Estimating above ground biomass as an indicator of carbon storage in vegetated wetlands of the grassland biome of South Africa

Naidoo, Laven van Deventer, Heidi Ramoelo, Abel Mathieu, Renaud Nondlazi, Basanda Gangat, Ridhwannah

Estimating above ground biomass as an indicator of carbon storage in vegetated wetlands of the grassland biome of South Africa

Naidoo, Laven 1van Deventer, Heidi 1Ramoelo, Abel 2Mathieu, Renaud 1Nondlazi, Basanda 1Gangat, Ridhwannah3
扫码查看

作者信息

  • 1. CSIR, Ecosyst Earth Observat, Pretoria, South Africa
  • 2. Univ Limpopo, Risk & Vulnerabil Assessment Ctr, Private Bag X1106, ZA-0727 Polokwane, South Africa
  • 3. Univ Witwatersrand, Sch Geog Archaeol & Environm Studies, Private Bag 3, ZA-2050 Johannesburg, South Africa
  • 折叠

Abstract

Wetlands store higher carbon content relative to other terrestrial ecosystems, despite the small extent they occupy. The increase in temperature and changes in rainfall pattern may negatively affect their extent and condition, and thus the process of carbon accumulation in wetlands. The introduction of the Sentinel series (S1 and S2) and WorldView space-borne sensors (WV3) have enabled monitoring of herbaceous above ground biomass (AGB) in small and narrow wetlands in semi-arid areas. The objective of this study was to assess (i) the capabilities of the high to moderate resolution sensors such as WV3, S1A and S2A in estimating herbaceous AGB of vegetated wetlands using SAR backscatter, optical reflectance bands, vegetation spectral indices (including Leaf Area Index or LAI measurements) and band ratio datasets and (ii) whether significant differences exists between the AGB ranges of wetland and surrounding dryland vegetation. A bootstrapped Random Forest modelling approach, with variable importance selection, was utilised which incorporated ground collected grass AGB for model calibration and validation. WorldView-3 (WV3) yielded the highest AGB prediction accuracies (R-2 = 0.63 and RMSE = 169.28 g/m(2)) regardless of the incorporation of bands only, indices only or the combination of bands and indices. In general, the optical sensors yielded higher modelling accuracies (improvement in R-2 of 0.04-0.07 and RMSE of 11.48-17.28 g/m(2)) than the single Synthetic Aperture Radar (SAR) sensor but this was marginal depending on the scenario. Incorporating Sentinel 1A (S1) dual polarisation channels and Sentinel 2A (S2) reflectance bands, in particular, yielded higher accuracies (improvement in R-2 of 0.03-0.04 and RMSE of 5.4-16.88 g/m(2)) than the use of individual sensors alone and was also equivalent to the performance of the high resolution WV3 sensor results. Wetlands had significantly higher AGB compared to the surrounding terrestrial grassland (with a mean of about 80 g/m(2) more). Monitoring herbaceous AGB at the scale of the wetland extent in semi-arid to arid grasslands enables improved understanding of their carbon sequestration potential, the contributions to global carbon accounting policies and also serving as a proxy for functional intactness.

Key words

Vegetation biomass/AGB/Wetland types/Carbon sequestration/Remote sensing/Earth observation/Sentinel sensors

引用本文复制引用

出版年

2019
International journal of applied earth observation and geoinformation

International journal of applied earth observation and geoinformation

SCI
ISSN:0303-2434
被引量18
参考文献量55
段落导航相关论文