Texas A&M University researchers are investigating the evolutionary effects deforestation has on river landscapes and floodplains | Image: Getty Images

The land around rivers has been attractive and vital to mankind throughout history for its fertile soil for agriculture, easy access to transportation and flood regulation. As civilizations continue to grow, trees are cleared to make room for progress, changing the landscape for better or worse.

Merging engineering with geosciences, an interdisciplinary team from Texas A&M University was awarded a 2019 PESCA grant to investigate how human-induced deforestation influences the evolution of lowland tropical river ecosystems. PESCA grants are awarded by the university’s Division of Research to support scholarly projects that have the potential to lead to external funding by agencies and endowments.

By utilizing real-world data, computer simulation technology and deep learning, the study aims to gain insight about how to preserve and sustain environments around rivers and ecosystems in areas impacted by human intervention.

Dr. Inci Guneralp, associate professor in the Department of Geography, is leading the project.

The team includes Dr. Anthony Filippi, associate professor in the geography department, Dr. Georgianne Moore, associate department head for graduate programs and associate professor in the Department of Ecosystem Science and Management, and Dr. Zhangyang “Atlas” Wang, assistant professor in the Department of Computer Science and Engineering.

“I think this project represents an exciting direction in research in general, how to use deep learning and artificial intelligence techniques to analyze the unprecedentedly large data sets collected in the geosciences,” said Wang.

At its core, deep learning is a branch of artificial intelligence that focuses on emulating human thinking and learning patterns. In doing so, it creates a computer “brain” that is able to draw conclusions and predict patterns without much human guidance. This is particularly helpful when deciphering and translating data sets that are too large or too complicated to do manually.

Before the team can apply deep learning techniques to the large data sets in the study, they must first create a computer simulation model using a small sample of data. This foundational model will help pinpoint key factors for the computer to make note of and act as a set of guidelines to steer the computer’s mind in the right direction.

This, in turn, will allow deep learning to be applied to bigger and future datasets in order to decipher and predict the long-term changes in geomorphic and forest patterns caused by deforestation. Armed with that information, researchers will be able to better understand the current state of threatened lowland tropical forest and river biomes and propose measures and policies to protect their resilience and future sustainability.

“As an engineer, it is exciting to see that our techniques can also help to generate societal values,” said Wang. “I think that the interdisciplinary projects we work on together – whether it’s for global warming, climate changes, deforestation or river flow – all really contribute to the quality of people's lives.”