Abstract:
Artificial Intelligence (AI) is revolutionizing the study of forest ecosystems, enabling researchers to analyze vast amounts of biodiversity and environmental data with unprecedented precision. This presentation explores how AI-driven methodologies, including machine learning and advanced computing, are transforming forest migration research. By leveraging global datasets and computational power, we uncover patterns of forest shifts, assess biodiversity changes, and predict future ecological trends. The integration of AI into forestry science not only enhances conservation efforts but also provides valuable insights for policymakers, industries, and communities reliant on forest ecosystems. This talk will highlight key findings from global research initiatives, such as the Forest Advanced Computing and Artificial Intelligence Laboratory (FACAI) and Science-i, demonstrating AI’s critical role in understanding and managing the dynamic nature of our forests in the face of environmental change.
Bio:
Dr. Jingjing Liang is a globally recognized leader in quantitative forest ecology, AI-driven environmental research, and international scientific collaboration. He is currently an Associate Professor at Purdue University and an International Consultant at FAO, with over 20 years of expertise in sustainable forest management, biodiversity conservation, and AI-enhanced ecosystem monitoring. As the Founder of Science-i and Coordinator of the Global Forest Biodiversity Initiative (GFBI), he leads a network of 500+ researchers across 55 countries, driving global collaboration in big data forestry and biodiversity science.
Dr. Liang’s research is widely published, with 93 articles in top-tier journals such as Science, Nature, and PNAS, accumulating 6,233 citations (h-index: 38, i10-index: 71). His work informs global forest policy and conservation strategies, securing over $6 million in research funding ($2.7 million as PI). At FAO, his contributions to AIM4Forests and For-Growth projects of the United Nations advance forest growth monitoring and carbon quantification, playing a key role in climate solutions.
A dedicated mentor and advocate for inclusive science, AI-driven forestry, and climate resilience, Dr. Liang has supported 2,878+ researchers worldwide. He also serves on the Forests Remaining Forests Committee, shaping the future of REDD+ crediting by developing a framework to recognize carbon removals from forests that remain forests—an area historically excluded due to monitoring challenges. His expertise directly influences the next iteration of the global carbon crediting standard, ensuring it fully accounts for the potential of forest-based carbon sequestration in climate mitigation efforts.
Summary:
AI - Game-changing tool for forest ecology
FACAI Lab: https://ag.purdue.edu/facai
Collects global data of forest health and growth
Quantified relationship between tree species diversity and forest productivity (2016)
Estimated number of tree species on Earth (2021)
We’ve discovered only a fraction of the species out there
Many tropical species have not yet been discovered
Estimate: 73k tree species total (60k discovered so far)
Deforestation may lead some species to go extinct before they’re discovered
Mapping locations and characteristics of planted forests across the world
Differentiating natural from planted forest rests on their different spatial structure (planted forests are more structurally homogeneous)
Identified species richness across the world
Forest Migration
Climate change has caused forests to
Die off in hot areas due to high heat
Establish in cold areas due to more favorable climate
Challenge: different agencies use very different forest type classification schemes (e.g. US vs Canada)
Used ground observed forest inventory data to establish a new ML forest type classification and forest type migration model
Hypothesis: forest types are portfolios of tree species (like a Markowitz portfolio)
Forest type shift is
A combination of the shifts of individual species +
Covariance term: more correlated species accelerate migration
AI-based forest type classification
Data
Forest trees
Covariates: bioclimate, topography, forest height, anthropogenic
Auto-encoder compresses these characteristics and then apply k-means classifier to cluster regions by type
Major North American forest types: boreal, east, west + many sub-types
Used classification to measure migration speed of forest types
Boreal forests have been moving at > 100 km/decade (major growth in Western Canada and Alaska)
Forest type migration speed is usually different from the speed of individual species: usually much faster
Migration has many effects
Socio-cultural
Disruption to timber supply
GFI-3D: Largest and most upto-date Global Forest Inventory Databases
Science-i: https://science-i.org/
>2m sample plots
92 countries
50k tree species
400 data contributors
Infrastructure for collecting and processing forest data