Hyper-temporal remote sensing for scaling between spectral indices and flux tower measurements

Abstract

Quantifying spatial and temporal variations of the environmental variables is a significant challenge in the Earth’s System studies. This can be addressed by tracking the exchange rate of carbon between land surface and the atmosphere through an effective integration of the tower-based flux measurements and remote sensing observations. However, confidence of these data with regard to extending into ecosystem or regional scales remains uncertain, particularly in heterogeneous landscapes. This paper studies how data with the necessary spatial and temporal sampling may be obtained for a typical temperate forest in southern England. A time series of the inter-calibrated Normalized Difference Vegetation Index (NDVI) from the Disaster Monitoring Constellation (DMC) over the flux tower extent suggested that the spatial variability of the canopy changed significantly during transitions from leaf-off to leaf-on conditions because of canopy patchiness at high spatial resolution, scales finer than a coarse satellite-sensor pixel. The availability of three acquisitions within five days during early March provided the opportunity to look at the sensitivity of the NDVI to extraneous factors, which showed the uncertainty in this spectral index to vary with the flux footprint coverage, peaking at around 5% early in the season. This is comparable to the error that could be introduced by uncertainty in atmospheric correction specifically the parameterisation of aerosols. Thus, in the case of Wytham Woods, it may be more important to improve the representation of aerosols in the atmospheric model used than replace the generalised flux tower extent with a realistic flux footprint.