Accuracy and Spatial Pattern Assessment of Forest Cover Change Datasets in Central Kalimantan

Abstract

The accurate information of forest cover change is important to measure the amount of carbon release and sink. The newly-available remote sensing based products and method such as Daichi Forest/Non-Forest (FNF), Global Forest Change (GFC) datasets and Semi-automatic Claslite systems offers the benefit to derive these information in a quick and simple manner. We measured the accuracy by constructing area-proportion error matrix from 388 random sample points and assessed the consistency analysis by looking at the spatial pattern of deforestation and regrowth from built-up area, roads, and rivers from 2010 – 2015 in Katingan district, Central Kalimantan. Accuracy assessment showed that those 3 datasets indicate low to medium accuracy level in which the highest accuracy was achieved by Claslite who produced 71 % ± 5 % of overall accuracy. The consistency analysis provides a similar spatial pattern of deforestation and regrowth measured from the road, river, and built-up area though their distance sensitivity are different one to another.

Keywords

• Area-proportion matrix
• Deforestation
• Gain
• Remote Sensing