Evaluation of vegetation indices for rangeland biomass estimation in the Kimberley area of Western Australia
Other literature type
Schut, A. G. T.
Lamb, D. W.
(issn: 2194-9050, eissn: 2194-9050)
The objective of this paper is to test the relationships between Above Ground Biomass (AGB) and remotely sensed
vegetation indices for AGB assessments in the Kimberley area in Western Australia. For 19 different sites, vegetation
indices were derived from eight Landsat ETM+ scenes over a period of two years (2011–2013). The sites were divided into
three groups (<i>Open plains, Bunch grasses</i> and <i>Spinifex</i>) based on similarities in dominant vegetation types. Dry and green
biomass fractions were measured at these sites. Single and multiple regression relationships between vegetation indices and
green and total AGB were calibrated and validated using a "leave site out" cross validation. Four tests were compared: (1)
relationships between AGB and vegetation indices combining all sites; (2) separate relationships per site group; (3) multiple
regressions including selected vegetation indices per site group; and (4) as in 3 but including rainfall and elevation data.
Results indicate that relationships based on single vegetation indices are moderately accurate for green biomass in wide open
plains covered with annual grasses. The cross-validation results for green AGB improved for a combination of indices for
the <i>Open plains</i> and <i>Bunch grasses</i> sites, but not for <i>Spinifex</i> sites. When rainfall and elevation data are included, cross
validation improved slightly with a Q<sup>2</sup> of 0.49–0.72 for <i>Open plains</i> and <i>Bunch grasses</i> sites respectively. Cross validation
results for total AGB were moderately accurate (Q<sup>2</sup> of 0.41) for <i>Open plains</i> but weak or absent for other site groups despite
good calibration results, indicating strong influence of site-specific factors.