Current Knowledge and Future Challenge for Visibility Forecasting by Computational Intelligence

Visibility forecasting, a challenging task of big data, is a complicated work in environmental simulation. There are many factors affecting the visibility change, such as humidity, cloud, and particulate concentration. The traditional statistical analysis cannot predict the visibility change very well, so the researchers try to apply various new methods. In this study, a review of the results of current knowledge, as well as future perspective about the forecasting of visibility, is presented. Because the environmental visibility is a typical type of big data, summarizing the meaningful information is benefit to the work of forecasting. Development of information granularity and computational intelligence help us to solve these problems. Various algorithms, such as grey theory, fuzzy theory, and neural network etc. provide superior capability in forecasting. However, how to effectively use the observed data is another great challenge. Finding scalable computational intelligence algorithms for visibility forecasting is essential. The knowledge from atmospheric dynamics and aerosol science helps us to explain the information derived from the big data of visibility. To spend more efforts in the interpretation of their relationship and develop more advanced mining methodologies helps us to construct the future model of visibility forecasting.