Performance of ratio‐based, soil‐adjusted and atmospherically corrected multispectral vegetation indices in predicting herbaceous aboveground biomass in a <i>Colophospermum mopane</i> tree–shrub savanna

Walter Svinurai, Abubeker Hassen, Eyob Habte Tesfamariam, Abel Ramoelo

Grass and Forage Science · 2018 · 15 citations · 37 references

Concepts

Abstract

Abstract Accurate and near‐real‐time estimation of herbaceous aboveground biomass ( AGB ) at farm level is crucial for monitoring pasture production and proactive management of stock in semiarid rangelands. Despite its importance, remote sensing has been rarely used by range ecologists and managers in Zimbabwe. This study aimed at assessing the performance of classical multispectral vegetation indices ( MVI s) when either singly regressed with measured herbaceous AGB or combined with other visible spectral bands in predicting herbaceous AGB in a Colophospermum mopane savannah. Field herbaceous AGB and corresponding Landsat 8 Operational Land Imager visible spectral data were collected during the 2016–2017 rainy season. Relationships between measured AGB and classical MVI s and extended models of MVI s combined with other visible bands were analysed using bootstrapped simple and stepwise multiple linear regression functions. When MVI s were singly regressed with measured AGB , ratio‐based indices yielded the highest r 2 value of 0.64, followed by soil‐adjusted indices ( r 2 = 0.61), while atmospherically corrected MVI s showed the lowest r 2 of 0.58 ( p = 0.00). A significant improvement in herbaceous AGB estimation was obtained using a combination of MVI s and other visible bands. Soil‐adjusted MVI s showed the greatest increase (44–46%) in r 2 , while atmospherically corrected and ratio‐based MVI s poorly improved (&lt;5%). The findings demonstrate that combining MVI s with Landsat 8 optical bands, especially green band, provides the best models for estimating AGB in C. mopane savanna rangelands. These findings emphasize the importance of testing band‐ MVI combinations when developing models for estimating herbaceous AGB .

References

37