Abstract
Tree species distribution is an important issue for sustainable forest resource management. However, the accuracy of tree species discrimination using remote-sensing data needs to be improved to support operational forestry-monitoring tasks. This study aimed to classify tree species in the Liangshui Nature Reserve of Heilongjiang Province, China using spectral and structural remote sensing information in an auto-mated Random Forest modelling approach. This study evaluates and compares the performance of two machine learning classifiers, random forests (RF), support vector machine (SVM) to classify the Chinese high-resolution remote sensing satellite GF-1 images. Texture factor was extracted from GF-1 image with grey-level co-occurrence matrix method. Normalized Difference Vegetation Index (NDVI), Ratio Vegetation Index (RVI), Enhanced Vegetation Index (EVI), Difference Vegetation Index (DVI) were calculated and coupled into the model. The result show that the Random Forest model yielded the highest classification accuracy and prediction success for the tree species with an overall classification accuracy of 81.07% and Kappa coefficient value of 0.77. The proposed random forests method was able to achieve highly satisfactory tree species discrimination results. And aerial LiDAR data should be further explored in future research activities.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have