Preliminary results on forest monitoring in Sumatra using ALOS-PALSAR

Shaun Quegan(1), Francesca Ticconi(1) and Yumiko Uryu(2)

(1) CTCD - University of Sheffield, Hicks Building, Hounsfield Road, Sheffield S3 7RH, United Kingdom
(2) WWF Indonesia, Kawasan Mega Kuningan, Jakarta 12950, Indonesia


Deforestation and changes in land use play a significant role in global climate change because carbon dioxide is released into the atmosphere when forests or forest products are burned or destroyed. In Indonesia the deforestation rate was very high over the last 23 years. More than 3 million hectares of forest have been lost due to fire and conversion of forest into land for farming and grazing. This study investigates the potential of the ALOS-PALSAR L-band radar sensor to monitor land cover changes and forest recovery. The Riau region in Sumatra has been chosen for study because it has excellent supporting ground data and forest maps based on field inventory available through the Indonesian section of WWF. The whole region covers about 11 million hectares, of which most is being exploited for agriculture or has been clear-cut. The remaining forest cover primarily consists of lowland dry forest, rubber plantations and swamp forest. Especially in the south of the region there are also many areas of forest re-growth. We are currently analysing images acquired in FBD (HH and HV) mode over the period July - August 2007 that cover part of the Riau region area. These are supplemented by images from November 2007 to cover a gap in the region for which no data were acquired in the period June – September 2007. At present we are still engaged in assessing the basic radiometric and geometric properties of the images, to ensure that we will encounter no problems when combining the 27 FBD images needed to cover the whole Riau region. The main purpose of the subsequent analysis will be to investigate the ability of PALSAR to distinguish forest from non-forest, to detect forest structure (primary, regrowing and plantation forest, tree density, etc.) and to distinguish variations in forest biomass to monitor forest re-growth. We will also look at more general land cover classes. In addition, we are investigating 8 ScanSAR images acquired in 2007 with a repetition time of 46 days over the region, in order to compare the information content of multi-temporal HH backscatter with that of dual-polarisation data.


Symposium presentation


  Higher level                 Last modified: 07.05.06