Auto-analysis of Trough System at 500 hPa Based on Gradient Algorithm

As the forecast products are increasing rapidly and the efficiency of manual process remains pretty low at present, the demand for intelligent assistant computer system of weather forecast is very urgent. It is the fundamental way to improve the automatic level of operational forecast in order to release the working pressure for detail forecast. The study on auto-analysis technology of weather systems is conducted as it is the key to build an intelligent forecast computer software with the thinking way of human kind. Based on synoptic notion and graphic features of trough system, under the guidance of computer graphics theory, trough system at 500 hPa are analyzed.The mathematic relations between gradient of geopotential height, its perpendicular direction and the nodes of trough system are proven by analysis of trough system conception, synoptic and graphics characteristics. The algorithm of trough system auto-analysis is developed under the strict constrain of atmospheric dynamic theory taking gradient of geopotential height as key element, and searching the relative minimal in its perpendicular direction within an optimal neighborhood which is determined by contrastive analysis. A set of software is developed using the geopotential height field of European Centre for Medium-Range Weather Forecasts at 500 hPa in format 4 of MICAPS which is the operational standard of China Meteorological Administration to meet the need of forecast centers at all levels of the country. The technical problems such as gradient calculating, direction zoning, neighborhood designating, primary nodes selecting, noise filtering, cluster analyzing are solved. Some fake troughs and over flow points can be identified with positioning according to the synoptic principals and basic experience, while the indices of density and location of primary nodes are applied to eliminate different noises, which concentrated in some certain areas. The multiple trough lines are merged by clustering and axial direction averaging. The auto-analysis technology and process of both westerly trough and traverse trough lines are developed and products are output in format 14 of MICAPS for forecasters. The algorithm and corresponding software system of auto-analysis of troughs is evaluated with parameters such as offset, length and general inclination angle. Recent experiments on cold wave samples in Ningxia show that the algorithm and the relevant software are stable, the effect in general is satisfying, and the method shows a better performance, especially in areas of middle and high latitudes. Through trail run, the efficiency of trough auto-analysis is improved further, and shortages of contour pursuing method are eliminated.