of campus, areas that are especially well served by lights and emergency call boxes include the Southwest section of campus (including the Education , Frohmayer Music, and Clinical Services buildings), the areas surrounding the Mathew Knight Arena and Jaqua Center, and the areas surrounding Cascade Hall, Onyx Bridge and Lawrence Hall.

Areas that could benefit from emergency call boxes include the sidewalks on Johnson Lane, the open area North of Lillis Hall, the sidewalks between Fenton and Deady Hall, the area North and East of Straub Hall, the area West of Willamette Hall, and the area South of Hayward Field. I would also recommend improving lighting on the sidewalks South side of Chapman Hall, South side of the Schnitzer Museum of Art, West and South side of Straub Hall, between Fenton and Deady Hall, surrounding McArthur Court, and in the open quad that features the Pioneer Man.

Q6: What are some issues you see in the available crime location data? What compromises or judgement calls do you have to make when entering data of this nature? Why do you think this is so for this particular data set?

The crime location data does not specify whether the crime occurred indoors or outdoors of a given building, nor does it specify which side of a building the crime occurred on, instead just listing the building name. I suspect this is done purely for time-saving purposes in reporting. This limitation creates a map in which the crime symbol is centered on the building and therefore gives us little indication as to whether lighting or emergency box access truly played a role.

In addition, some crimes, particularly those related to sexual harassment, often do not specify the campus location, instead just generally stating “On Campus” or “Campus Property.” Perhaps this is done to protect the anonymity of the sexual harassment victim. These line items had to be excluded from the map as there was no way to join them to a specific campus location.

Q7: What information would make your report more complete? Can you think of any other data that might be important to include that you did not or were unable to? Include a discussion of the shortcomings of using buffers alone to identify areas of concern.

Having more specific information about the location of crimes (such as which side of the listed building, or any level of specificity for sexual assault-related crimes) would make this analysis more complete. There were also cases in which it was unclear whether a theft or offense occurred inside or outside of a building (for example, a laptop theft is more likely to have occurred indoors than outdoors in which case lighting would have been irrelevant). It would have also helped to record GPS points for all the missing lights, which we failed to do due to weather and time constraints (we only recorded 20 of the missing lights).

One constraint of using buffers and calculating the difference on the sidewalk to determine areas of concern is that this excludes the analysis of lawn areas that may also be traversed by foot. Additionally, buffers do not make clear how obscured from plain sight certain areas of campus are, be it by walls, vegetation or other barriers that could enclose a given space or make it easier to conceal oneself.

Lab Assignment 1: Projections and Classification

Q1 : What information does the Source tab provide about the states shapefile?

It specifies the folder where the shapefile is saved in the R: drive.

Q2 : What coordinate system is this layer in? Is it a geographic or projected coordinate system? What is the difference between these two types of coordinate systems?

The layer is in GCS_North_American_1983, a geographic coordinate system. Geographic coordinate systems georeference locations using latitude and longitude on a curved earth, and represent the most comprehensive georeferencing system. Projected coordinate systems operate on a flattened surface of the earth in order to better translate into 2-dimensional media such as paper maps. Projections often use a cylindrical, planar or conic surface to flatten a portion of the earth’s surface. Different projections work best for different map purposes and for different regions of the earth, though they all come with inherent distortions.

Q3 : Compare the different projections. How does the shape of the continental US change with each projection?

Albers Equal Area Conic: The continental US appears horizontally shorter than Plate Carree projection, and smaller overall than the Mercator projection. It appears roughly the same size as the Robinson projection but without Robinson’s slant. In the Albers projection, the country’s upper boundary has a curvature that does not appear in the other three projections.

Mercator: The continental US becomes much larger overall than in the Albers Equal Area Conic projection, and the country’s upper boundary is straight (as opposed to the curvature visible in the Albers Equal Area Conic).

Plate Carree: The continental US remains horizontally as long as in the Mercator projection, but narrows vertically. The country’s upper boundary is straight (as opposed to the curvature visible in the Albers Equal Area Conic).

Robinson: The continental US becomes horizontally shorter than in the Mercator and Plate Carree projections and appears “italicized,” with a visible slant towards the northeastern direction. The country’s upper boundary is straight (as opposed to the curvature visible in the Albers Equal Area Conic).

LAB_1_Layout_1 — FIGURE 1: The Effect of Four Distinct Mapping Projections on the Shape, Area, Size, and Direction of the Continental United States

Q4 : How does the position of the cities in relation to each other appear to change between projections (give an example of some cities)?

Looking at Salem and Eugene in OR…

In the Mercator and Plate Carree projections, Eugene appears almost directly South of Salem, whereas in the Albers Equal Area Conic and Robinson projections, Eugene appears Southwest of Salem, especially in the Robinson.

The distance between the cities appears largest in the Mercator projection, while the distance between the cities in the Albers Equal Area Conic projection appears the smallest.

Q5 : What spatial properties (i.e. shape, direction, area) does each projection distort?

Albers Equal Area Conic: shape

Mercator: area, distance

Plate Carree: shape, area

Robinson: shape, direction

Q6 : Use the measure tool to measure the planar distance between cities. How does this distance change between projections? Create a table with your findings.

Vertical (North-South) distances were assessed by measuring the distance between Salem, OR, and Eugene, OR. Horizontal (East-West) distances were assessed by measuring the distance between Bend, OR, and Eugene, OR. See table below:

Projection Type	Distance (miles) between Salem and Eugene	Distance (miles) between Bend and Eugene
Albers Equal Area Conic	61.23	89.99
Mercator	85.47	125.44
Plate Carree	61.36	125.44
Robinson	69.99	96.42

As the table reveals, the Mercator projection displayed the greatest vertical and horizontal distances. The Plate Carree projection displayed horizontal distance as elongated as the Mercator projection, but had shorter vertical distances. The Albers Equal Area Conic displayed the shortest distances (both vertical and horizontal) of any projection.

Q7 : What variables does this dataset contain?

I chose a data set from ArcGIS online quantifying the 2013 homeless population by State. The dataset contains the following variables by state: percent change in homeless population from 2012 to 2013, 2013 overall population, and 2013 homeless population.

Here is a link to an Atlantic article, as well as a report, discussing the homelessness trends that were explored by the U.S. Department of Housing and Urban Development using this data:

http://www.theatlantic.com/business/archive/2013/12/homelessness-is-up-in-new-york-city-but-its-down-everywhere-else/282315/

http://www.endhomelessness.org/page/-/files/2014_State_Of_Homelessness_final.pdf

Q8 : What classification methods did you use? How does each classification method bias the interpretation of the data?

I used natural breaks (Figure 2), quantile (Figure 3), and equal interval (Figure 4) classification methods.

A natural breaks classification method (Figure 2) works well for this particular data because it allows me to separate positive and negative values, in other words, to differentiate and have a clean break between States that experienced an increase in the homeless population and those that experienced a decrease. With natural breaks, the middle values are particularly prominent on the map, while the extremes are fewer and as such, more pronounced. In this way, the map’s audience may assume that most states either saw a slight-to-moderate increase, or a slight-to-moderate decrease in the homeless population, with only a handful of states in the extremes of either significant homelessness reduction or increase.

FIGURE 2: Choropleth Map Using Natural Breaks Classification Method

The quantile classification method, on the other hand, has a much more even distribution of colors, therefore placing less emphasis on any one State, even those that are on the more extreme ends of the spectrum. Additionally, there is no clean break between States who experienced an increase or decrease in the homeless population, as there was one category that had both positive and negative values. This would make legibility difficult, given the nature of the data.

FIGURE 3: Choropleth Map Using Quantile Classification Method

Given this dataset, perhaps the most misleading classification method is equal interval. The equal interval classification is skewed dramatically by the dataset’s outlier, North Dakota’s 200% increase in homelessness, and therefore results in a map in which all other states are lumped into two categories, and in which North Dakota is the only State that visually stands out. Here, like in the quantile classification, there is no clear divide between positive and negative values. The only way in which this map is effective is in highlighting North Dakota as an outlier.

FIGURE 4: Choropleth Map Using Equal Interval Classification Method