Mapping rubber stand ages in Luangnamtha district (Northern Laos) using NDVI and LSWI from Landsat images

摘要

Over the recent past, rubber tree plantations have rapidly expanded across vast tracts of Northern Laos. Remotely assessing rubber stand age is invaluable to monitor and predict the impacts of such an extensive and rapid land use change. This study presents the classification of rubber plantation areas versus non-rubber plantation areas, and the potential of two vegetation indices, namely NDVI and LSWI, for mapping rubber stand ages. Spectral signatures were derived from Landsat OLI images of Luangnamtha district acquired on 5th March and 14th March 2017, including Band 4 (RED), Band 5 (NIR), and Band 7 (SWIR), captured at the time of the year corresponding to the beginning of the foliation of rubber tree, with a spatial resolution of 30m. The work shows that the classification has a high accuracy of 92% and 0.89 for the overall accuracy and the Kappa coefficient, respectively. The producer’s and user’s accuracies of rubber stand ages were 75-92%, and 75-91% for the three age groups ( 12 years old). Overall accuracy and Kappa coefficients were 87 and 80%, and 0.79 and 0.67 for NDVI and LSWI indices, respectively. We also found that LSWI values were larger than NDVI when rubber stand age is greater than 7 years old, but NDVI showed better overall accuracy than LSWI at the beginning of the rubber tree foliation.