Improvement of Automatic Present Weather Observation with In Situ Visibility and Humidity Data

The present weather is an indication of current weather condition that directly affects our daily activities such as public welfare and road safety. Sources and dynamics of this phenomena have long been a great interest for meteorologists since they are results from atmospheric turbidity and types of precipitation. Long history of human eyed observation is still valuable, yet it is subjective and relatively inconsistent. Over a decade, automatic present weather sensors have been in daily operation by Korea Meteorological Administration (KMA), along with human-eye observation. While manual observations are available only 10% of automatic present weather sensors as of 2019, initial study shows it is far from perfect match between them otherwise higher spatial resolution of present weather service is available. 8 years of human-eyed present weather records in 19 sites over Korea are compared with visibility sensors and auxiliary measurements, such as AWS (Automatic Weather Station). As clear condition agrees with high probability, next best categories follow fog, rain, snow, mist, haze and drizzle in comparison with human-eyed observation. Fog, mist and haze are often confused due to nature of machine sensing visibility. Differences between instrument manufacturers are also found indicating nonstandard present weather decision. Such ambiguous weather conditions are improved with empirically induced criteria in combination with visibility and humidity. Accuracies of present weather for haze, mist, and fog are all improved by 61.5%, 44.9%, and 26.9% respectively. The results shows that automatic present weather sensing is feasible for operational purpose with minimal human interactions. Causes of differences are further investigated by different sensing types between manufacturers and by geographical dependencies due to observation environment, such as mountain or flat area, and distance from the sea. The same correction method are applied over 291 sites of automatic present weather sensors. Automatic present weather with the correction method has accepted an initial review of replacement of human-eye based observation in KMA. The result is beneficial for present weather measurement and enhancing its application to improve traffic warning system from roadside automatic present weather sensors.