Apply tools including vector operations, map algebra, point pattern analysis, hypothesis testing, measuring spatial dependence, and other tools for quantitative spatial analysis for counter mapping the U.S. immigration system.

You work should be informed by readings and class discussions on data sovereignty, data ethics and power.

1. Background on Counter Data/Mapping

2. Review the story: Borderlands: The Line Within

Background on Borderlands and the Separados / Torn Apart project

Separados / Torn Apart Project

• Volume 1: A rapidly deployed critical data & visualization intervention in the USA’s 2018 “Zero Tolerance Policy” for asylum seekers at the US Ports of Entry and the humanitarian crisis that followed.

• Volume 2 of Torn Apart is a deep and radically new look at the territory and infrastructure of ICE’s financial regime in the USA. This data & visualization intervention peels back layers of culpability behind the humanitarian crisis of 2018.

• Credits for Torn Apart project Volume 1 and 2

Data Sources

• Borderlands Archives Cartography

• ICE Detention Facility Lists

Borderlands Film

• https://www.youtube.com/watch?v=6KKDkFo1IQ8 

Other Resources

• NYT on Borderlands Can a Film Find Something New to Say About the Border Crisis? “Borderland: The Line Within” takes a surprisingly multidisciplinary approach fueled by both personal history and government data

• Snapshot of ICE Detention: Inhumane Conditions and Alarming Expansion September 2024

• Immigration Justice Center Research

Your first task is to explore the follow data visualizations from Separados / Torn Apart

• Detention facilities (as of 2018)

• Redacted sites

• People deported by ICE in 2012-2017

• Cumulative ICE awards 2014-2017 to contractors by U.S. congressional district

Describe how the  Separados / Torn Apart project is engaging in counterdata science? Provide two examples. (approx. 200 words).

• Create a new ArcGIS Pro project in your course working directory titled borderlands_labs

• Download the data

  • iceFacs.csv contains ICE detention facilities active from 2014-2018.

    • facility_type_details.csv contains details about the facility type fields in iceFacs.csv

  • historical_IceFacs.csv contains all known former ICE detention facility locations

  • removals_2020.csv contains the count of removals "re-displacement" of people in 2020.

• Set the ArcGIS Pro project Environment settings for all geoprocessing tools to project in the Output Coordinate System: USA Contiguous Albers Equal Area Conic USGS

• Change the your map coordinate system to USA Contiguous Albers Equal Area Conic USGS

Why does this lab use an Albers Equal Area Conic projected coordinate system?

The first data set you will explore contains the location of ICE detention facilities.

• Rename your first map to Part1_borderlands

Load and display ICE facility locations in ArcGIS Pro

• Add the table iceFacs.csv to your home geodatabase

• Add the table to your Part1_borderlands map

How many detention facilities in the iceFacs data set have had their location redacted by US ICE or the National Immigrant Justice Center? What percentage of all facilities? The NIJC redacts the address location of facilities detaining children in Office of Refugee Resettlement (ORR) facilities.

Display the facility locations using the latitude and longitude information contained in the table. 

• Note: you will have to make some edits to your attribute table in order for this work.

• Use the XY table to point geoprocessing tool

Subset the points to only those within the contiguous United States

• You will need to load in an additional layer to complete this step. Load in a polygon feature class of the US states.

  • Note you may need to set a Search Distance when using the tool to select facility locations within the contiguous US.

• If the layer contains Alaska and Hawaii, you will need to remove these states before using it to subset your facility points.

• Add the layer to your geodatabase before proceeding. Be sure to add the version saved in your geodatabase to the Contents pane and remove the original.

Create two charts to summarize the detention facility data

Review the attribute table and create a chart visualization in ArcGIS Pro for the iceFacs data.

• See the Symbolizing vector features tutorial for a review of creating charts from attributes in ArcGIS Pro

Chart 1

• Create a line chart that shows the count of facilities by date of first use

  • Set the interval size to 1 Years

  • Add Data Labels

  • On the General tab in the Chart Properties window, edit the title of the chart to be more descriptive of the data being featured.

• Provide a brief interpretation of the chart (a sentence or two). What does the chart communicate?

Not all immigration detention centers in the U.S. are directly funded by ICE (Immigration and Customs Enforcement), but ICE is the primary agency responsible for most immigrant detention. Here’s a breakdown of how funding works in these facilities:

• ICE-Operated Facilities: ICE directly funds and operates some detention centers. These are run by ICE itself and are fully funded by the agency's budget.

• Privately Operated Detention Centers: Many immigration detention centers are run by private companies, like GEO Group and CoreCivic, which contract with ICE to house detained immigrants. ICE pays these companies through contracts, which makes the funding indirectly from ICE, even if the day-to-day operations are handled by private firms.

• Local Government Facilities: Some local jails or detention centers have contracts with ICE to hold immigrants alongside other detainees. This arrangement, called "intergovernmental service agreements" (IGSAs), allows ICE to pay local governments to house detainees. This setup provides extra funding to local facilities, though they’re technically managed by local authorities.

While ICE is the main source of funding through these various channels, state and local governments, along with private companies, are involved in operating and managing the facilities.

Chart 2

• Create a bar chart to show the number of facilities by the attribute Type.Detailed

  • Sort the bar chart by Y-axis Descending

  • On the General tab in the Chart Properties window, edit the title of the chart to be more descriptive of the data being featured.

  • Use the facility_type_details.csv table included in the data download to interpret the categories

• Provide a brief interpretation of the chart (a sentence or two). What does the chart communicate? 

• Use what you learned completing the Map algebra:  Zonal and Global operations (in-class lab) to generate a raster which represents the distance to an ICE facility in the contiguous US.

Create a map layout to display the distance to the nearest ICE facility in the contiguous US (not submitted on Brightspace)

Reclassifying raster cells

What percentage of the the contiguous US is within 50 kilometers of an ICE facility?

Compute zonal statistics by State

Compute zonal statistics for ICE facilities by state in the contiguous US. See Map algebra:  Zonal and Global operations for help.

• Choose whether you will place the output in a file folder using a the file extension .tif OR add the output to your geodatabase without a file extension.

• Join the results to the a feature class of US states and symbolize to display the median distance to facilities by state

  • Set up clear class breaks

  • Manually edit the Label to by easy to interpret values in kilometers

Create a layout to display the result of distance to ICE facilities by US state.

Re-displacement statistics

Next, we are going to fast forward to 2020, the last year for which a table is easily accessible, to map removal tabulated by ICE jurisdictions (areas of responsibility, AOR) responsible for Enforcement and Removal Operations (EOR).

An investigation by the US Government Accountability Office (GAO) in July 2024 found several problems with the statistics publicly reported by ICE.

What GAO found

The number of U.S. Immigration and Customs Enforcement (ICE) arrests and removals declined from 2019 through 2021, and then increased in 2022, according to GAO's analysis of ICE data:

• The number of arrests varied from calendar years 2019 through 2022 but increased overall, from 133,541 arrests for 2019 to 154,204 arrests for 2022.

• While increasing slightly in 2022, since 2019 removals have declined significantly from 276,122 for 2019 to 81,547 for 2022.

While we do not have access to these data without a Freedom of Information Act request, we can map ICE's published removal counts by ICE jurisdiction.

Map removal counts by ICE jurisdiction

• Add the removals_2020.csv table in the downloaded folder to your geodatabase 

  • Data source: https://www.ice.gov/spotlight/statistics

• Add the table to your map and open to explore the attributes

We will need features to which we can join these data.

• Navigate to the Living Atlas search in the Catalog pane under the Portal tab.

  • Search for and add the USA Major Cities feature class

  • Add the layer to your geodatabase

    • Add the result to your map

    • Open the table to explore the attributes. 

We can conduct a table join to add the removal data to these features:

• Right click on the cities feature class in your Contents pane.

• Add a join

  • IMPORTANT: Unselect the "Keep all input records" box so that the result only contains the matching records, and not all cities in the USA Major Cities data.

  • Set up the join and Validate Join

    • Scroll to the bottom of the results, you should see that 35 records matched.

• Symbolize your points with the field containing the total count of removals in 2020.

Finalize a layer of removal counts by ICE jurisdiction

• Implement class breaks that are easy to interpret.

• Calculate the total number of removals in 2020 using the removals table attributes. Include the count of total removals in your description.

• Use the Lines visualization from Volume 2 of Separados as inspiration for how you symbolize the data. Note: you are not expected to replicate this symbology.

Most residents of the US do not realize the extent of the borderland. Customs and Border Patrol has jurisdiction in any areas of the US within 100 miles of the border. 

In order to add another important layer to our story, and practice using vector operations, you will answer the questions:

• How many people reside within the borderlands? What percentage?

• How many facilities are within the border area (100 miles from US border line)?

• How many people live within 25 miles of detention facilities? What percentage?

Create a layer presenting the extent of the borderlands

To do this, you will create an inward (negative) buffer that includes all areas in the US within 100 miles of the border line.

First, we will use a new tool to dissolve the state borders in the contiguous states layer you created earlier. This will give us a layer that contains only one polygon for the contiguous US.

• Using the US state feature class you created earlier in the lab, create a single polygon to represent the contiguous US boundary:

• • Select the US contiguous state layer in your Contents pane.

  • Open the geoprocessing tool Dissolve (you can learn more about the tool here)

    • Input Features: you states feature class

    • Output Feature Class: US_contiguous_boundary

    • Unselect Create multipart features

    • Select Unsplit lines

    • Run the tool

  • NOTE: check the attribute table to see how many rows are present. If you have over 100 rows then the buffer tool you will run in the step below will take substantially longer to run.

Create a buffer

• Open the buffer tool in the proximity geoprocessing toolkit. 

• Create an inward (negative) buffer

  • Use the dissolved layer you just created as the input features.

  • Rename the output feature class to us_borderland.

  • Set the distance to:   -100 US Survey Miles.

  • Change Side Type to Exclude the input polygon from buffer.

  • Accept the remaining default settings and Run the tool.

• NOTE: the buffer should run in under a minute. If this is not the case for your layer, then you might need to bring in a different layer from the Portal or previous tutorials that contains the US states, then subset to the contiguous US, and create a new dissolve.

Clipping a layer

Next, you will learn how to clip out features using a polygon layer. In this example we will use the recently created buffer layer to clip out all counties that fall inside the borderlands layer.

First, add a layer from the Portal that contains the total population for all US counties.

• Navigate top the Portal tab of the Catalog pane.

• Select the icon for Living Atlas

• Search for "ACS Population Variables - Boundaries"

  • Add this group layer to your map

  • Add the county feature class within the group layer to your geodatabase

  • Remove the group layer from your Contents pane 

  • Add the new county layer in your geodatabase to your Contents pane.

• Navigate back to the Geoprocessing window's main landing page.

• Type "Clip" in the Search bar.

• The Clip tool from the Analysis Tools toolset should be the first tool returned

• Open the Clip tool.

• Select the US counties layer as the layer to be clipped.

• Select the borderland buffer layer as the Clip Features.

• Name the output facilties_ontheborder.

• Click Run.

Note that this output differs from the output of a Select Layer by Location geoprocess in that the latter would return (select) the entire polygon features even if just part of that feature fell within the 100 mile buffer polygon.

The clip tool "cookie cuts" out the polygons thus limiting the output to just the areas of the clipped layer that fall inside the polygon.

The clipped output will always carry over the attributes from the input layer (i.e. the layer that was clipped). The one difference is that the number of features may be less in the clipped layer if the clipping feature did not encompass all the features from the input layer.

In our example, the clipped county layer has the same attributes from the original county layer, but it has  fewer records. Note that it does not carry over attributes from the clipping feature (i.e. the borderlands buffer).

Now, let's answer one of our initial questions: How many people reside within the borderlands?

• Open the attribute table for the clipped county layer.

• Right-click on the Total Population attribute name and select Explore Statistics.

• Scroll through the resulting row to identify the Sum.

• Add this value to your document with responses to the questions throughout the Part 1 activity. 

Carry out the same steps on the Total Population attribute in the original county layer (with all US counties). Use the results to answer the following question: What percentage of the US population resides within the borderland?

Next, we'll use the data to answer the question: How many detention facilities are inside the borderland? What percentage?

• Use the Clip or Select by Location geoprocess to answer the question.

Buffer and spatial join to calculate the population near facilities

Note: these data are not perfectly matched for this analysis. We have facility location information from the FOIA request by the National Immigrant Justice Center in 2018. And, we are using population statistics from the 2018-2022 Census Bureau 5-year estimates. Ideally, we would pull data from the Census for the years that center on 2018, the 2016-2020 ACS 5-year estimates.

• Create a 25 mile outward (positive) buffer around the locations of ICE facilities

Conduct a spatial join of the county feature class to the 25-mile buffers around ICE facilities.

• Select your 25-mile buffer layer for Target Features.

• Select the county feature class with total population as the Join Features.

• Provide a descriptive name for the Output Feature Class.

• Keep the Join one to one Join Operation (see note below for why).

• Match Option: Intersect

• Open the Fields drop down and remove all attribute fields except DETLOC,  Name, Name_1,  Geographic Identifier - FIPS Code, and B01001_001E. 

  • • The Total Population variable name is B01001_001E.

    • You can select/highlight multiple fields and then click the red X by one of the selected fields to remove multiple fields at once.

If a conflict in attribute name occurs, the software will append a number to the end on the attribute name. For example, both input layers have a column called "NAME".

Map the facility buffer zones by total population within 25 miles

• Create a copy of the layer created using the spatial join. 

• Rename the layer "facilities_countypop" and add the layer to your geodatabase using the Data --> Export Features process to add a feature class to the home geodatabase.

• Symbolize the result using Graduated Colors

Do you observe any patterns? Where do more people live near ICE facilities?

In order to map the total population near a facility we needed the total population within 25-miles for each facility. However, there are many facilities in some cities, so there are many duplicate population entries in the Total Population attribute field.

Use the Delete Identical geoprocessing tool to remove duplicate county population data. 

• Use the original result of your spatial join as the Input Dataset. Do not use the layer you used to map the buffer zones by total population.

• Select Geographic Identifier - FIPS Code from the Field(s) list.

• Use  Explore Statistics to confirm that the total population living in counties within 25 is 64,028,509.

  • The total population in the contiguous U.S. using the County data is 334,369,975.

  • Therefore, approximately one fifth of the US total population lives within 25 miles of an ICE detention facility.

Tell a data story

Create one map layout with multiple maps to tell a story. 

• Include at least three map frames, include:

  • Search for and add in an additional layer from the data resources at your disposal (mySidewalk, ArcGIS Online and Living Atlas, the FAQ page, do some web searches)

  • Map removal counts by ICE jurisdiction

  • The extent of the borderlands

• Symbolize your layers and construct your map layout elements to represent the people, lived-experiences, and history that form the context of these data.

  • Consider what we learned in readings and class discussions on data sovereignty, data ethics and power (positionality, visceralization in data visualization, etc.). 

  • You should be intentional about how you symbolize layers (e.g, don't accept default symbology).

• Include a brief map description, title, north arrow, scale bars for each map extent, and data citations.

PART 1: Submit on Brightspace

1. A document with brief responses to the questions throughout the Part 1 activity (see highlighted text)

2. Two chart images

3. A map layout for ICE facility zonal statistics by state

4. A map layout from the Tell a data story section

5. A brief response to the following question:

   • What additional features or datasets would you add if you were to continue and extend this project? In other words, what's missing? 

You are welcome, but not required to construct one single document to contain all of the above.

Layout reminders

• Include a north arrow, legend, scale bar, brief map description, and data citations.

• Use clear, easy to interpret class breaks.

• With the exception of your "Tell a story" layout, each layout should use the same map extent, and the north arrow and scale bar should be placed in the same location.

See below for what to submit after you complete Part 2.

Required readings

• Chapter 11 Point Pattern Analysis

• Chapter 12 Hypothesis testing

Tessellated quadrat count and density

Load in a raster layer of maximum monthly temperature for the contiguous US.

• Download the file and place in your project working directory (home folder for the ArcGIS Pro project)

• Click Yes if prompted to calculate statistics for the layer.

For this part of the lab, you will use a raster containing the maximum temperature across the contiguous US to create tessellated quadrats for ICE facilities at different levels of maximum temperature.

1. Use the Reclassify tool to generate a raster of maximum temperature classified by:

   • Equal interval

   • 12 classes

2. Use the Raster to Polygon tool to convert the integer raster created in step 1 above into a polygon feature class.

   • Create multipart features.

3. Tally the number of points in each tessellated surface using the spatial join workflow covered in the Point Pattern Analysis in-class tutorial.

   • Save a version of the joined feature class to your home geodatabase.

4. With the layer from your geodatabase added to the map, edit the attribute table to calculate area and density.

5. Create a layout: Symbolize the layer and generate a new layout to interpret and communicate the result.

6. Create a chart: Finally, use the chart tool to plot count of ICE facilities vs temperature range. 

   • Make a scatter plot and explore the result. 

Submit on Brightspace:

• Tessellated quadrat density map layout (.png).

• Scatter plot of ICE facilities vs temperature range.

Average Nearest Neighbor (ANN)

We can assess point data for point pattern’s spatial interaction, a second order effect. 

ArcGIS Pro offers a tool (Average Nearest Neighbor or ANN) that tests whether or not the observed first order nearest neighbor is consistent with a distribution of points we would expect to observe if the underlying process was completely random (i.e. IRP). ArcGIS’ average nearest neighbor (ANN) tool computes the 1st nearest neighbor mean distance for all points. It also computes an expected mean distance (ANNexpected) under the assumption that the process that lead to the observed pattern is completely random.

• If ANNratio is 1, the pattern results from a random process. 

• If it’s greater than 1, it’s dispersed. 

• If it’s less than 1, it’s clustered. 

In essence, ArcGIS is comparing the observed ANN value to the ANNexpected one would compute if a complete spatial randomness (CSR) process was at play.

ArcGIS’ tool also generates a p-value (telling us how confident we should be that our observed ANN value is consistent with a perfectly random process) along with a bell shaped curve in the output graphics window.

Your turn!

Use the Average Nearest Neighbor tool to answer the following questions:

• Are ICE facilities randomly distributed, clustered, or dispersed?

• On average, how far apart are ICE facilities from their first nearest neighbor?

Steps

• Set the ice facilities as the input feature.

• Keep the default Euclidean distance option.

• Check the box next to Generate Report.

• ArcGIS’ ANN tool offers the option to specify the study surface area. If the area is not explicitly defined, ArcGIS will assume that the area is defined by the smallest area encompassing the points.

  • Set the study extent area to 7,779,067,000 (do not include the commas when you input). The units are adopted from the layer's coordinate system—meters in this example.

The output is not a data layer but a report saved as an HTML file. The filename is listed in the Geoprocess' Details window. 

• In the Analysis tab, click on the History button.

• Click on the Average Nearest Neighbor tool. This will bring up a side window.  (You might need to hover your cursor above the  tool for a second or two to see the window pop up).

  • Click the pop out button as needed to open the full window.

• In the side window, click on Report File link (this should open an HTML file saved in your project folder).
  
  

Submit on Brightspace:

• Given the ANN tool results, write your response to the following  questions (approx. 150 words).

  • Are ICE facilities randomly distributed, clustered, or dispersed?

  • On average, how far apart are ICE facilities from their first nearest neighbor?

PART 2: Submit on Brightspace

1. Tessellated quadrat density map layout (.png).

2. Scatter plot of ICE facilities vs temperature range.

3. ANN: Given the ANN tool results, write your response to the two questions specified (approx. 150 words).

Layout reminders

• Include a north arrow, legend, scale bar, brief map description, and data citations.

• Use clear, easy to interpret class breaks.

• With the exception of your "Tell a story" layout in Part 1, each layout should use the same map extent, and the north arrow and scale bar should be placed in the same location.