Long-term cloud characterization at the AGORA ACTRIS-CCRES station using a novel classification algorithm

Abstract. The Western Mediterranean is a climatic hotspot with strong variability in cloud processes. However, Cloudnet sites there are scarce compared to northern Europe. This study presents for the first time a five-year cloud statistical analysis at the AGORA ACTRIS-CCRES station in Granada (Spain), using 94 GHz Doppler radar, microwave radiometer, and ceilometer data. Analyses focus on single-layer clouds and their interannual variability in macrophysical and microphysical properties. A new cluster-based algorithm (CBA) is introduced for cloud classification, reducing spurious correlations found in earlier methods. The CBA shows single-layer cloud minima in summer, with annual occurrences of 5.0 % for ice, 3.6 % for precipitating ice, 3.4 % for mixed-phase, 3.2 % for precipitating mixed-phase, and 1.4 % (1.2 %) for liquid (precipitating liquid) clouds. Liquid clouds are observed at 1–2 km, thin (∼200–300 m), with a droplet radius of 5 μm and liquid water paths of 12 g m⁻². Mixed-phase clouds occur at 5–6 km, nearly 1 km thicker, with larger droplets (10.8 μm) and ice water paths of 3.5 g m⁻². Ice clouds dominate at 7–8 km, the thickest type, with higher ice water paths (8.5 g m⁻²) but smaller particles (∼39 μm) than mixed-phase (∼45 μm). Across all phases, precipitating clouds have lower bases, greater thickness, and higher water content and particle sizes than non-precipitating clouds. These results provide benchmark data for satellite and model evaluation. The algorithm can be applied to other Cloudnet sites, supporting consistent European cloud statistics.