Current issue: 58(4)
This study proposes an original method for tree species classification by satellite remote sensing. The method uses multitemporal multispectral (Landsat OLI) and hyperspectral (Resurs-P) data acquired from determined vegetation periods. The method is based on an original database of spectral features taking into account seasonal variations of tree species spectra. Changes in the spectral signatures of forest classes are analyzed and new spectral–temporal features are created for the classification. Study sites are located in the Czech Republic and northwest (NW) Russia. The differences in spectral reflectance between tree species are shown as statistically significant in the sub-seasons of spring, first half of summer, and main autumn for both study sites. Most of the errors are related to the classification of deciduous species and misclassification of birch as pine (NW Russia site), pine as mixture of pine and spruce, and pine as mixture of spruce and beech (Czech site). Forest species are mapped with accuracy as high as 80% (NW Russia site) and 81% (Czech site). The classification using multitemporal multispectral data has a kappa coefficient 1.7 times higher than does that of classification using a single multispectral image and 1.3 times greater than that of the classification using single hyperspectral images. Potentially, classification accuracy can be improved by the method when applying multitemporal satellite hyperspectral data, such as in using new, near-future products EnMap and/or HyspIRI with high revisit time.
A large quantity of non-viable (empty, insect-attacked and shriveled) seeds of Juniperus polycarpos (K. Koch) is often encountered during seed collection, which should be removed from the seed lots to ensure precision sowing in the nursery or out in the field. The aims of this study were to evaluate different modelling approaches and to examine the sensitivity of the change in detection system (Silicon-detector in the shorter vis-a-vis InGsAs-detector in the longer NIR regions) for discriminating non-viable seeds from viable seeds by Near Infrared (NIR) spectroscopy. NIR reflectance spectra were collected from single seeds, and discriminant models were developed by Partial Least Squares – Discriminant Analysis (PLS-DA) and Orthogonal Projection to Latent Structures – Discriminant Analysis (OPLS-DA) using the entire or selected NIR regions. Both modelling approaches resulted in 98% and 100% classification accuracy for viable and non-viable seeds in the test set, respectively. However, OPLS-DA models were superb in terms of model parsimony and information quality. Modelling in the shorter and longer wavelength region also resulted in similar classification accuracy, suggesting that prediction of class membership is insensitive to change in the detection system. The origins of spectral differences between non-viable and viable seeds were attributed to differences in seed coat chemical composition, mainly terpenoids that were dominant in non-viable seeds and storage reserves in viable seeds. In conclusion, the results demonstrate that NIR spectroscopy has great potential as seed sorting technology to upgrade seed lot quality that ensures precision sowing.