In the open ocean, the observation and quantification of biodiversity patterns is challenging. Marine ecosystems are indeed largely composed by microbial planktonic communities whose niches are affected by highly dynamical physico-chemical conditions, and whose observation requires advanced methods for morphological and molecular classification. Optical remote sensing offers an appealing complement to these in-situ techniques. Global-scale coverage at high spatiotemporal resolution is however achieved at the cost of restrained information on the local assemblage. Here, we use a coupled physical and ecological model ocean simulation to explore one possible metrics for comparing measures performed on such different scales. We show that a large part of the local diversity of the virtual plankton ecosystem – corresponding to what accessible by genomic methods – can be inferred from crude, but spatially extended, information – as conveyed by remote sensing. Shannon diversity of the local community is indeed highly correlated to a ‘seascape’ index, which quantifies the surrounding spatial heterogeneity of the most abundant functional group. The error implied in drastically reducing the resolution of the plankton community is shown to be smaller in frontal regions as well as in regions of intermediate turbulent energy. On the spatial scale of hundreds of kms, patterns of virtual plankton diversity are thus largely sustained by mixing communities that occupy adjacent niches. We provide a proof of principle that in the open ocean information on spatial variability of communities can compensate for limited local knowledge, suggesting the possibility of integrating in-situ and satellite observations to monitor biodiversity distribution at the global scale.