Abstract: Traditional radiosonde balloon can float a long distance from its releasing place especially when reaching a high level above ground, while special radiosondes consisting of wind profiler and satellite remote sensing information are snapshots of atmospheric status and have no spatial drifts of particular location. The spatial derivatives (such as divergence) calculated through triangle method are closely related to the triangle's three-culminations position. The balloon floating and inhomogeneity introduced by the mixed use of data from traditional and special radiosondes can dramatically change the relative position of those culminations. The balloon floating feature and its impact on the divergence calculated through triangle method is a subject of potential application. Based on traditional radiosonde data of three stations in Eastern China, namely Nanjing, Anqing and Hangzhou with time coverage from 2006 to 2013, statistical features of balloon drift are investigated. And three experiments are designed to investigate how the balloon drifting impact the divergence computed through triangle method. The first experiment does not take the balloon drift into account, representing the traditional case which regards the radiosonde is right above the releasing point. The second experiment regards three balloons floating freely controlled by the atmospheric circumstance, which reflects the true physical processes of balloon motion. The third experiment is to simulate the inhomogeneity of traditional and special radiosondes: The balloon from Nanjing is assumed to have no horizontal motion as an analogue of special radiosonde and balloons from other two stations are freely floating as representatives of traditional radiosondes. Result shows that the balloon floats eastward all year round except in July and August. The float distance is larger in winter contrast with other seasons, with its maximum of about 120 km. In July and August, the balloon floats eastward within the low level and change its direction to westward at higher level. In meridional direction, the balloon floats in the manner of monsoon. In tropospheres, it floats northward in summer and southward in winter. Above 100 hPa in stratosphere it floats oppositely, southward in summer and northward in winter. The whole layer maximum mean drift distance is about 30 km in July and August, and the distance is larger than 100 km in winter above 100 hPa. Divergence differences between the second and third experiments to the first experiment are researched. The absolute difference increases with the height and reaches its maximum between 200-100 hPa in each month. The second experiment's relative differences are larger from June to September, with the extreme value about 7%. Relative differences of the third experiment are larger than those of the second experiment. In July and August, the relative difference of the whole layer is slightly less than 9%. In winter months, they are larger contrast to other months, when the relative difference of the third experiment is 25% at 200 hPa and greater than 50% at 50 hPa in January. It indicates that when the divergence of high level is calculated through triangle method, the balloon drift should be taken into account. When both traditional and special radiosonde data are used, for the inhomogeneity of measurement causes large differences, the balloon drift must be considered.