Accuracy of cirrus detection by surface-based human observers

Abstract The longest cirrus time series are ground-based, visual observations captured by human observers (SYNOP). However, their reliability is impacted an unfavourable viewing geometry (cloud overlap), and misclassification due to low cloud optical thickness, especially at night. For the very first time, this study assigns a quantitative value uncertainty. We validate 15 years of SYNOP (2006–2020) against data from lidar flown onboard Cloud-Aerosol Lidar Infrared Pathfinder (CALIPSO) spacecraft. develop dedicated method match reports (with hemispherical field view) with samples (along-track profiles). Our evaluation eye’s sensitivity revealed that it moderate, best. In perfect conditions (daytime no mid/low-level clouds) probability correct detection was 44–83% (Cohen’s kappa coefficient <0.6), fell 24–42% (kappa <0.3) Lunar illumination improved detection, but only when Moon’s phase exceeded 50%. Cirrus depth had clear impact on detection. When clouds all levels were considered (i.e. real-life conditions), moderate poor: detected 47–71% <0.45) during day, 28–43% <0.2) night, decreased increasing low/mid-level fraction. These coefficients suggest agreement CALIPSO close random. findings can be directly applied estimations frequency/trends. reported probabilities serve as benchmark for other ground-based methods.