Graduate Project
 

Wall-to-Wall Forest Mapping in Southeast and Southcentral Alaska: A New Application of the Gradient Nearest Neighbor Approach

Public Deposited

Downloadable Content

Download PDF
https://ir.library.oregonstate.edu/concern/graduate_projects/c821gs79m

Descriptions

Attribute NameValues
Creator
Abstract
  • Boreal and temperate biomes host nearly half of the earth’s forested ecosystems. The temperate rainforests of the Pacific coast of North America constitute nearly half of all temperate rainforests on earth. Along the northern extent of this region, the perhumid and sub-polar rainforests of southeast and southcentral Alaska are among the largest intact tracts of temperate rainforest in existence. These forests are globally significant for their role in storing and cycling carbon and are regionally and locally valued for their cultural significance, their provision of ecosystem services, and their economic importance. The cumulative impacts of historic management and uncertainties regarding future conditions under a changing climate have largely gone understudied in this important ecosystem. A relative dearth of spatially comprehensive information exists to describe detailed forest attributes at a resolution relevant for both informing management decisions and at an extent necessary to meet regional and national monitoring objectives. This study demonstrates one approach to providing wall-to-wall forest attribute data across the forested areas of coastal southeast and southcentral Alaska using the Gradient Nearest Neighbor (GNN) method. I leverage field surveys from the USDA Forest Service Forest Inventory and Analysis (FIA) program collected across a 26-year timespan (1995-2020) with a set of spatially continuous environmental predictors and annual Landsat Timeseries (LTS) to produce spatially explicit 30-m predictions of forest structure and composition across the region. Spectral harmonization across sensors, a multi-step cloud masking procedure, and the spectral segmentation algorithm, LandTrendr, were implemented in Google Earth Engine (LT-GEE), to produce spatially complete annual imagery for model development. Model predictions were generally more precise and less biased in the boreal forest biome of the western Kenai Peninsula, lending support for further exploration of the LandTrendr-GNN approach to broader monitoring efforts across Interior Alaska. In the coastal temperate rainforest ecoprovince, models tended to truncate distributions and overpredict some observation estimates, but overall agreement revealed relatively strong alignment with design-based estimates in this heterogeneous region.
Resource Type
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Rights Statement
Funding Statement (additional comments about funding)
  • Partial funding for this work was provided by the US Forest Service, Pacific Northwest Research Station.
Publisher
Peer Reviewed
Language

Relationships

Parents:

This work has no parents.

Items

Thumbnail Title Date Uploaded Visibility Actions
file details IrvineDanielR_04Mar2022.pdf 2022-03-21 Public Download