Frameworks: Cyberinfrastructure for Urban Tree Resilience and Environmental Equity (uTREE)

With 89% of the U.S. population and 68% of the world population projected to live in cities by 2050, rising urbanization will worsen already significant environmental challenges such as excessive heat, poor air quality, and rainwater runoff. With urban trees recognized for their potential to ameliorate these issues, many large cities are allocating resources for obtaining the accurate up-to-date tree inventories necessary for managing urban forests. However, cities currently collect data through labor-intensive, one-time manual surveys of public trees or via coarse low accuracy canopy cover estimation. In addition, these methods lack access to the significant portion of trees on private property. As a result, urban research and management communities lack critical information about tree density, tree species, locations of trees across different land types, and changes of tree counts over time and events. This project will address time-critical gaps of data completeness, accuracy, and equity and of tool availability with the development of the first-ever cloud-based cyberinfrastructure (CI) supporting national (and potentially worldwide) urban tree inventory estimation that in turn will improve multidisciplinary urban research, engineering, and planning to yield safer and more livable cities into the future. Moreover, sustaining the CI will need gradually less computation as the system benefits from trained generative modeling and transfer learning. In addition to broad dissemination to science and engineering communities and to urban stakeholders and practitioners, the effort will target recruitment of underrepresented minorities through campus programs in order to broaden participation of diverse graduate students in ecological and computer science-related disciplines. Researchers armed with the increased computational power and advancement of recent novel urban canopy parameterization models, urban ecological models and urban social science linkages are now providing dramatically improved abilities to perform comprehensive predictions. However, the lack of high-resolution tree data integrated with such powerful simulation tools, particularly in under-resourced communities, remains a significant barrier to transformative impacts of urban canopy estimations and predictions. This CI will use a spatio-temporal generative artificial intelligence approach capable of determining highly accurate locations and species of urban trees in cities. Further, the team will address non-trivial challenges in data management and CI interoperability to enable research workflows that seamlessly integrate big remote sensing data, desktop simulation tools, HPC-optimized simulation models, and web-based interactive dashboards to provide a set of diverse tools capable of performing mitigation simulations and visualizations. Building on team members' prior experience and a prior NSF Elements project, current tools will be extended to automatically link urban tree-related data to the community’s urban parameterization models. The project will impact a broad community represented by committed partners of WRF/NCAR, National DOE Urban Integrated Field Labs, WUDAPT, NSF LTERs, NSF Accelnet: GLASSNET, i-Tree, CRTI, KIB/KAB, and ArbNet. The uTREE CI will be extensible, portable, and scalable to serve a large and multi-disciplinary community of 60,000 researchers and 500,000 urban practitioners and at the outset will impact an initial 3.6M urban citizens and 9M trees (Chicago and Indianapolis). Next-generation urban computational and ecological researchers and managers will gain critical exposure to data science concepts in urban tree research and computational data analytics and visualization. Nearly one-third of the Earth's land surface is covered by forests, which host the majority of terrestrial biodiversity. Accurate mapping and monitoring of forests across large regions and over time is critical for mitigating climate and natural hazards, managing natural resources and protection of vital ecosystems. While traditional ground-based measurements of plant species and size provide the most accurate data on forest structure and above ground biomass, these methods become impractical when covering large areas with high-frequency repeat cycles. Airborne and Space-based remote sensing techniques provide a timely and cost-effective way to assess forest structure and biomass on regional to global scales. Satellite missions from NASA and ESA have sensors that gather data with significantly more frequent repeat cycles compared to in situ measurements or aerial surveys. While these satellite missions offer global coverage, some provide only sparse data on forest structure and need to be combined with other data sources for producing comprehensive and accurate wall-to-wall maps. There is a lack of efficient frameworks that utilize multi-source remote sensing data to produce wall-to-wall forest structure or above ground biomass at temporal and spatial scales necessary for effective forest management or use in hazard mitigation and monitoring applications. Without significant improvements to existing methodologies and looking beyond traditional data sources, efficient and accurate monitoring of forest structure and above ground biomass will remain limited. OpenForest4D will allow a wide range of users to generate on-demand and up-to-date research-grade forest structure and above ground biomass estimates across a range of timescales. This will be achieved by applying novel statistical models and artificial intelligence methodologies on a fusion of multi-source remote sensing data from ground, airborne and spaceborne platforms. Providing these cyberinfrastructure services through easily accessible interactive web-based interfaces, along with educational resources focused on the underlying domain science, will facilitate transformative research in forest sciences and ecology and encourage broad community participation. OpenForest4D's web-based educational resources, published curriculum materials, and live webinars will help develop a diverse, globally competitive STEM workforce. This award by the Office of Advanced Cyberinfrastructure is jointly supported by NSF's National Discovery Cloud for Climate initiative. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.