We improve coastal ocean carbonate system estimates by assimilating glider pH and alkalinity data into a regional biogeochemical model. Joint assimilation with physical observations successfully improves pH estimates while maintaining physical estimates. A hybrid approach combining dynamical models with statistical methods produces accurate pH estimates without requiring biogeochemical models, offering an alternative solution for ocean acidification monitoring.

A dataset of phytoplankton cell counts, pigments, particulate organic carbon and optical properties enables comparison of three methods to estimate phytoplankton carbon (C_phyto) in oligotrophic waters, where uncertainties in phytoplankton productivity are still large. Two methods based on chlorophyll concentration and particulate backscattering, are scalable to global scale while cell counts reduce bias from non-algal material. This comparison clarifies uncertainties in optical C_phyto estimates.

Understanding what shapes lake water quality is vital in a changing world. We studied dissolved organic matter, a key part of water quality in lakes and the carbon cycle, to analyse its environmental drivers and make predictions, by using machine learning. Tested in lakes in Ireland and Spain, it showed good predictions, even when relying only on climate-soil data, and Julian day. This helps explain how land and climate conditions influence freshwater resources. It can be reproduced worldwide.