Search

Earth Observation has become a progressively important source of information for land use and land cover services over the past decades. At the same time, an increasing number of reconnaissance satellites have been set in orbit with ever increasing spatial, temporal, spectral, and radiometric resolutions. The available bulk of data, fostered by open access policies adopted by several agencies, is setting a new landscape in remote sensing in which timeliness and efficiency are important aspects of data processing. This study presents a fully automated workflow able to process a large collection of very high spatial resolution satellite images to produce actionable information in the application framework of smallholder farming. The workflow applies sequential image processing, extracts meaningful statistical information from agricultural parcels, and stores them in a crop spectrotemporal signature library. An important objective is to follow crop development through the season by analyzing multi-temporal and multi-sensor images. The workflow is based on free and open-source software, namely R, Python, Linux shell scripts, the Geospatial Data Abstraction Library, custom FORTRAN, C++, and the GNU Make utilities. We tested and applied this workflow on a multi-sensor image archive of over 270 VHSR WorldView-2, -3, QuickBird, GeoEye, and RapidEye images acquired over five different study areas where smallholder agriculture prevails. [ABSTRACT FROM AUTHOR]

The use of satellite image time series analysis and machine learning methods brings new opportunities and challenges for land use and cover changes (LUCC) mapping over large areas. One of these challenges is the need for samples that properly represent the high variability of land used and cover classes over large areas to train supervised machine learning methods and to produce accurate LUCC maps. This paper addresses this challenge and presents a method to identify spatiotemporal patterns in land use and cover samples to infer subclasses through the phenological and spectral information provided by satellite image time series. The proposed method uses self-organizing maps (SOMs) to reduce the data dimensionality creating primary clusters. From these primary clusters, it uses hierarchical clustering to create subclusters that recognize intra-class variability intrinsic to different regions and periods, mainly in large areas and multiple years. To show how the method works, we use MODIS image time series associated to samples of cropland and pasture classes over the Cerrado biome in Brazil. The results prove that the proposed method is suitable for identifying spatiotemporal patterns in land use and cover samples that can be used to infer subclasses, mainly for crop-types. [ABSTRACT FROM AUTHOR]

The production of land cover maps through satellite image classification is a frequent task in remote sensing. Random Forest (RF) and Support Vector Machine (SVM) are the two most well-known and recurrently used methods for this task. In this paper, we evaluate the pros and cons of using an RF-based kernel (RFK) in an SVM compared to using the conventional Radial Basis Function (RBF) kernel and standard RF classifier. A time series of seven multispectral WorldView-2 images acquired over Sukumba (Mali) and a single hyperspectral AVIRIS image acquired over Salinas Valley (CA, USA) are used to illustrate the analyses. For each study area, SVM-RFK, RF, and SVM-RBF were trained and tested under different conditions over ten subsets. The spectral features for Sukumba were extended by obtaining vegetation indices (VIs) and grey-level co-occurrence matrices (GLCMs), the Salinas dataset is used as benchmarking with its original number of features. In Sukumba, the overall accuracies (OAs) based on the spectral features only are of 81.34 % , 81.08 % and 82.08 % for SVM-RFK, RF, and SVM-RBF. Adding VI and GLCM features results in OAs of 82 % , 80.82 % and 77.96 % . In Salinas, OAs are of 94.42 % , 95.83 % and 94.16 % . These results show that SVM-RFK yields slightly higher OAs than RF in high dimensional and noisy experiments, and it provides competitive results in the rest of the experiments. They also show that SVM-RFK generates highly competitive results when compared to SVM-RBF while substantially reducing the time and computational cost associated with parametrizing the kernel. Moreover, SVM-RFK outperforms SVM-RBF in high dimensional and noisy problems. RF was also used to select the most important features for the extended dataset of Sukumba; the SVM-RFK derived from these features improved the OA of the previous SVM-RFK by 2%. Thus, the proposed SVM-RFK classifier is as at least as good as RF and SVM-RBF and can achieve considerable improvements when applied to high dimensional data and when combined with RF-based feature selection methods. [ABSTRACT FROM AUTHOR]

Remote sensing studies of vegetation phenology increasingly benefit from freely available satellite imagery acquired with high temporal frequency at fine spatial resolution. Particularly for heterogeneous landscapes this is good news, given the drawback of medium-resolution sensors commonly used for phenology retrieval (e.g., MODIS) to properly represent the fine-scale spatial variability of vegetation types. The Sentinel-2 mission acquires spectral data globally at 10 to 60 m resolution every five days. To illustrate the mission's potential for studying vegetation phenology, we retrieved phenological parameters for the Dutch barrier island Schiermonnikoog for a full season of Sentinel-2A observations in 2016. Overlapping orbits resulted in two acquisitions per 10 days, similar to what is achieved globally since the launch of Sentinel-2B. For eight locations on the island's salt marsh we compared greenness chromatic coordinate (GCC) series derived from digital repeat RGB-cameras with vegetation index series derived from Sentinel-2 (NDVI and GCC). For each series, a double hyperbolic tangent model was fitted and thresholds were applied to the modelled data to estimate start-, peak-, and end-of-season (SOS/PS/EOS). Variability in Sentinel-2 derived SOS, when taken as the midpoint between minimum and peak NDVI, was well-explained by camera GCC-based SOS (R 2  = 0.74, MSD = 8.0 days, RMSD = 13.0 days). However, EOS estimates from camera GCC series were on average almost two months before NDVI-based estimates. This could partially be explained by the observed exponential relationship between GCC and NDVI, as well as by the combined effect of viewing angle differences and the presence of non-photosynthetic elements in the vegetation canopy. A two-layer canopy radiative transfer model incorporating reduced chlorophyll levels in the upper layer provided a physically-based explanation of the viewing angle effect. Finally, we applied the phenology retrieval approach to NDVI series for all pixels of the island in order to map spatial patterns of phenology at fine resolution. Our results demonstrate the potential of the Sentinel-2 mission for providing spatially-detailed retrievals of phenology. [ABSTRACT FROM AUTHOR]