Global assessments quantify the net exchange of greenhouse gases (GHGs) between land and the atmosphere but lack a high-resolution, observation-based representation of underlying gross emissions and removals (1–3). Here, we integrate satellite-based data with GHG inventory methods to generate annual estimates of gross GHG emissions, gross carbon removals and net GHG fluxes from vegetation and soil across global land at 30-m resolution from 2016 to 2024. The monitored land processes, which exclude agricultural methane and nitrous oxide emissions, were a net sink of -3.6±4.1 Gt CO2e yr-1, comprised of 20.2±2.6 Gt CO2e yr-1 gross emissions and -23.8±3.1 Gt CO2 yr-1 gross removals; the relative contributions of vegetation and soil varied across landscapes. Gross emissions were spatially concentrated, with half of global emissions occurring in 4.5% of emitting land, while half of global removals occurred across 19% of land with removals. Our approach complements global carbon budgets (1,4) and inventory-based reporting systems (5–7) by resolving the geography of emissions and removals within a consistent geospatial framework, enabling more targeted emissions reductions, sustained carbon removals, and transparent tracking of progress across scales.