Intra-pixel scaling bias in satellite data for coastal monitoring: Evidence from photosynthetic biofilms

Abstract

Microphytobenthos (MPB) refers to photosynthetic, unicellular organisms that live in sediment, particularly in the intertidal zone of estuarine mudflats, where they can form biofilms. MPB provides a range of ecosystem services, including its role in carbon fluxes, sediment stabilisation, nutrient cycling, and the trophic network. MPB biomass is usually estimated by measuring the concentration of chlorophyll a at the sediment surface during low tide, which can be estimated locally and then extrapolated spatially using remote sensing methods, to produce biomass maps. Satellites are of particular interest since they provide long time series at high frequencies for studying MPB phenology. However, the MPB exhibits high spatial variability at the sub-satellite pixel scale, which is known to induce a scaling bias that results in the biomass being underestimated. This study used an unmanned aerial vehicle (UAV) to investigate the fine-scale spatial patterns of MPB, quantified the scaling bias and modelled it using generalized additive models (GAMs). Twenty-seven flights were performed over four different sites in different seasons in 2024 and 2025. The spatial resolutions of four satellites were compared: Pléiades Neo (1.2 m), Pléiades 1 A and 1B (2 m), SPOT 6 and 7 (6 m), and Sentinel-2 A, 2B and 2C (10 m). The bias values ranged between −14.2% (Pléiades Neo) and − 68.5% (Sentinel-2 A, 2B and 2C) when all the results were combined. This study emphasised the importance of spatial structure in quantitative remote sensing and proposed a practical framework for modelling the bias.