Detection of seasonal changes of rivers in tropical forests from JERS-1 SAR images of Amazon, New Guinea, and Congo Basin

We have recently developed a new, fully automated method for detecting and visualizing narrow rivers in Amazon forests from 3-look JERS-1 SAR images using strong isolated scatterers lined up intermittently along rivers. The resulting approximate range of waterways correspond astonishingly well with waterways visible in the cloudless areas of the near IR images of JERS-1 VNIR data observed one week later. We applied this method to more than 250 images of multi-temporal 4 look JERS-1 SAR scenes of tropical forest areas, such as the Amazon from May, 1996 to February, 1997, New Guinea from November, 1995 to March 1996, and Congo Basin of February and November, 1996. The resulting change in brightness (or length) of rivers corresponds quite well with the change in the average monthly rainfall data of the nearby areas. We therefore conclude that the brightness of the waterways in the JERS-1 SAR images are qualitative indicators of water flow in these rivers.