摘要：The increasing demand for palm oil and bioenergy has promoted the expansion of tropical farmland covered with oil palms (Elaeis guineensis), resulting in increased competition with food production as well as environmental degradation. Moreover, oil palm cultivation may have increased greenhouse gas (GHG) emissions through deforestation. The overall impact estimation of oil palm related land-use change requires spatiotemporal land-use maps. So far, the Roundtable on Sustainable Palm Oil (RSPO) has not established guidelines on how to measure and evaluate oil palm related land-cover change. While remote sensing methods are suitable in general, the use of Landsat images in the tropics for the monitoring and modeling of land-cover changes has been restricted due to the influence of cloud cover. This study presents a novel approach for mapping tropical land-cover change using the Google Earth Engine (GEE) cloud-based platform and the System for Automated Geoscientific Analysis (SAGA) GIS. Spatiotemporal land-use and land-cover changes in relation to oil palm cultivation are assessed using a median pixel composite mosaic of Landsat 5, 7 and 8 image scenes for the time periods 1999-2005 and 2009-2015. The proposed approach yields an overall accuracy and kappa coefficient of 70.33 % and 0.62 for the first image composite period, and 84.5 % and 0.80 for the second image composite period respectively.
其他摘要：The increasing demand for palm oil and bioenergy has promoted the expansion of tropical farmland covered with oil palms (Elaeis guineensis), resulting in increased competition with food production as well as environmental degradation. Moreover, oil palm cultivation may have increased greenhouse gas (GHG) emissions through deforestation. The overall impact estimation of oil palm related land-use change requires spatiotemporal land-use maps. So far, the Roundtable on Sustainable Palm Oil (RSPO) has not established guidelines on how to measure and evaluate oil palm related land-cover change. While remote sensing methods are suitable in general, the use of Landsat images in the tropics for the monitoring and modeling of land-cover changes has been restricted due to the influence of cloud cover. This study presents a novel approach for mapping tropical land-cover change using the Google Earth Engine (GEE) cloud-based platform and the System for Automated Geoscientific Analysis (SAGA) GIS. Spatiotemporal land-use and land-cover changes in relation to oil palm cultivation are assessed using a median pixel composite mosaic of Landsat 5, 7 and 8 image scenes for the time periods 1999-2005 and 2009-2015. The proposed approach yields an overall accuracy and kappa coefficient of 70.33 % and 0.62 for the first image composite period, and 84.5 % and 0.80 for the second image composite period respectively
