Vegetation Spectral Reflectance Along a North-South Vegetation Gradient in Northern Australia

Abstract

This research demonstrates that frequently used vegetation indices should not be applied comprehensively across subtropical, semi-arid and arid zones. Zonal, species specific band combinations are considered more sensitive to changing conditions. In three areas within a subtropical to arid climatic gradient, ground radiometer readings were undertaken to establish basic spectral data for the quantification of green vegetation cover in Northern Territory, Australia. The density of woody green vegetation cover was 32% in the subtropical zone, 28% in the semi-arid zone and 9% in the arid zone. Dry season reflectance data on different woody species, dead and alive herbaceous cover, plant litter and soil (cover types) were obtained on the equivalent of Multispectral Scanner (MSS) wavebands. Relationships between ratio and linear band combinations (including vegetation indices) and percentage cover types were examined by regression analysis to determine what the band combinations were in fact predicting along the gradient. Results showed that completely different band combinations were necessary to predict green vegetation cover in subtropical as opposed to semi-arid and arid areas. Commonly applied indices like the NDVI (Normalized Difference Vegetation Index) were negatively associated with soil variables and not related to green vegetation cover. Further tests where projected foliage cover was artificially increased from 20-100% resulted in a suite of different spectral response curves for each climatic zone. These were categorized as: (a) plants which totally darken or (b) partially darken the soil background in all wavebands and (c) plants which are more highly reflective in the near-infrared than the background soil. The subtropical species divided evenly among the three response curve types but the semi-arid and arid species mostly darkened or partially darkened the local soil response. The need for different band combinations to predict green vegetation cover through the climatic zones results from the unique spectral response patterns of climatically variable plant species in relation to their soil background. This work has value in environmental and range monitoring in Northern Territory and other subcontinents spanning similar climatic belts.