International Journal of Remote Sensing · 2017 · 76 citations · 85 references
Natural ResourcesForestryGeoinformaticsForest InventoryLandscape ModelingRemote SensingLand Cover ClassificationRemote Sensing Image AnalysisRemote Sensing IntegrationVegetation Cover AssessmentVegetation IndicesVegetation MonitoringCanopy Cover EstimationCanopy Cover MappingForest Canopy Cover EstimationGray-level Co-occurrence MatricesMultispectral SensingMultispectral Vegetation IndicesRemote Sensing AnalysisTree Cover Assessment
The availability of accurate and updated spatial information of tree cover in semi-arid and arid silvopastoral systems (SPSs) is crucial to understand their spatial patterns and trends. Although remote-sensing techniques have been proved useful in estimating tree canopy cover in general, more research is required to investigate the capabilities of new high spectral and spatial resolution satellites, such as Sentinel-2A, in predicting tree canopy cover in semi-arid environments. The objective of this study was to explore the capabilities of Sentinel-2A multispectral data, in combination with a stochastic modelling technique, for mapping montado tree canopy cover percentage (CCP) at pixel level. The stochastic gradient boosting algorithm was used to predict tree CCP using Sentinel-2A spectral data, vegetation indices, and textural information as predictor variables. The results of the study showed that the combination of multispectral bands with the selected vegetation indices and grey-level co-occurrence matrix (GLCM) texture features performs well, presenting a coefficient of determination (R2) of 82.8% and an error prediction of 8.68%. The analysis also showed that normalized difference vegetation index (NDVI) and Plant Senescence Reflectance Index (PSRI), as well as homogeneity texture feature, were the most important predictor variables to undertake the complex montado tree canopy cover estimation. In addition, this study demonstrated the usefulness of narrow spectral bands provided by the Sentinel-2A sensor for accurately estimating tree CCP (e.g. Red Edge1 – B5 – for PSRI and NIR2 – B8a – for NDVI computation). The modelling procedure used here emphasizes the effectiveness of stochastic models for predicting tree canopy cover from a complex semi-arid silvopastoral system by using Sentinel-2A multispectral data.
85
R: A Language and Environment for Statistical Computing
R Core Team · 2000 · 353K citations · Full text
Natural Sciences, Computer Science, Statistical Computing +10
Leo Breiman · Machine Learning · 2001 · 126.2K citations · Full text
Greedy function approximation: A gradient boosting machine.
Jerome H. Friedman · The Annals of Statistics · 2001 · 29K citations · Full text
High-Resolution Global Maps of 21st-Century Forest Cover Change
Matthew C. Hansen, Peter Potapov, Rebecca Moore et al. · Science · 2013 · 11.6K citations