Forest Cover Index for tree cover detection using Landsat-7 multispectral imagery

Abstract

Forest cover monitoring using satellite imagery is important to U.S. military terrain analysis. Mobility models, cover and concealment, and precise surface feature delineation all rely on an accurate forest/nonforest cover layer. However, the challenge remains in distinguishing trees from other vegetative land covers when relying on single date imagery. A Forest Cover Index (FCI) algorithm was previously developed on Worldview-2 imagery and was designed to separate forest cover from nonforest covered areas. In this research, the FCI algorithm was applied to Landsat-7 imagery using the analogous red (636-673 nanometers (nm)) and near-infrared bands (851-878 nm) from a peak summer image on 16 August 2012. The results obtained with Landsat-7 imagery proved satisfactory with an overall accuracy (tree cover versus non-tree cover) of > 83 percent, according to testing with two different accuracy assessment tests. The application of the FCI to Landsat-7 imagery broadens the applicability of the FCI to freely available imagery.

A multi-biome study of tree cover detection using the Forest Cover Index

“…While there was no consistent difference in seasonal accuracy, perhaps due to the disparate nature of the biome conditions, seasonal changes such as flooding, snow, and tree senescence did reduce the accuracy of the FCI in specific images. This aligns with previous research by Becker et al (2018) and Feliciano-Cruz et al (2019) who found clear delineation with the FCI between tree cover and the other land covers in the image, including agriculture and built-up land, Becker et al (2018) also found a seasonal component.…”

“…Accuracies were lower in comparison to Becker et al (2018); however, Becker et al (2018) only examined one study site and used a less robust non-random sampling accuracy assessment technique, which could have inflated their accuracy. The Kappa values of this research were consistent with the Kappa value that Feliciano-Cruz et al (2019) derived from an error analysis using random sampling.…”

“…This research found that images where tree cover was clearly delineated from other vegetation and built-up land covers performed similar to Becker et al (2018) andFeliciano-Cruz et al (2019). This was apparent in the image from the tropical and subtropical grasslands, savannas and shrublands biome, which had clearly delineated tree cover alongside agriculture (Figure 6).…”

“…The NDVI was used as a preliminary step to mask non-vegetation from the FCI input (Tucker 1979;Rouse 1973), because Becker et al (2018) and Feliciano-Cruz et al (2019) found confusion between tree cover and land covers that exhibited similar low FCI values, such as water, which led to the addition of the NDVI pre-mask in this research as a first step to eliminate land covers that yielded low FCI values. A default NDVI threshold of 0.4 was used to mask infrastructure, water, and other nonvegetative land covers.…”

“…Future research will focus on automating the identification of the threshold that separates tree from other land covers, exploring use of the output for machine learning applications, and incorporating ancillary data such as digital surface models and existing tree cover maps. While the Normalized Difference Vegetation Index (NDVI) is often used to study vegetation, prior research developed the Forest Cover Index (FCI) using the red and near infrared (NIR) bands from multispectral sensors to further distinguish tree cover within mixed vegetation (Becker et al 2018;Feliciano-Cruz et al 2019). The FCI was previously tested at the U.S. Department of Agriculture's Beltsville Agricultural Research Center in Beltsville, MD on urban land cover, closed canopy forest cover, and dense agriculture (Becker et al 2018).…”