Abstract: Area observation is realized by the geostationary meteorological satellites of the United States and Europe. The timeliness of satellite observation is greatly improved, and an instance can be accomplished within several minutes. Geostationary meteorological satellites of FY-2 series have the capability of specific area observation, but due to the strip image navigation problem, the area observation of FY-2 satellites is not put into operational use. Therefore, the attitude determination technology of FY-2 satellites' area observation is studied and 3 solving methods are proposed, which are based on area images, attitude prediction, and the relation model between crude and precise attitude.The attitude determination method based on area images uses area images to calculate satellite attitude directly. This method is applicable to the condition that area images can be obtained continuously. When area images are accumulated for twenty-four hours, the method navigates the area image with the attitude gained by the method based on area images. The navigation deviation is in 2.5 infrared pixels. FY-2 satellite receives full disk images under normal observation mode, and receives high-frequency area images under emergency observation mode when special weather occurs. When using the attitude to navigate satellite image at the beginning of area observation, the image navigation accuracy maybe not ideal because the area images quantity is not enough. On this occasion, the attitude solving method based on attitude prediction can be used. With this method, the average deviation of image navigation can be reduced to 1 pixel. Based on the precise attitude which is gained through full disk images, this method predicts the future satellite attitude according to the variety rules of attitude. It is applicable to predict navigation parameter in future 24 hours with mathematics model while precise orbit attitude parameter is obtained. The attitude solving method based on relation model of crude-precise attitude uses the crude attitude which is calculated from telemetry data and the relation between the latest crude-precise attitude, and estimates precise attitude from the crude attitude which is calculated in real time. Using the attitude gained by this method to navigate the image, the max navigation deviation of area image is 4.9 infrared pixels, and the average deviation of former 24 hours is 3.6 infrared pixels. When precise attitude is not available, the attitude obtained by the method based on relation model of crude-precise attitude can be used to navigate area image during emergency situations.