This dataset provides long-term daily aggregated eddy-covariance, meteorological, radiation, and soil observations collected at the Kayoro cropland site in northern Ghana between 2013 and 2025. The observations originate from six stations representing a gradient of land-use types, including near-natural savanna (Nazinga, Mole), cropland (Kayoro), rainfed rice field (Janga), and (semi-)degraded grassland (Sumbrungu, Gorigo). Turbulent fluxes were derived from high-frequency (20 Hz) measurements using standard EC processing procedures implemented in EddyPro (v7.0.9), including despiking, coordinate rotation, spectral corrections, and Webb–Pearman–Leuning (WPL) density corrections. Quality assessment follows established stationarity and turbulence criteria, and quality flags are provided for all fluxes. The dataset is aggregated to a uniform 30-minute temporal resolution, with no gap filling applied. All variables are provided in standardized units and naming conventions consistent with FLUXNET guidelines to facilitate comparability and reuse. The dataset supports studies of land–atmosphere interactions, surface–atmosphere coupling, and the impacts of land-use and climate variability in a data-scarce region of the global tropics.

The half-hourly datasets contain the full set of processed eddy-covariance fluxes, radiation components, meteorological variables, turbulence parameters, and soil measurements. The daily datasets contain aggregated values derived from the half-hourly observations, using appropriate aggregation methods (e.g. means for state variables and fluxes, sums for precipitation).