Image interpolation smooths satellite cloud-data movement

Image interpolation smooths satellite cloud-data movement

Anyone who has watched weather forecasting on television in the last few years will immediately realize that when satellite data are superimposed on a map of the earth, the movement of cloud data is not smooth. Instead, clouds move in a jerky manner across the area being described by the forecaster. The reason is that cloud data are often generated from a global temperature (gray-level) threshold of satellite images.

When such data are combined with frequently occurring missing image data and a temporal sampling of 30 minutes per image, discontinuous sequences result. Now, thanks to assistant professor Rasmus Larsen at the Technical University of Denmark (Lyngby, Denmark), this is about to change.

Using grey-level thresholds based on ground temperatures in satellite IR images and satellite image temperature data, Larsen generates cloud maps that are more accurate than those produced with global thresholding techniques. Furthermore, to produce continuous images, cloud motion is estimated and interpolated on a per-pixel basis, allowing smooth transitions to occur.

Once this is accomplished, a series of interpolated images corresponding to three-minute time intervals can be inserted in the image sequence, resulting in a continuous motion sequence. Rasmus Larsen can be contacted at tel: +45-4588-1433; Fax: +45-4588-1397; or e-mail: [email protected].