Estimating Forest Characteristics for Longleaf Pine Restoration Using Normalized Remotely Sensed Imagery in Florida USA

Abstract

Effective forest management is predicated on accurate information pertaining to the characteristics and condition of forests. Unfortunately, ground-based information that accurately describes the complex spatial and contextual nature of forests across broad landscapes is cost prohibitive to collect. In this case study we address technical challenges associated with estimating forest characteristics from remotely sensed data by incorporating field plot layouts specifically designed for calibrating models from such data, applying new image normalization procedures to bring images of varying spatial resolutions to a common radiometric scale, and implementing an ensemble generalized additive modeling technique. Image normalization and ensemble models provided accurate estimates of forest types, presence/absence of longleaf pine (Pinus palustris), and tree basal areas and tree densities over a large segment of the panhandle of Florida, USA. This study overcomes several of the major barriers associated with linking remotely sensed imagery with plot data to estimate key forest characteristics over large areas.

Community Experiences in the use of Modified Taungya System for Restoring Degraded Forests and Improving Livelihoods in Ghana

“…3 of the restoration practices of the Commission. Furthermore, managing the rehabilitated forests effectively is predicated on accurate information pertaining to such characteristics of trees as the height and diameter growth (Hogland et al, 2020)…”

21st Century Planning Techniques for Creating Fire-Resilient Forests in the American West

“…Our case study highlights new and novel computational tools and techniques used to create fine-grained timely information that accurately describes the existing condition of forest lands and integrates seamlessly with wildfire risk, spread, and suppression analyses to quantify prescription costs and anticipated revenues for forest plan implementation. Case study objectives include: (1) to quantify the existing forest condition using remotely sensed imagery, field data, the Rocky Mountain Research Station (RMRS) Raster Utility spatial modeling tools [22], and ensemble of generalized additive models (EGAMs) [23]; (2) to spatially define desired future condition and prescriptions that leverage the Potential Operational Delineations (PODs) framework to both harden fire control boundaries and bring POD interiors to a more fire-resilient condition; and (3) to spatially map a cost revenue assessment (CRA) that quantifies the existing supply chain and delivered costs [24] associated with implementing desired future conditions (Figure 1). Using these spatially explicit datasets, the RMRS Raster Utility spatial modeling tools, batch processing, and function modeling [25], we then further demonstrate how managers can quickly and easily prioritize treatments based on wildfire risk, budgetary constraints, and anticipated revenues.…”

“…However, coupled with finely grained remotely sensed information, such as Sentinel 2 imagery [34] and the National Elevation Dataset (NED) [35], data such as multiparty monitoring plots and TIGER road networks and statistical and machine-learning relationships can be leveraged to produce finely grained surfaces depicting the existing forest condition [23] and the costs to move biomass from the forest to a given sawmill [24]. In this study, we demonstrate how existing forest monitoring plots [32], TIGER line files [33], Sentinel 2 imagery [34], and PODs derived from fire managers and common fire modeling tools [11] were used to quantify various forest metrics and treatment costs which in turn were used to inform, justify, and improve restoration decisions.…”

“…To aid in reading, we provide a list of all acronyms used in this study in Appendix A. Topologically Integrated Geographic Encoding and Referencing (TIGER) road inventories [33], and derived products such as those produced by the Landfire project and Wildfire Risk Simulation Software (FSim) exist, these resources alone do not provide the wall-to-wall information needed, at the appropriate scales, to address the finely grained spatially explicit questions or the justification for related prescriptions designed to restore the landscape to a fire-resilient condition. However, coupled with finely grained remotely sensed information, such as Sentinel 2 imagery [34] and the National Elevation Dataset (NED) [35], data such as multiparty monitoring plots and TIGER road networks and statistical and machine-learning relationships can be leveraged to produce finely grained surfaces depicting the existing forest condition [23] and the costs to move biomass from the forest to a given sawmill [24]. In this study, we demonstrate how existing forest monitoring plots [32], TIGER line files [33], Sentinel 2 imagery [34], and PODs derived from fire managers and common fire modeling tools [11] were used to quantify various forest metrics and treatment costs which in turn were used to inform, justify, and improve restoration decisions.…”

Improving Estimates of Natural Resources Using Model-Based Estimators: Impacts of Sample Design, Estimation Technique, and Strengths of Association

“…Managers of natural resources need accurate information describing the resources they manage to make informed decisions [1][2][3]. Owing to high acquisition costs, managers typically employ sampling and estimation techniques to describe various aspects of a given population.…”