The multi-scale storm cell identification and tracking algorithm uses a blend of a sequential morphological technique and a K-means clustering technique to identify and track storm areas. The algorithm may identify and track a variety of data fields, such as radar reflectivity, vertically integrated liquid or satellite cloud tops. The automated tracking of storm cells is a very challenging problem, as storms morph, merge, split, intensify, and weaken in time; hence, one cannot simply match a current storm image with one from 5 or 15 minutes previous without accounting for these changes. Our algorithm uses storm intensity, overlap with previous images, and the distance of storm centroids between images to associate storm cells between images and estimate their motion using a constant-acceleration Kalman filter. The presentation explores the accuracy of the storm tracking on a variety of multi-sensor data fields.