Large-area Enhanced Forest Inventory modeled on 3-D Airborne Laser Scanning data

Project Description Maine’s economy depends heavily on its forest resource base: it accounts for over 6% of the total GDP and has an estimated total annual economic impact of $8-10 billion. The sound, scientifically-based management of the forest resource requires a significant investment in inventory programs. Current methods for monitoring the forest are limited as using traditional ground-based sampling techniques are expensive ($3 million annually spent in Maine alone), imprecise, not real-time, difficult, and spatially coarse. New capabilities for the aerial collection of high quality, detailed remotely sensed information on 3-D forest structure over large areas are providing inventory information more accurately, efficiently and at lower cost relative to traditional methods. Working with the large volumes of data and increasingly complex analytics requires computer science research and development designed to make progress on processing workflow solutions.

---

New capabilities for the aerial collection of high quality, detailed remotely sensed information on 3-D forest structure over large areas are providing inventory information more accurately, efficiently and at lower cost relative to traditional field methods. Over the last decade, studies have demonstrated the capabilities of Airborne Laser Scanning (ALS) as a potential operational tool in forest inventory, providing accurate estimates of a wide range of commonly-used measurements. There is a fast-growing need for leveraging the growing collection of ALS data across Maine for usable and reliable Enhanced Forest Inventory (EFI) data products to support management and decision-making in the state’s forest industry. However, working with continually evolving sensor technology, large volumes of data and increasingly complex analytics requires research and development designed to make progress on workflow solutions and advance processing methodologies.

Area-based extrapolation of EFI variables requires a statistical model that relates a sample of plot measurements to a set of explanatory covariates calculated from the 3-D profiles of gridded ALS data. This project will be carried out at the University of Maine as a collaboration between the Advanced Computing Group (ACG) and the School of Forest Resources. The software development will involve porting and rewriting the code for calculating ALS metrics, from old and unsupported code written for Windows, to code that will run in a parallel computing environment on the ACG’s computer cluster. Model development will make use of extensive plot-based inventory information gathered from existing sources previously measured by SFR investigator teams, our Maine-based forest landowner cooperators and other State and Federal forest inventory programs. The result will be the software tools and computing workflow to accurately and efficiently produce geospatial data deliverables of inventory metrics to be used by forest industry practitioners, agency decision-makers and other researchers.