Undergraduate Research with GIS
our merry band of 12 undergraduate GIS students presenting their research posters at SUSRC.
Today was the Salisbury University Student Research Conference (SUSRC). I love the SUSRC because it represents what is so great about Salisbury University: the undergraduate experience. Our university prides itself in giving undergraduates a graduate school experience. As a friend once told me, when you come to Salisbury University, you’ll find yourself in the lab, with a Professor, with a PhD.
Salisbury University has really shown me the joy of working with undergraduates. I get to watch them enter as immature 18 year old freshmen, get excited about quantitative geography when they take Statistical Problem Solving in Geography their sophomore year, and continue to mature as they take GIS Programming and Advanced GIS, along with so many of the other great course offerings in Geography and Geosciences.
The last couple of years in my Advanced GIS class, I decided to have the students work on their Final Projects in the middle of the semester so that they could present their work at SUSRC in the form of a poster. So, we dive right in on a lot of advanced techniques (network analysis, modeling, spatial interpolation, spatial statistics, terrain analysis, etc.), and then stop for 4 weeks to work on final projects. I then finish out the year with more lectures on computational geometry, algorithms, and geodatabase design (which don’t really effect their research project).
Each year the posters get better and better, and the course has begun to turn into a sort of capstone GIS/quantitative geography class. For the student projects, my expectations are:
The statistical analysis must be at a level to rival our work in Statistical Problem Solving in Geography.
The maps, charts, and graphs must rival the work they do in Cartographic Visualization.
Their analysis must be at the level of work they’ve done in Introduction to GIS, Advanced GIS, GIS Programming, and Spatial Modeling.
Their research idea should complement their particular track in the major (atmospheric science, plannning, human geography, geology, GIS, etc.).
So, it is a way to tie up all that they have learned into a project they can be proud of, and then show off to potential employers.
So for today’s very long post, I want to introduce you to my students, and the work they showed off. I think each of them would be an asset to any company, and I’m proud of the fact that most of them already have jobs, funding for graduate school next Fall, or a summer internship lined up – we work them very hard at Salisbury, and it really pays off.
I am so proud of each and every one of them – they feel like family, and each day I can’t wait to go to work so I can see them.
So let’s meet the students (I’m missing three of them, as their pictures didn’t come out – I will try to add them later)….
Today, Caitlin is one of the hardest working students in the Department. Not only does she work very hard at school, but she finds that she works way too many hours at a local restaurant to help pay for school. You don’t find students with that kind of responsibility and work ethic every day.
Caitlin’s project was to look at forest fire distribution in California. She was interested in determining the relationship between forest fire frequency and human and natural phenomenon. After collecting data from 1980 and 2010, and quantifying the data into a set of quadrats, Caitlin ran multiple regression analysis for many different factors. None of the results showed any statistical significance (in this case, like Edison, she learned many ways how to not make a light bulb), but that is ok, as it showed there simply wasn’t a statistical correlation of those factors with forest fire frequency. Another thing Caitlin was interested in was the degree of spatial clustering of forest fires. Using the same quadrats, she invoked the iterator functions in ESRI’s model builder to run Moran’s I indices for the last 30 years. The little chart on the top of her poster shows the Morans I values for each year. As one would expect, there is spatial autocorrelation in each year, but the distribution of the clustering is random over the 30 years – meaning, forest fires in the State are neither becoming more or less clustered over time.
This past April, Caitlin won a GIS competition to become an intern with a GIS company outside of Washington, DC.
Connor (a Sophomore, yes, that’s right, a Sophomore) came to SU ready to hit the ground running. Usually when you see a kid with this much enthusiasm his Freshman year you worry: they tend to get ahead of themselves, and fall face-first into a buzz saw as they bite off more than their 18 year old brain can handle.
In Connor’s case, he did in fact learn a lot about the challenge it was to take a course with mature Juniors during his Freshman year, but he never wavered. Connor is always thinking about geography: what it means, and how it can be applied. We talked a lot last year about whether he should be taking Advanced GIS as a Sophomore, and I told him I would let him take the course, but that I didn’t recommend it. Boy was I wrong! Connor has really thrived in Advanced GIS. As a Human Geographer, Connor is very interested in how geography shapes culture.
For his project, Connor looked at human development indices (hdi) in Brazil for income, education, and equality (one of our faculty members collected mountains of data as part of an NSF team). Connor first conducted a k-means cluster analysis to determine which settlement areas were similar from strictly their attribute ordering. Once he established a series of clusters, he performed a Moran’s I test to determine if the attribute based clustering also had a spatial clustering component. His tests indicated that there was significant spatial clustering for those like-areas. He also found the top and bottom 20% for each hdi category, and then identified which areas were in all three (i.e. which areas were in the top 20% for income, education, and equality). While the results were not surprising, it still made for a beautiful map that showed the spatial autocorrelation for each group, and did yield some very interesting patterns that many faculty were interesting in discussing and speculating about. Did I mention that Connor is only a Sophomore!
Jessica loves the outdoors, and is fanatical about nature, so it was no surprise that her project was on a multi-criteria analysis of spotted owl habitats in Washington State. For this project, Jessica read up on a number of scholarly articles that examined the habitat criteria for the spotted owl, and then blended the many criteria into her own analysis using model builder. One of the biggest issues Jessica had was dealing with the voluminous amount of data for things like land use, tree canopy, and elevation. Fortunately, she is like a pit bull, and doesn’t give up easily. I was so proud of her, as each day, working with all that data was a struggle, but she kept fighting through it. We had to dial back some of her area, but the results of her analysis were very interesting. In fact, her analysis had a significant overlap with habitat areas that the State of Washington already identified, and by pulling in criteria from other scholarly papers, Jessica found other areas that appear to be a suitable habitat for the spotted owl.
During the first week of GIS Programming, Austin realized he wasn’t in Kansas anymore! Unfortunately, Austin didn’t have much in the way of the prerequisites to take GIS Programming. In fact he told me “Dr. Lembo, I’m in way over my head, but this stuff seems really cool“. Since the add/drop date was a month away, we decided to just take it day-by-day. That was a great decision. Every day you could see Austin getting more excited about what he was learning, getting better at programming, and finding all kinds of ways to experiment – he definitely caught the programming bug! In fact, in Advanced GIS, rather than using the GUI, Austin just programs all of his labs using Arcpy! And boy, was he good at it. So, for his final project we decided to really try and push him by having him develop an Internet-based project with Arcserver. Austin had never worked with client/server protocols before, but if anyone had the tenacity to work at it, figure things out, and deliver a product, it was Austin.
Austin used the Arcserver API to develop an application for field reconnaissance by allowing users to select a geographic area and then return a Viewshed for the area as a GeoPDF. As you can see from the poster, Austin’s cartography skills are really good. But, his ability to work with the API and develop an application that runs on a tablet or phone is what made this project really cool. In fact, Austin was the only person at the poster session who was demonstrating his work using an iPad. So, that area of the room was pretty crowded with people wanting to see the cool tool everyone was talking about.
Carl’s research focused on nutrient loading in the Chesapeake Bay. With his scripting skills, Carl created watersheds for 28 different USGS gauge stations around the Bay, and quantified the land use within each contributing area. From there he used published export coefficients for the land use to create a lumped model for estimating nutrient loading. I typically stop students at that point, as it is sufficient for an undergraduate research project. Carl took it step forward and programmed a spatially explicit model originally defined by Endreny and Wood using Arcpy and numpy – and it recreated the entire model in a matter of weeks.
Walker has been working with a handful of climate stations around the region over the years, and used inverse distance weighting to interpolate the impact of synoptic weather events throughout the region based on the readings from his stations. After generating his surfaces, he performed a cluster analysis (using 5 clusters) to find similar areas. The similar areas were then turned into a map of distinct synoptic regions throughout the State – something I had never seen done before. It was remarkable how similar his statistically generated regions were to the physiographic regions defined by the Maryland Department of Natural Resources. I know Walker has big plans for graduate school, but as a side hobby his is also interested in extending this work in an attempt to generate synoptic regions for the entire eastern seaboard – Roll Tide!
Elkins Internship Award where he worked as a GIS analyst in the City of Baltimore Planning Department. When he’s not in the GIS lab, Tyler spends many hours working with SU athletics, and you can see Tyler sitting court side announcing SU basketball games (I love hearing him shout over the loudspeaker “another threeeeeeeeeeeeeee”).
For his research, Tyler borrowed my Phantom UAV and conducted a study to determine the heights where people could no longer discern if what they were seeing on the ground was a human or not. Because people were ranking the data, Tyler used a Kruskal Wallace test and Mann Whitney pairwise test to determine which heights appeared to be the breaking point. This was an interesting study as there was certainly a height where it became difficult to identify people – something Tyler hopes is used to better understand how to perform rooftop rescues and reconnaissance of people stranded during floods and natural disasters.
As a biology major, Chris was interested in animal habitats. So, he decided to develop a multi-criteria habitat model for Florida Manatee migrating to the Chesapeake Bay. The first thing I learned from this project was that yes, Manatee do migrate up to the Bay! Chris researched some papers that discussed the optimal habitats for Manatee such as the abundance and type of submerged aquatic vegetation, the water depth, and water temperature among other things. The real trick in this project was assembling the massive amount of data. When it was all done, his multi-criteria model found a number of spots where the Manatee were likely to be found. Finally, he collected the sighting locations and placed them on top of his map to see how is model fit with the actual sightings. I have to thank Chris for making this a great semester, but especially thank him for serving in our Armed Forces!
Here on the Eastern Shore of Maryland, nitrogen loading in the Bay is a big issue. And, a lot of that nitrogen is believed to come from chicken manure. Maggie focused her project on creating a multi-criteria analysis to identify suitable locations for anaerobic digestion system for the manure. Now, most of us (including Maggie) know that anaerobic digestion of chicken manure is not very efficient. But, Maggie was not concerned as much with the ability to generate lots of energy, but rather to find some way to incentivize farmers to remove the excess manure from their fields. To complete the work, Maggie researched a number of papers that evaluated optimal locations anaerobic digestion systems for chicken and other manure (i.e. cows). She collected the necessary data for Wicomico County and put her criteria into model builder to identify the best places to locate the digestion systems. Maggie is now starting to tinker with open source GIS, as she knows the Peace Corps can make extensive use of GIS and a free and open software system will allow them to accomplish these tasks in an affordable way.
As you might imagine, I’m exhausted at this point in the semester. Juggling 12 research projects (13 actually, when you count Tyler) takes a lot of concentration and energy. But, it is worth it. As you can see, these students aren’t just people who sit in my class, they are friends. And getting to know them, hearing their stories, and watching them grow over their four years at Salisbury University make this such a fun job – trust me, being exhausted at the end of a semester is a small price to pay for the privilege of being a college professor.