The Statistical Pattern Recognition of the Weather Conditions Based on the Gray-Scale of Image

In this paper, gray-scale of various types of cloud images collected over the Kunming Area were analyzed based on statistical theory and methods in order to achieve recognition of the pattern of the weather conditions. The results show that there are remarkable differences in normal distribution on the gray-scale histogram and the recurrence plot for different weather conditions. It is shown that the gray-scale method is simple, feasible, timely, reliable, and accurate. That would provide theoretical support and methods for meteorological and other related departments. DOI: 10.4018/jaec.2012070105 International Journal of Applied Evolutionary Computation, 3(3), 78-87, July-September 2012 79 Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. Overall, weather forecast have developed to a new era when a variety of the latest achievements of science and technology and methods are applied synthetically based on the platform of man-machine interactive processing system and the numerical weather prediction from the traditional semi-empirical theory of qualitative methods of the climate theory, mathematical statistics and forecaster’ experience; and the way of forecasting gradually shifts to the in-depth consideration of the inherent laws of atmospheric motion from the external phenomena of atmosphere (Qiu, 2000). More specifically speaking, the kinematic method and extrapolation method, the most popular weather forecast technology 50 years ago, had followed the style of the Bergen School. Since the operational model of numerical weather forecast appeared, the national meteorological centers gradually adopted the numerical power and probabilistic forecasts in order to guide the weather situation or element prediction after 6 ~ 12 hours (Zeng, 1979). Before 1966,people like Zeng, Chou, and Zhou Xiaoping have made many contributions to the development of the numerical techniques, initialization and mode (Chen & Xue, 2004; Zeng, 2002). Afterwards, scientists had a lot of research on the predictability of the various types of weather phenomena. Lorenz (1963) proposed that the atmosphere is a chaotic system and probability concepts should be adopted to forecast weather change. His “attractor” research has been widely used in the field of fluid mechanics. An important theoretical foundation was provided by Lorenz for numerical statistics (MOS), probability statistics forecast and numerical ensemble prediction. However, it was found that a tiny error in initial conditions would enlarge with the proliferation of larger scales in many trials of the modes. Besides, with the wide application of computer, synoptic meteorology had changed gradually from description by artistic experience of weather phenomena and process to analysis of dynamic and thermodynamic diagnostic in order to incorporate synoptic meteorology and atmospheric dynamics. By the 20th century, the grid points of the mode in two-dimensional (z, y) has increased from 102×102 to 103×103 with the development of large-capacity and highspeed computers. Undoubtedly, development of these technologies is great power for data collection, exchange, processing of numerical weather prediction, weather, climate, ocean observing and other areas atmospheric science. In addition, the World Meteorological Organization (WMO) is embarked on a Global Weather Observing System Research and Predictability experiment (THORPEX) for 10 years (Shapiro & Thorpe, 2004), further enrich results achieved from the ensemble forecasting and adaptive observation to improve dynamical prediction and probabilistic forecasts of disastrous weather and rainfall. A short memory and long-mutation features exists in weather changes (Ding, 1991). It is concerned that the following practical problems: whether the weather in cloudy changes would be turned fine, rainy, or continue to remain cloudy especially turned rainy and fine is what people care about. The results of the kinematics and other research have become very inaccurate because of many unusual climates, even the exact opposite in recent years. The images method combined with the statistical analysis was used in this paper (Fu, Dai, & Ye, 2009; Hu, Hong, Ru, Wang, & Chen, 2010; Huang, Zuo, Zhang, & Liu, 2010), and that is a new perspective to the gray-scale histogram and recurrence plot analysis method for real-time forecasting of the weather conditions. The results show that the method is simple, reliable, accurate and real-time.