GIS Internship: Santa Rosa County Property Appraiser's Office

 This semester, I am completing my internship with the Santa Rosa County Property Appraiser's Office (SRCPA) in Milton, Florida, as part of my undergraduate GIS certificate program at the University of West Florida. My role primarily revolves around cadastral mapping, where I assist with projects such as converting the county's parcel database to Esri’s ArcGIS Parcel Fabric, georeferencing historical aerial imagery, and improving data quality for land records management.

I have been in this position for six weeks now and have really enjoyed the experience. Although the commute is about an hour each way, I genuinely enjoy the drive. Milton reminds me of my hometown, which makes the trip something I look forward to.

One of my first challenges was learning how to work with ArcGIS Pro mosaic datasets, a tool my office hadn’t used before. Our team was unsure how to crop georeferenced images efficiently, so I took the initiative to research solutions independently. Through this process, I learned how to use footprints for cropping, perform color balancing, and export GeoTIFF files, significantly improving our workflow.

Currently, I am preparing to process the historical aerial images of Santa Rosa County that I have georeferenced in GDAL. These mosaics will be finalized and published to the SRCPA website, where they will be made publicly available for residents, businesses, and other stakeholders.

As part of this course, I have also joined the Florida Association of Cadastral Mappers (FACM), a professional organization focused on cadastral mapping across the state of Florida. Membership is affordable, with a lifetime fee of just $5 for students or $40 for professionals, making it a valuable resource for networking and professional development. Since cadastral mapping is at the core of my internship work, this group is an excellent fit and will allow me to learn from other professionals in the field.

For more information on the FACM, visit their website: https://www.facm-online.org/

GIS4930 - Data Quality - Fundamentals

We kicked off the Special Topics in GIS course with a lab focused on data accuracy and precision, two important concepts for evaluating GPS data quality. In this exercise, we worked with a set of 50 waypoints collected using a handheld GPS unit. The goal was to figure out how tightly the points clustered together (precision) and how close they were to a known reference point (accuracy). Using ArcGIS Pro, we calculated both horizontal and vertical precision, created buffer zones to visualize the spread of the data, and compared our results to the reference point. To wrap up, we explored error metrics and built a cumulative distribution function (CDF) to get a clearer picture of how GPS errors are distributed across the dataset.

---

Horizontal Accuracy: 3.24 m
Horizontal Precision (68%): 4.5 m

Horizontal precision is a measure of how closely repeated GPS measurements align with one another. In this lab, it was calculated as the distance within which 68% of the collected waypoints fall.

Horizontal accuracy measures how close the average GPS location is to a known reference point. This was determined by comparing the average waypoint location to a surveyed reference point.

In simple terms:

• Precision = consistency (how tightly grouped the points are)

• Accuracy = closeness to the “true” location

---

After finishing our GIS work in ArcGIS Pro, we switched to Excel to calculate detailed error metrics for the GPS points. Using formulas, we compared each collected point to a benchmark location and calculated the distance error for every point. From these values, we determined statistics like the minimum, maximum, mean, and percentiles.

One key metric we calculated was the Root Mean Square Error (RMSE), which is a single number representing the typical error across the whole dataset. In this lab, our RMSE was 3.06 meters, meaning that, on average, the GPS points were about three meters away from the benchmark location. RMSE is especially useful for quickly evaluating and comparing the overall quality of GPS data.

---

The Cumulative Distribution Function (CDF) scatter plot can be used to visually determine data metrics, such as the minimum, median, and maximum error, as well as percentiles (68%, 90%, and 95%). However, it cannot be used to directly determine the mean or RMSE.

The CDF provides a clearer picture of the dataset than individual metrics alone, as it shows how all the errors are distributed. It helps identify whether most GPS points are clustered together or if there are a few outliers with significantly higher errors, extending the line out into a “tail.” This insight goes beyond summary numbers and helps in evaluating the overall quality of the GPS data.

M5 - Applications in GIS - Hazards: Damage Assessment

In this week's lab, we kept focus on Hurricane Sandy, but this time using GIS tools to conduct damage assessments and evaluate the storm's effects on individual properties. 

We began the lab by using previously learned GIS and cartography skills to prepare a map illustrating Hurricane Sandy's track, including the intensity of the storm as it migrated from the Caribbean Sea into the Atlantic Ocean and up the east coast of the United States.  We had some fun with symbology and also learned to add a grid and graticules to our map layout.  This was just a small reminder that I enjoy the cartography aspect of GIS. 

Next, we utilized Survey123, an excellent tool provided by ESRI for creating, managing, and analyzing surveys, to develop a citizen "self-reporting" damage survey. We learned how to construct a comprehensive survey that would allow citizens to report storm damage, including the type of damage (with a text input for details), images, the date and time, and the exact location of the damage. This week’s lecture highlighted how valuable this method of data collection can be immediately after a significant storm like Hurricane Sandy.

We proceeded through the lab by taking historical aerial imagery of the coastline both pre- and post-Hurricane Sandy, and merging them into a mosaic dataset that we could utilize to visualize the before and after.  We learned how to use the Swipe and Flicker tools for comparison, which really aided in damage assessment tasks later in the lab. 

We moved on to learning more about data creation and specifically attribute domains.  Attribute domains in ArcGIS Pro are predefined sets of valid values or ranges that restrict the data entered into a field in a geodatabase. They ensure data integrity by limiting the choices for attribute values, helping to maintain consistency and accuracy in data entry.

This made the task of examining the before and after imagery of a study area, creating point features, and analyzing property damage parcel by parcel much more streamlined and accurate in terms of our data entry.  

In this process, we also used our previously learned skills to create symbology that accurately conveys damage type when viewing our data on the map. 

Finally, multi-ring buffers around the coastline were created at 100 m, 200 m, and 300 m intervals. Using the Tabulate Intersection tool, I summarized the counts of structures within each buffer zone, linking the damage assessments to the corresponding distance ranges.

The analysis shows a strong link between how close buildings are to the coastline and the amount of damage they experience. Structures within 100 m of the shore had the highest damage rates, with fewer issues as you move further out to 200 m and 300 m. This indicates that being near the coast makes buildings more vulnerable. While we can estimate that, say, X% of buildings within 100 m are damaged, we should do more local studies to confirm these numbers before applying them to other areas.

M4 - Applications in GIS - Hazards: Coastal Flooding

Module 4 of Applications in GIS introduced us to a critical use of GIS, especially relevant to us here on the Emerald Coast, analyzing coastal flooding.

We started the lab by working with LiDAR data, learning the process of creating a usable Digital Elevation Model (DEM) from raw .laz files. We specifically used data collected pre- and post-Hurricane Sandy in a coastal area of Cape May County, New Jersey, to visualize the incredible force a storm such as this one can have on the landscape. 

Pre- and post-Sandy DEM illustrating barrier island breech at Mantoloking Road, Mantoloking, New Jersey 

This direct comparison vividly demonstrated the immense force a storm like Sandy can have on the landscape, revealing extensive coastal reshaping, with clear evidence of significant erosion, loss of buildings, and even a dramatic breach of the barrier island. It also highlighted the critical challenge of accounting for post-disaster recovery and rebuilding efforts in our analysis.

Visualizing 2-meter storm surge in Cape May County, New Jersey

Shifting gears a bit, we then focused directly on storm surge analysis in Cape May County, New Jersey. We learned how to define flood zones by reclassifying elevation data to show exactly which areas would be underwater during a specific surge height, like the 2-meter surge experienced during Hurricane Sandy. This allowed us to calculate the exact percentage of the county that might have been inundated, revealing that approximately 47% of Cape May County was likely affected.

Finally, we moved to Florida for what I found to be the most eye-opening part: a direct comparison of different elevation data sources. Using a 1-meter storm surge scenario, we stacked up flood predictions from a traditional USGS DEM against a more advanced, high-resolution LiDAR-derived DEM. This allowed us to pinpoint exactly how different data sources can change our understanding of flood risk, helping us to calculate and visualize critical errors of omission and commission (where one model got it right and the other didn't) between the two models. This analysis was key to understanding the real-world accuracy challenges in flood mapping.

However, our lab also made us aware of some key assumptions and simplifications. Assuming storm surges always cause uniform flooding is unrealistic, as water levels vary and can reach isolated low areas. We also didn't account for other causes like heavy rain, large waves, or tidal timing. For greater accuracy, advanced computer models should simulate different flood scenarios and water movement, considering all these factors.

M3 - Applications in GIS - Visibility Analysis Using 3D

Screenshot of a 3D Scene of Downtown San Diego, California in ArcGIS Pro (I used to live within the area pictured, close to Petco Park)

This week, we were introduced to another interesting application in GIS: analyzing visibility and line of sight using 2D and 3D scenes in ArcGIS Pro.

For this module, we were tasked with completing four web courses through the Esri Academy. I enjoy how these Esri courses are structured; they are easy to understand and comprehensive. I appreciate that, as we learn new tools and processes, the courses provide practical exercises in various methods frequently used in ArcGIS Pro.

Our first course was an Introduction to 3D Data. It was a general overview for navigating and exploring 3D scenes (including some really helpful shortcut keys), functional surfaces, and (rasters and triangulated irregular networks (TINs.)) The course also introduced us to the various 3D features, such as vector features in 3D, meshes, and multipatch features.

Screenshot of an ArcGIS Pro Line of Sight analysis of a hypothetical parade route for the purpose of planning security.

Next, we learned to perform a Line of Sight Analysis. A line of sight calculates intervisibility between the first vertex—the observer—and the last vertex—the target—along a straight line between the two. A line of sight considers any obstructions provided by a surface or multipatch feature class. Visibility between these points is determined along the sight line.

Line of sight analysis is invaluable in various applications, including planning security for large events such as parades, sporting events, or concerts. This data is also beneficial in urban planning, allowing us to study the impact a proposed new building would have on existing views. Additionally, line of sight analysis is crucial when planning a new air traffic control tower, ensuring that the ATC team can effectively monitor the entire tarmac.

Screenshot of an ArcGIS Pro Viewshed Analysis, utilized for planning the positioning of artificial lighting in a campground. The color variance illustrates the overlap of light output, with darker colors indicating areas receiving illumination from multiple light sources.

We then jumped into performing Viewshed Analysis in ArcGIS Pro. The Viewshed tool visualizes visible areas from a vantage point. This tool has many practical uses, including how to locate guard towers, fire watchtowers, or event surveillance.

One exciting process we engaged in was modeling light visibility using the Viewshed tool. We simulated assisting a campground director in strategically placing new artificial lighting to maximize illumination for campers' outdoor activities after dark. By visualizing light distribution at heights of 3 meters and 10 meters, we were able to make informed decisions on the optimal placement of lighting across the property.

Screenshot of a 3D scene I authored within ArcGIS Pro illustrating an area of downtown Portland, Oregon

Finally, we learned to author a 3D scene and share a 3D scene as a hosted scene layer package. Sharing a scene online in 3D gives the viewer a greater sense of realism, making it easier for the viewer to make decisions and achieve solutions with greater understanding.

This was a fun module, and I look forward to completing more of the training courses in the Esri Academy.

M2 - Applications in GIS - Analyzing Forestry with LiDAR

Distribution of Tree Heights in the Study Area

In Module 2 of Applications in GIS, we jumped into working with LiDAR (Light Detection and Ranging) technology and its applications in forestry. The primary focus was on decompressing .las files, creating Digital Elevation Models (DEMs) and Digital Surface Models (DSMs), and calculating forest height using LiDAR data sourced from the Virginia LiDAR application.

I'm really fascinated with LiDAR and excited to work with the data type more. However, I realized that this data type can be quite cumbersome when it comes to processing. This week, I struggled with the UWF virtual machine we use to run ArcGIS Pro, which was lagging significantly. Unfortunately, I couldn't get the LAS layer to rasterize, so what you see in the included map isn't what it was supposed to represent.

Canopy Density Analysis: Visualizing Forest Coverage

3D Visualization of Forest Structure Using LiDAR/Topographic Representation of the Terrain

M1 - Applications in GIS - Crime Analysis

Week one of Applications in GIS introduces us to crime hotspot mapping and the different methods used to identify and analyze areas with elevated crime rates, enabling law enforcement and community planners to allocate resources effectively and implement targeted interventions.

In the first part of this week’s lab, we conducted an exploratory crime analysis focused on burglaries in Washington, DC. Using ArcGIS Pro, we isolated burglary reports and analyzed their distribution across census tracts. By counting the number of incidents per tract and normalizing this data against the number of housing units, we calculated a burglary rate, which provides a clearer understanding of crime patterns. The map pictured above illustrates the burglary rates across the census tracts.

In part two of this week's lab, we explored various methods of crime hotspot mapping, focusing on techniques like Grid Overlay, Kernel Density, and Local Moran's I. Each method offers unique insights into crime patterns, allowing us to assess their effectiveness in identifying high-crime areas.

The graphic above illustrates the results of our analysis, highlighting the different hotspot areas identified by each method, which can aid law enforcement in making informed decisions regarding resource allocation and crime prevention strategies.

After completing the three different methods of hotspot analysis, we were tasked with deciding which method we believe would be the most useful for a police chief working to predict and prepare for future crime patterns.  

From the reading materials we have covered this week, as well as seeing and working with these data in the real world, I believe the Kernel Density method of hotspot mapping would be the most ideal for a police chief when making decisions on how to police their region.

While the Grid Overlay method captured a much larger area, the crime density doesn't seem as high as I would have thought, which leads me to believe it includes some areas with lower crime rates. This large area could strain department resources if they attempted to police it in its entirety.

The Local Moran's I method returned a more balanced area compared to the other two methods; however, its crime density is the lowest of the three, suggesting it may not be as helpful for targeting high-crime areas.

The Kernel Density method returned the smallest area but the highest crime density. This indicates it is more accurate in pinpointing high-crime areas. The department can focus on a smaller area, utilizing its resources more efficiently and effectively.

This insight into crime analysis and mapping techniques was fascinating, and I could definitely see myself enjoying working in this application of GIS.  I believe that more education in criminal justice would be necessary to truly be of help to law enforcement, but I'm always open to learning. 

I am excited for lab two, which covers LiDAR (Light Detection and Ranging). LiDAR is a remote sensing technology that uses laser light to measure distances and create precise, three-dimensional information about the Earth's surface and other objects.