Usability of Sentinel-1 C-band VV and VH SAR data for the detection of flooded oil palm

Abstract: Flooding in oil palm plantations in Southeast Asia is a common problem. The oil palm’s habitable range is restricted to lowland equatorial areas, and the high rainfall and degraded landscapes associated with oil palm cultivation leaves oil palm prone to flooding. Flooding diminishes yields in mature palms, increases mortality in young palms, and disrupts operational processes. This reduces the economic viability of plantations, which may lead to their abandonment. Oil palm expansion is associated with environmental degradation and high carbon emissions; when expansion occurs in flood-prone areas a high environmental cost is paid for little or no economic gain. Climate change is expected to bring about more variable and extreme rainfall events in Southeast Asia, therefore the problem of inundation in oil palm plantations is set to worsen. Improving our knowledge of flood events in plantations will inform decision-makers in the oil palm industry and policy-makers in government of the sensitivity of certain landscapes to oil palm cultivation so that the expansion of oil palm in flood-prone areas is avoided. Existing under-productive stands may be taken out of production and rehabilitated to restore the ecosystem, social, and economic functions of the peatland and forest landscapes that they replaced. Few studies have applied remote sensing technology to the problem of inundation in oil palm plantations. To the best of the author’s knowledge, this is the first study to report the backscatter characteristics of flooded oil palm. This study tested the ability of Sentinel-1 C-band VV and VH data to detect the presence/absence of flooding in oil palm stands of all growth stages in a study area in Jambi Province, Indonesia. Smallholdings were the predominant production system although industrial holdings were also present. Classes were defined to represent the different growth stages of oil palm in flooded and non-flooded conditions, and the backscatter characteristics and separability of the classes were determined. C-band successfully detected the presence of flooding in very young oil palm, but not in older oil palm. C-band’s short wavelength and limited canopy penetration ability meant the signal saturated early and did not generate the double bounce effect that is characteristic of flooded forests. New SAR data (L-band and P-band) coming online in 2024 provides exciting opportunities for future research as the longer wavelengths are expected to achieve superior canopy penetration than C-band and, in the case of P-band, will permit the detection of flooding in all growth stages of oil palm.