Using Forest Inventory Data with Landsat 8 Imagery to Map Longleaf Pine Forest Characteristics in Georgia, USA

Abstract

This study improved on previous efforts to map longleaf pine (Pinus palustris) over large areas in the southeastern United States of America by developing new methods that integrate forest inventory data, aerial photography and Landsat 8 imagery to model forest characteristics. Spatial, statistical and machine learning algorithms were used to relate United States Forest Service Forest Inventory and Analysis (FIA) field plot data to relatively normalized Landsat 8 imagery based texture. Modeling algorithms employed include softmax neural networks and multiple hurdle models that combine softmax neural network predictions with linear regression models to estimate key forest characteristics across 2.3 million ha in Georgia, USA. Forest metrics include forest type, basal area and stand density. Results show strong relationships between Landsat 8 imagery based texture and field data (map accuracy > 0.80; square root basal area per ha residual standard errors < 1; natural log transformed trees per ha < 1.081). Model estimates depicting spatially explicit, fine resolution raster surfaces of forest characteristics for multiple coniferous and deciduous species across the study area were created and made available to the public in an online raster database. These products can be integrated with existing tabular, vector and raster databases already being used to guide longleaf pine conservation and restoration in the region.

Spatially Quantifying Forest Loss at Landscape-scale Following a Major Storm Event

“…Variables were calculated within a 4x4 pixel moving window to emulate the spatial extent of our plots. We then used an iterative variable selection script and methods from [17] to select a variable subset for each model.…”

“…Areas where the presence model estimates that no trees are present are set to zero BA. For our tree presence hurdle model, we used a softmax neural network and selected predictive variables from the first ten principal components, following the iterative variable selection R script and methods from [17]. The softmax neural network model output is the probability of each pixel having any tree BA present.…”

“…This 1-m spatial resolution land cover classification is derived from 2013 NAIP imagery and has both probabilistic and most likely classification datasets. Pixels with values above the upper limit of modeled basal area values when the model was limited to areas that fell within the prediction variables domain, as described in [17], were also masked. These pixels mostly occurred in urban areas and around roads that were not masked using the land cover classification product.…”

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

“…Remotely sensed data were re-scaled to a common radiometric scale using an enhanced aggregate no-change regression (EANR) methodology. Similar to aggregate no-change regression (ANR) [6,23], EANR seeks to leverage strong linear relationships among Landsat, Sentinel 2, and aerial image bands to bring finer spatial resolution imagery to the same relative radiometric scale as coarser imagery. The main differences between EANR and ANR are: (1) an added aggregation step to bring finer resolution imagery to the same spatial scale as the reference imagery, (2) a normalization of raw digital number (DN) values, (3) a trimming procedure to mitigate confounding effects of land use/cover changes for images acquired at different dates, and (4) a sampling scheme to extract spectral aggregates within overlapping image boundaries.…”

“…Alternatively, relating remotely sensed data to field measurements of existing forest characteristics to generate fine to medium grained spatially explicit information at a much lower cost than traditional inventory methodologies has shown great promise [5,6]. However, in practice the implementation of these techniques and the use of modeled outputs has seen mixed success.…”