The University of Florida (UF) is transforming patient care and disease prevention, crop safety and yields, and machine learning research. Soon, UF could change the way land use is determined, not just in-state but nationally and potentially across the globe.
“Just a few years ago, the typical process for county planning and zoning decision-making was to schedule a meeting, gather all interested parties in a room, and try to determine what the best land use designation for available parcels would be,” mused Paul Zwick, smiling at the memory.
Zwick, professor emeritus in the School of Urban and Regional Planning, knows the process well. More than a decade ago, he and fellow College of Design, Construction, and Planning (DCP) professor Peggy Carr developed LUCIS, which stands for Land-Use Conflict Identification Strategy. The software application was conceived to deliver informed decisions based on data analysis and not just best guesses.
The LUCIS model uses as its foundation commercial geoprocessing software that allows it to provide analysis on a parcel of land’s suitability for multiple major land-use categories (e.g., residential, commercial, agricultural), determine potential conflicts among those categories, and analyze future land-use scenarios based upon identified land-use conflicts.
Using LUCIS, Zwick and Carr could deliver several data-driven planning scenarios previously unavailable at the actual planning stage. These options included an environmentally friendly scenario, along with an agriculture productivity model and an urban growth forecasting model, to name a few.
For municipalities, the ability to have models that answer so many questions before assigning a designation to undeveloped or little-used parcels of land represented a transformation of the planning experience. County administrators could now show conservationists, developers, water management district officials, and anyone else how the various scenarios would impact the community.
Speeding up the process
But times change, as do expectations. While LUCIS brought land-use planning out of the paper, pen, and best-guessing age, today’s technology-infused world requires more expedited decision-making and more planning options. The LUCIS application, running on a laptop and localized server in Zwick’s DCP office, took as long as six months to complete processing of a municipality’s land use analysis.
Enter then-graduate student Changjie Chen. Chen – from Mudanjiang, China – was working on his master’s degree in Urban/Regional Planning, with Zwick on his thesis committee.
“Paul’s research very much interested me,” says Chen. “I came to UF because the urban planning department was already top-15 when I applied, and very focused on research. Research is extremely important to me. I had studied Paul’s research extensively, so getting to work with him was such an important opportunity for my studies and my career. It is my dream come true.”
Since his research interest required speeding up the process of land use analysis, the first thing Chen had to learn was the geoprocessing software (GIS), standard in commercial use, that powered the LUCIS environment.
“I didn’t know much about GIS when I arrived from China, except that it stood for ‘Geographic Information System!’" says Chen. "The more I studied, the more it interested me. And then this focus on learning everything about geoprocessing with GIS got me very focused on programming."
The more I learned about LUCIS, the more I realized that you cannot get all of the functionality needed to conduct the best land-use planning from the available commercial software packages on a desktop or laptop environment.
Chen worked with UF Information Technology Bioinformatics Specialist Oleksandr “Alex” Moskalenko to make the commercially available, industry-standard GIS software do what was required to work in a high-performance computing environment.
Significant modifications were necessary for the modelling to work on HiPerGator, the University of Florida’s supercomputer for research. Chen and Moskalenko decided to use programming language Python to expand the set of land-use models produced as well. To make this happen, Chen had to rewrite all code in the LUCIS simulations from the commercial software into Python—an 18-month commitment.
“Alex helped me with the setup process, including databases and in creating a computing environment on HiPerGator, which we named ‘geopython’,” added Chen.
The result of Chen’s dedication?
- Instead of taking six months, the modeling can now be done in five minutes running parallel processes on HiPerGator.
- With the original LUCIS application, it took six months to generate the three most common land use scenarios. Now? In five minutes, UF can produce 100 land-use scenarios for any municipality or government entity that wants them.
Thesis advisor and mentor Zwick is duly impressed:
“What Changjie has done is incredible! The benefits and implications for Florida or any other state with the data available can transform the way city and county governments plan for growth. Imagine knowing the impact to roads or whether a parcel’s soil quality supports its agricultural designation. The efficiencies gained in reviewing and planning optimal space use is amazing; and Changjie rewrote all of the models to make this ‘from six months to five minutes’ possible. This is his work.”
Zwick is thrilled that the next generation has taken his research and built on it. He also knows that 30 or 40 years after a parcel of land was initially developed, the time to rethink its use is at hand:
"Changing demographics necessitate changes in land use. There are real estate implications, for example—a proposed retirement community has a set of criteria that is very different from say, a mixed-used development aimed at millennials."
The computing power required to deliver land-use models in such a short period of time is significant, and the work couldn’t have been produced without a pilot research project grant of 32 HiPerGator compute cores to run the analyses. UFIT Director of Research Computing Erik Deumens was extremely impressed by what Chen wanted to try:
“I became interested in Changjie’s research after I learned he was working with Alex [Moskalenko] to implement GIS analysis through Python on HiPerGator. A university is for exploring new ideas and the focus of Changjie’s research certainly met all of the criteria for a new and innovative way of helping municipalities be better prepared for growth.”
As for Chen, times have changed for him, too. Since first arriving in Gainesville in 2012, he has earned a master’s degree and this spring, earned his Ph.D. in urban planning. Chen is looking forward to publishing additional articles on parallel computing and hopes to become a research university faculty member, just like his mentor. Chen is also working on completing a second master’s degree in statistics this December, making him a triple Gator grad. What else is on his horizon? Hopefully teaching as well as research.
“I taught for the first time this spring, an undergraduate course in Python,” said Chen. “It was really enjoyable. I loved teaching students and helping them in their academic career. I would like to teach more.”
The now retired Zwick is by no means on the sidelines when it comes to the future of urban planning and the possibilities generated by “LUCIS version 2.0”:
The modeling and analyses available for land use were limited. Those days are over! Now, we can come into a meeting with multiple land-use scenarios and modify them, based on the specific criteria a county has. Marketed effectively, this application could be the urban planning success equivalent to Gatorade!
Marketing considerations notwithstanding, the models created by UF researchers can quickly deliver quality data to impact decision-making across Florida. The predictor models can also cut down on conflicts between developers, farmers, residents, environmental activists, citizens, and government staff—everyone who has a vested interest in how land is developed (or not developed) in the Sunshine State.