Jackrabbit in the Spotlight / Kamal Albousafi

If terms like Kalman filtering, normalized difference vegetation index and recursive algorithms sound like a foreign language to you, then you’re probably not watching math videos for fun.

Kamal Albousafi, a senior math major from Sioux Falls, does. “I love math,” said the outgoing Albousafi, who does admit to being a bit nerdy when it comes to math. “The deductive power to use information and data to gain insight and understanding drew me to math and statistics,” he said.

Albousafi found his nerdiness very helpful this summer as a Research Experience for Undergraduates program participant at South Dakota State University. Working under math faculty Hossein Moradi and Jung-Han Kimn, Albousafi used satellite images and dynamic modeling to predict crop health.

The eventual goal is to be able to use the images and more intricate modeling to predict crop yield.

“This research has several practical applications,” he said. “For example, it could provide farmers with crop yield estimates weeks in advance, helping them make informed decisions during harvest. It could also identify potential crop health issues early on, allowing for corrective actions that might prevent devastating losses—and, in turn, help combat food scarcity.”

Research part of REU program at State

Albousafi, who was born and raised in Sioux Falls, never had any interaction with crops prior to doing a smaller satellite data research project under Moradi at the beginning of the year. This summer he took a 10-week dive into it.

Research Experience for Undergraduates is a 10-week, National Science Foundation-funded program that provides a stipend for selected students to work with faculty at a host institution. 

Four SDSU students (mathematics and statistics, computer sciences, physics, and mechanical engineering) were part of the program this year, all working on separate projects but sharing workspace in a large office space in the Chicoine Architecture, Mathematics and Engineering Hall.

Kimn said, “Kamal was an outstanding leader during our summer REU program. The shift from classwork to research can be challenging, and I've often found it difficult and sometimes discouraging. Kamal's character and personality motivated the other students to persevere through the occasionally tedious research process. 

“His positivity and quick wit brought energy and excitement to the REU program. when I shared a simple, lighthearted anecdote to boost morale, he was often quick to add his own upbeat reaction, lifting everyone’s spirits even more. 

“However, it was his ability to stay focused and deliver results that truly made him an exceptional role model. He helped ensure the 10 weeks were both highly productive and enjoyable.”

Model proves 80% accuracy

From NASA’s Landsat 5 and 8 satellites, Albousafi took 40 years’ worth of overhead images of a farm in Edmunds County from— some 800 images — and extracted each image’s average normalized difference vegetation index, a standard measurement for quantifying vegetation greenness. 

This index gives a metric to measure the health of crops, in this case corn and soybeans, at a given time.

He then obtained historical temperature, precipitation, and drought records for the corresponding time period. Landsat produces an image every 16 days. Using his model, Albousafi would then use the historical climate data to predict what the normalized difference vegetation index would be for the next time the satellite passed over. 

After training the model on data from 1980 to 2020, Albousafi then applied the predictive model to 2021-23 and “got solid results,” predicting with 80% accuracy, he said.

The modeling and predicting were done using an 85-acre field in Edmunds County, west of Aberdeen, and was accurate regardless of crop rotation or harvest timing, he said.

Albousafi had high praise for the Research Experience for Undergraduates program. Though he only knew one of the four fellow students beforehand, he said they all became “quite chummy” by the end of the summer. To that he credits professor Kimn, who would incentivize the students’ interaction with daily non-math trivia, social events, and his joking yet supportive personality.

Provides opportunity to give poster presentation

Additionally, Albousafi was able to give a poster presentation on his work at the summer South Dakota EPSCoR conference at the Sioux Falls Arena.

He hopes that is just the beginning of his presentations. Albousafi also has his sights on the SDSU Data Science Symposium in February.

This semester Albousafi is taking 15 credits, which includes two graduate courses. He will earn his bachelor’s degree in May 2025, and through the accelerated option he will earn a master’s degree in math and statistics in May 2026.

In the spring semester he will pick up his summer 2024 research program again in conjunction with a senior capstone class. He hopes to write a “couple algorithms that will allow for more robust predictions allowing for non-linear and other unordinary relationships between time points.

“Right now, the model is limited to using simple, straight-line relationships between images to estimate crop health. However, since the data tends to follow more of a repeating, cyclical pattern, the model could be more accurate if it could account for more complex, non-linear interactions between the images.”

Ideally, that could give Albousafi the material he needs to publish a paper in a scientific journal.

Ultimately, he would like to find a job in precision agriculture where he could turn his math nerdiness into a well-paying yet satisfying occupation.