Detecting Fire and Grazing Patterns in Tallgrass Prairie Using Spectral Mixture Analysis

Abstract

Global grasslands are typically under management practices (such as fire and grazing) that alter nutrient cycling, ecosystem composition, and distribution of organic matter from the unmanaged condition. We evaluated landscape‐level response to fire and grazing treatments in the Konza Tallgrass Prairie Research Natural Area, Kansas, using spectral mixture analysis of Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) data acquired 31 August 1990. Spectral mixture analysis derives the fractional abundances of spectrally unique components in the landscape. The reflectance spectra of these components are called endmembers. Endmember fractions values were compared against ground values of live biomass, current standing dead biomass, and litter for 12 watersheds. Analysis of variance (ANOVA) was performed on 37 watersheds with known burning and grazing histories for each of the remote sensing variables. Seven endmembers were selected from the AVIRIS data using a manual endmember selection method: nonphotosynthetic vegetation (NPV), soil, rock, shade, and three green vegetation endmembers (GV1, GV2, and GV3). Each vegetation endmember correlated differently to biomass measurements and revealed unique relationships to management treatments. From regressions, ANOVAs, and image analysis, these three endmembers were inferred to represent canopy vertical structure or leaf area index (LAI), greenness, and fractional cover of grass, respectively. There was a stronger relationship between the sum of GV1 and GV3 fractions and live grass biomass values than there was with the (unsummed) individual fractions. In an ANOVA, the sum separated both burn and grazing treatments as well as the treatment interaction. The NPV fraction was strongly correlated with ground measurements of litter and standing dead biomass, and significantly separated burn treatments. The soil fraction differentiated grazing treatments, and analysis of the soil fraction image revealed a spatial coherence of grazing patterns along drainages. Similar analyses were perfomed on the Normalized Difference Vegetation Index (NDVI), a commonly used two‐band index computed from red and near‐infrared reflectance. NDVI, shown in previous studies to estimate the fraction of photosynthetically active radiation absorbed by green vegetation (FPAR), was a poor indicator of canopy biomass, but it successfully separated fire treatments. Broad‐scale assessment of the state and structure of managed grassland systems requires the identification of several indicator variables. Spectral mixture analysis, unlike NDVI, not only separated treatments but also allowed for the identification of five remotely sensible factors affected by the management treatments, namely, vertical structure, percentage cover or patchiness, greenness, and distribution of soil and litter.

Grass species differentiation through canopy hyperspectral reflectance

“…Mixed canopies introduce the additional challenge of interactive effects (Peterson et al 1988, Wessman et al 1997. Paradoxically, the two situations in which linear spectral mixture analysis clearly failed had an extremely large proportion of the C 4 component, which, a priori, would be the easiest to estimate, based on isolated species signatures.…”

“…Linear spectral mixture analysis linearly estimates the proportion of individual identities, in this case species, from the mixing of their reflectance spectra (Wessman et al 1997, Clark 1999, Bateson et al 2000, Dennison and Roberts 2003. This technique was applied to evaluate mixed canopies.…”

Monitoring desertification in a Savannah region in Sudan using Landsat images and spectral mixture analysis

“…Elmore et al, 2000;Wessman et al, 1997), whereas others employ libraries of endmember spectra (library endmember), which are produced from reflectance measurement in a laboratory (e.g. Smith et al, 1990).…”

Spectral mixture analyses of hyperspectral data acquired using a tethered balloon

“…Ground feature areal fraction information has been successfully provided by spectral mixture interpretation in many studies (e.g. Garcia-Haro et al, 1996;Jasinski & Eagleson, 1990;Kootwijk et al, 1995;Meer, 1995;Wessman et al, 1997). In addition, SMA theory has been used to quantify vegetation biomass, fraction of absorbed photosynthetically active radiation (FAPAR), leaf area index (LAI), and net primary productivity (NPP) (e.g.…”