A method for computing the trajectory is presented, and then a case study is made with the method by using a ten-level primitive equation model. Comparing the results obtained from the data with different time intervals ranging from 30 minutes to 12 hours, including the conventional observations only twice a day shows that using those data with the intervals equal to or less than 3 hours can lead to accurate trajectory computations while significant discrepancies are found with 12 hour interval.