Flood Mapping and Emergency Management: An Interview with Jude Kastens
Jude Kastens is an associate research professor at the University of Kansas. He links the ARISE project to the Kansas Applied Remote Sensing program, where he works as a numerical analyst and all-purpose data modeler. But that’s just scratching the surface.
In this interview with freelance writer Melinda Cordell, Dr. Kastens shares his work on flood mapping, emergency management, and environmental modeling, reflecting on his years of experience in those arenas. He also discusses the real-world impacts of his projects and the partnerships that have helped bring them to life.
Q. As a former river rat who’s had to paddle home in a canoe during the Missouri River floods of ’93 and ’11, I’m very interested in this research. Could you share a bit about how you got your start in flood mapping?
A. Back in ’07, during the severe flood in southeastern Kansas, I was in the throes of developing a hydrologically-informed model called FLDPLN (pronounced “floodplain”) to identify floodplains, or river valleys, using elevation data. A colleague – also a student at KU – was working with emergency management to identify floodwaters in satellite imagery. This is tricky because usually it takes at least a day or two to access a new image, and often there are clouds obscuring what’s happening on the ground.
To help focus the flooded image mapping process, we used the local stream gauge data to calibrate a FLDPLN model delineation of river valley lowlands, which is where most Kansas flooding occurs. After the flood, through comparisons to flooded imagery, we realized that the gauge-calibrated model was doing a very respectable job of mapping the actual flood itself.
Q. Can you tell us about your involvement in flood mapping and some of the specific events you’ve worked on?
A. Following the 2007 event, we were awarded a sequence of GIS policy board projects to build our Kansas flood-mapping library. We assembled the data and the information, but we still needed to develop a delivery mechanism to make it accessible.
When the 2019 flood hit, we got the opportunity to test our mapping methods in a live situation. Working extended hours for many days while events unfolded, I made several real-time and forecast flood maps for the Kansas Division of Emergency Management (KDEM) and Kansas Water Office (KWO). Our efforts were generally viewed as a success, and subsequently we received more funding from KWO to build out automated flood mapping tools. The system is now very fast and can update in just a couple minutes, with results featured in the Kansas Flood Mapping Dashboard that is accessible to anyone through the Kansas Applied Remote Sensing (KARS) website.
Q. The 2011 Missouri River flood was a peculiar, prolonged disaster situation. What made it unique in your career?
A. In the summer of 2011, the lower Missouri river experienced a three-month flood from the slow release of massive amounts of stored floodwater out of upstream reservoirs. The river valley by Nebraska was inundated, while crops in the uplands just outside the flooded area were suffering from drought.
State emergency managers in Kansas were getting calls from people wanting to know if the Missouri flooding would cause water to back up on the Kansas River. The answer was probably not a lot, and almost certainly not like in 1993 when the Kansas River was flooding on its own (not the case in 2011). Through our simulations, we were able to verify our assumptions for only minor flooding along the Kansas were correct, assuaging some concerns.
Q. Your projects sound very collaborative. Can you tell us about your team?
A. We’ve had great team members the whole time to help move things forward, including colleagues who have moved on as well as another who is now at the University of Alabama and remains actively involved while providing a bridge to our burgeoning involvement with NOAA’s national flood inundation mapping efforts.
Under Belinda Sturm’s Kansas NSF EPSCoR leadership, the ARISE project started to coalesce not long after the 2019 flood. Considering our established flood mapping capabilities and long-time work with KDEM, all this work blended well with the community impact and natural disaster focus of ARISE.
Concurrently with our ARISE involvement, Professor Xingong Li here at KU has been fantastic. Our labs began a partnership in 2021 that has kicked into high gear all our flood-mapping developments. Xingong has overseen automation and web-enabling of our most critical mapping services and applications, including the development of open-source tools that make the general implementation of our entire approach accessible to others wanting to evaluate the system.
Working with me on ARISE first was graduate student Kenneth Ekpetere, who finished his PhD last spring, and now Master’s student Abu Sabiq Mahdi. Both have brought unique and useful talents to the project and have been a pleasure to work with. Also, PhD student and colleague David Weiss, who, like Dr. Li, isn’t directly involved with ARISE but continues to be invaluable to all of our flood-mapping efforts. Our latest flood mapping work is dovetailing nicely with our ARISE objectives, fostering new in-project collaborations that ultimately should enhance the tools we have been developing for the state.
Q. The 2007 Coffeyville oil refinery spill on the Verdigris River was quite the snafu. Can you walk us through that experience?
A. The FLDPLN model was in late developmental stages when the 2007 flood hit, and I was watching things mostly from the sidelines as the colleague I mentioned earlier was in the thick of it. The stream gauge upstream from the refinery became stuck for a substantial amount of time, then suddenly recovered and lurched upward. This caught the downstream refinery off guard, which had to perform a greatly accelerated shutdown. They didn’t quite make it. When the protective levee overtopped, one storage tank was compromised, sending thousands of gallons of oil into parts of Coffeyville and then down into Oklahoma. It was a huge mess that led to some major community and infrastructure changes in and around the city and refinery. As a consequence of this 2007 disaster, along with severe flooding again in 2019, this stretch of the Verdigris River has been our most studied area for testing.
Q. What about the long-term vision of your work – what’s next?
A. It’s been a long process to get where we are now. The state remains very supportive of our Kansas efforts, and meanwhile we have been nurturing expanded involvement with the NOAA national flood inundation mapping program, leveraging our Kansas system for testing and implementing improvements.
Helping Kansans will always be our top priority, but we are confident that our approach can have durable value elsewhere considering the unsettled state of real-time flood mapping technology. Thanks to ARISE for giving us another stepping stone toward bringing our work to the public sector and to our emergency management officials and personnel.