Dr. David E. Willmes on Solving Global Food Insecurity


David Willmes (top) and Danny Nsouli (bottom). Graphic: Danny Nsouli

Interviewer: Danny Nsouli

Welcome to the latest installment of the Knowledge-Driven Podcast. In this series, Cyber Security Software Engineer Danny Nsouli interviews technical leaders at MITRE who have made knowledge sharing and collaboration an integral part of their practice.

Approximately 2 billion people lack regular access to sufficient quality food. The issue of global food insecurity is one that is constantly being looked at and, fortunately, MITRE is stepping up to help mitigate this problem. Today, Dr. David Willmes will be discussing his team’s project and how it uses significant crop and consumption data to better understand the factors at play in this global problem.

Click below to listen to podcast:


Podcast Transcript
Danny: Hello, everyone. My name is Danny Nsouli, and welcome to MITRE’s Knowledge-Driven Podcast. Today, I will be discussing global food insecurity with Dr. David E. Willmes. David, would you like to introduce yourself and tell the listeners a little bit about your role and experience at MITRE?
David: Sure. As Danny mentioned, my name is David Willmes, and I am a Principal Artificial Intelligence Advisor here at MITRE. I am in the Advanced Processing and Exportation Technologies Department here. I’ve been working at MITRE for a little over a year, but I’ve been involved with Federally Funded Research and Development Centers for over a decade now.
Danny: Great, and David, can you give us a little bit of background on the term global food insecurity and how it’s affecting different countries around the world?
David: Yeah. Sure. First of all, food security is characterized by, let’s say, three different factors. There’s food availability, food access, and food use. So [by] food availability, we mean making sure that there’s just enough food that is … enough crops are grown, enough food is produced. Food access is making sure that the population has the ability to obtain that food through, typically, through purchasing power, and food use is making sure that that population has the knowledge and the tools to properly prepare the food and to meet their caloric and nutritional requirements. So that means making sure that there’s, say, adequate water and sanitation. So food availability, food access, and food use—those are all necessary.
David: So you can infer from this that simply growing more food will not necessarily result in more food security. Right? You need to have the ability to properly prepare that food. So let’s …maybe you put some numbers here. Right? Today, there’s about one in nine people [who] are undernourished. That’s about 820 million people. Another one-and-a-half billion people face moderate food insecurity. So that’s about a third of the world that has some sort of at least moderate to severe food insecurity.
David: This number of hungry and malnourished people across the globe is strongly correlated to the rise in human conflict and forced migration. So this food scarcity, not only is it caused by a lot of violent conflict, it also has a role as a driver in a lot of this conflict. That has not been cited in quite as much detail. So there’s this reinforcing loop between food insecurity and conflict. More food insecurity results in more conflict, which can result in more food insecurity, and it kind of spirals out of control that way.
Danny: So you mentioned human conflict, but are there other external factors besides that you found also affect food security?
David: Yes, absolutely. Think for a moment that while the world as a whole likely will produce enough food, say, for the next 10 years or so, the issue is more the changes that are going to be occurring in the next 10 to 20 years. One of these is the climate. The climate can affect food production through crop yield changes. For example, for about every one degree of Celsius rise in mean temperature, there’s about a 10% drop in crop yields. Of course, that’s very variable based on what crop there is and where in the world, but figure that’s a pretty reasonable number.
David: So it affects food production but also potentially increases the frequency and intensity of natural disasters that may affect the food supply, both from affecting the crops as well as affecting the dissemination of those crops through the supply lines. Climate change can also increase human conflict and increase the migration of vulnerable populations. So in this way, you can see climate change as being a threat multiplier, where we already have all these issues with conflict and migration, and having this just exacerbate that problem.
Danny: That’s interesting. I didn’t realize the scope of that factor.
David: It is going to be a, I think, a big issue going forward in the next few decades.
Danny: I understand that you are currently working on a project to help mitigate these problems. Could you give us an overview of how it factors into this issue?
David: Yes. That’s right. Our research project is called AURAC, which stands for Anticipatory Understanding for Resilient Architecture to Climate. We’re developing a large system that traces crops, from the planting of the crops all the way to the consumption of the crops by the, consumption of that food, by the food-insecure population. For example, the first component is classifying and identifying crop land for staple crops. The next component is modeling crop yields under different environmental conditions, including different climatic conditions. We are then aggregating environmental and food system data and identifying and characterizing the food system vulnerabilities through building predictive models.
Danny: So by that, I get the sense that your project is trying to strengthen global food security by using the benefits of data and predictive models as to not leave as much to chance. Is that correct?
David: Yes. That’s a good way of putting it. A lot of our technologies are based on, say, deep learning and artificial intelligence. There’s also a lot of causal modeling that’s going on with our predictive models, and there’s also a lot of crop yield simulation work that we are utilizing based on a lot of the crop simulation software that’s out there.
Danny: I also wanted to ask just simply, how did this idea come about within your team at MITRE?
David: We have a group here led by Dr. Alex Schlichting, who’s kind of … who came to our group and asked for some ideas on trying to understand how to mitigate a lot of the climate change issues that he’s been seeing. I’m working now in computer vision, so a lot of our ideas had to do with applying a lot of computer vision and deep learning technologies to this problem. So from there, one of the main issues associated with climate is food insecurity. So how can we use a lot of the technologies that we at MITRE have a lot of expertise in to try to focus on those problems?
Danny: That’s great, and can you also talk about why you think MITRE specifically is suited for such a project?
David: Sure. Well, MITRE is a system integrator FFRDC, so because we are a system integrator, we are well-positioned to serve across different government sponsors, academia, NGOs, industry to validate and verify existing models. So we have that role to be able to validate and verify. We are also connected across different federal sponsors, and US, the US is a major source of agricultural technology and innovation. So while we can provide information to United States government sponsors, they will then be armed with a lot of the information that we provide to try to assist these other countries and their food security problems. Hopefully, that can help mitigate a lot of the food security hotspots that might occur down the road. We are also a trusted third party. So due to our FFRDC nonprofit status, we are well-positioned to handle sensitive and proprietary data sources and integrate those into an existing framework that we are building.
Danny: Do you have any real-world use cases you’ve been working on?
David: Sure. One of our use cases is looking at the wheat bread basket of Northern India. About 60% of the food in the world is grown in one of six major bread baskets, and Northern India is bread basket for wheat. So that’s our first use case. Now, India is not currently under a severe food crisis like some other parts of the world. So while it might not be the scenario that I’d think would have the most immediate impact, down the road, in a decade or so, as climatic changes affect India, we want to make sure that they don’t become more food insecure. The economic model for Indian farms is such that the government buys a Minimum Support Price to farmers. We call that the MSP. So the MSP to farmers allows them to purchase produce at competitive prices for the food-insecure population.
David: This food-insecure population is determined by a ration card system. This population can then buy that food at a subsidized price. Now, there’s one MSP for each crop across the whole country of India. We can model how changes to the MSP for wheat can influence whether the farmers sell the wheat to the government shops or on the open market. We, in our modeling, we are able to modify the MSP through some lever and see how that affects food security for the Indian population. You would think that just, “Oh. Let’s increase the MSP, and that makes food more available.” However, if you increase the MSP too high for wheat, it can incentivize the farmers to grow wheat at the expense of other crops, and those other crops may be necessary to provide the vulnerable population the nutrients that it needs.
David: So there’s a lot of interplay between the MSP price, farm production, incentivizing the farmers, storage and transportation of the crops. All of that kind of is built into our model, called a system dynamics model. So modifying this MSP can actually have a bunch of cascading effects on the use of the fields to grow different crops and affect the crop yields that way. There’s actually a current situation in India right now where the government of India, just a few months ago, decided that they’re going to eliminate this Minimum Support Price for the farmers. I’m not sure if you’ve seen this in the news, but there’s been a lot of protests by these farmers. They’ve been going through the Capitol in Delhi and just protesting in the streets the fact that they’re losing this MSP.
David: And so this Minimum Support Price to the farmers is necessary for them to make a living. So we are trying to make sure, then, that we can provide some information to allow the government in India to recognize that maybe removing this might cause more food insecurity based on the farmers not being able to provide enough crop. So there’s a lot of different economic factors, factors involving the production on the farms, the transport and storage of the crops. All these different factors affect food security in India and not just obtaining the crop yields.
Danny: Interesting. Jumping off that, would you agree that the hardest part about your project is the diversity in obstacles and variables you have to keep in mind for each country?
David: Absolutely, and we are focusing on the wheat bread basket of Northern India right now, but we are modularizing our system so we can take different factors that affect different parts of the world and swap them out and still get a functioning model. For example, this whole idea about the Minimum Support Price is pretty specific to India. So that would be a module. The economic module there would be swapped out for something in a different part of the world. So in parts of the world that have a lot of human conflict right now, in those cases, we are going to be building different modules to take into account how the food-insecure population is going to be able to obtain the nutrients that they need.
Danny: Going back to your project development, can you paint us a picture of what this project looks like in action and how the developers interact with it?
David: Yes. We are actually building a dashboard that allows a user to modify certain policy levers, say, and other factors that will affect food security and see how those different changes to those levers affect certain districts. So this dashboard is actually a map for the user, so the user can see how different districts are going to be affected. The models that go into that include, I mentioned economic models of production, dissemination through transportation and storage facilities, and different proneness to flooding and other natural disasters. So a user can actually see where hotspots might occur if they start changing things like the MSP in India, for example, or the availability of storage in a certain district.
Danny: Will there be opportunities in the future for an average person, like a listener of this podcast, to be able to use this project for themselves?
David: Yes. We hope to be able to take this dashboard that we are currently building and, say, and that we’d like to be able to beta test with people within MITRE first, but after we get a functioning dashboard that we feel comfortable with, we will share that to the outside world and let different users experiment with the dashboard and see how their changes to certain policies and certain capabilities will affect food security.
Danny: That’s really cool, and David, is there anything else you’d like to add about your project before we go?
David: Yes. I would like to emphasize our use of artificial intelligence and building deep neural networks to do a lot of the classification that we’re doing. We are trying to classify crop land based on satellite remote-sensing imagery. This is necessary, because in a lot of these conflicted regions, regions where there’s a lot of environmental crises, the agriculture is through small family farms and not the big industrial farming that you might expect from someplace like in the United States. So being able to identify crop land in a lot of these areas is a bit more difficult than just looking at a government database. What we’re trying to do, then, is use a lot of this deep-learning technology to get a good understanding of where the crop fields are, and we’re using that as an input to our crop yield models that can then be used to understand where food-insecure areas may be.
Danny: That’s great. It’s nice to hear that there’s a lot of attention to detail going into this.
David: Yes. I think that’s a very important aspect of the project.
Danny: All right. Well, thank you for coming on to discuss your work. I’d like to give a quick thank you to MITRE and the Knowledge-Driven Enterprise for making this show possible. Again, thank you, David, for coming on to share. I’m sure our listeners learned a lot.
David: My pleasure. Thank you, Danny.


Danny Nsouli is an Associate Cyber Security Software Engineer. He has a passion for computer graphics and enjoys learning about front-end solutions for consumer-facing project components such as data visualizations.

© 2021 The MITRE Corporation. All rights reserved. Approved for public release.  Distribution unlimited. Case number 21-0437

MITRE’s mission-driven team is dedicated to solving problems for a safer world. Learn more about MITRE.

See also:

Project Demodocus: Bringing Accessibility to the Masses

The Power of Geospatial Data In Developing Countries

Jen Choi and Josh LeFevre and the power of “Yes, And”

Getting Students Excited About STEM (and MITRE), with Willie Hill

Project Demodocus: Bringing Accessibility to the Masses

Turkeys vs Swans, with Imanuel Portalatin

Justin Brunelle: Lessons from MITRE’s Innovation Program

Rachel Mayer on the Fight Against Maternal Mortality 

Theodore Wilson: Thinking Like a Turtle

Marcie Zaharee and MITRE’s Open Innovation Challenge

Dan Frisk and Paula Randall on bringing innovation to government

Interview with Awais Sheikh on Deciphering Business Process Innovation

Interview with Jackie Morin on her journey from intern to senior engineer

Interview with Jay Crossler on why passion is the key to success

Interview with Dan Ward, Rachel Gregorio, and Jessica Yu on MITRE’s Innovation Toolkit

Interview with Tammy Freeman on Redefining Innovation

Interview with Dr. Sanith Wijesinghe on Agile Connected Government

Is This a Wolf? Understanding Bias in Machine Learning


Pin It on Pinterest

Share This