Our ongoing research focuses on computational remote sensing of terrestrial ecosystem dynamics at local to global spatial scales, and daily to decadal temporal scales. Current research areas in our lab include computational remote sensing, multi-scale land surface phenology, intelligent agriculture, and invasive species and biodiversity.
Computational Remote Sensing
The increasing proliferation of airborne and satellite platforms facilitates unparalleled monitoring of evolving earth system dynamics. The exploding remotely sensed data, along with artificial intelligence and computational advances, provide unique opportunities to improve the understanding of dynamic and complex biogeographical phenomena. In this big data era, our group is interested in developing advanced computational models on high-performance and cloud computing platforms to analyze massive time-series remote sensing data (including optical, RADAR, and LiDAR) for monitoring terrestrial ecosystem dynamics. We are also interested in synthesizing derived remotely sensed products with process-based models and artificial intelligence (particularly deep learning), to understand ecosystem responses to climate change and human activities. We have been working on the Blue Waters petascale supercomputer, Microsoft Azure cloud computing platform, Keeling high-performance computing cluster, and Google Earth Engine to devise computational space-time modeling across varying ecosystems (e.g., agriculture and forest) at regional to global scales. Research includes building large-scale remote sensing data repositories and cyberinfrastructure, fusing optical, RADAR, and LiDAR data of varying resolutions, and advancing model-data integration for understanding ecosystem structures, functions, and responses to climate change.
Multi-scale Land Surface Phenology
Land surface phenology, as the recording of seasonally recurring events (e.g., leaf emergence and senescence), plays a crucial role in characterizing the biogeographical system structures and functions. The phenological dynamics of vegetation over the course of a year regulate the terrestrial gross primary productivity, biogeochemical cycling, water-energy-carbon fluxes, and biotic interactions. Our group integrates multi-scale remote sensing (e.g., UAV, Pheno-Cam, and satellite data), field observations, and process-based phenological models to study land surface phenology across biogeographical systems. We are specifically interested in 1) advancing time-series remote sensing in characterizing land surface phenological dynamics at multi-scale and multi-source levels; 2) enhancing the theoretical understanding of land surface phenological responses to climate and anthropogenic changes; and 3) assessing the feedbacks of land surface phenological dynamics to biogeographical systems and atmosphere (e.g., carbon and water cycling).
The global demand for agricultural crops is rapidly increasing with the continuing growth of the worldwide population, posing significant threats to food security. The escalating crop demand consequently results in the intensification and extensification of agricultural productions, which overstrains ecosystem services and causes drastic environmental degradation. There is an increasing need to transform agricultural systems into resource-efficient systems that are both productive and environmentally sustainable. Built upon the concepts of digital agriculture and precision agriculture, we are interested in advancing monitoring and modeling constituents underlying intelligent agricultural systems, with the synthesis of multi-source remote sensing, crop and ecosystem models, deep learning, and farm experiments. Our research has focused on intelligent field-level crop type mapping, crop phenology and condition retrieval, crop yield estimation and forecasting, and farming practices prioritization from local to global scales. We also investigate the influence of changes in climate and extreme weather on agroecosystems to advance the science underpinning resilient and sustainable agriculture.
Invasive Species and Biodiversity
Invasive species reduce biodiversity, threaten the functioning of natural ecosystems, and cause substantial economic losses at the global scale. The rapid spread of invasive species over landscapes alters fire regimes, causes decline or extinction of native populations, and has devastating impacts on ecological systems. Understanding how the invasive species interact with native communities and change landscapes over space and time remains a grand challenge, particularly with the complications of climate change and human interventions. We use a combination of remote sensing, UAV, field observations, ecological models (e.g., population dynamic models), and computational models (e.g., agent-based models), to investigate patterns, dynamics, causes, and consequences of species invasion at regional to continental scales. We study the characteristics and behaviors of invasive species in both native and non-native habitats, to enhance our understanding of species invasion mechanisms and to inform the cost-effective control and ecosystem restoration strategies. Research includes the modeling of patterns and invasion processes of Tamarisk, an exotic species profoundly altering riparian zones and native biotic communities in the southwestern US. A range of control strategies, particularly bio-control, are evaluated to inform conservation practices for the systematic restoration of affected riparian ecosystems.
The above ongoing research has been generously supported by the University of Illinois at Urbana-Champaign, National Center for Supercomputing Applications, National Science Foundation, United States Department of Agriculture, and Microsoft. We greatly appreciate their support for our research.