Abstract:Early warning systems for food security require accurate and up-to-date information on the location of major crops in order to prevent hazards. A recent systematic analysis of existing cropland maps identified priority areas for cropland mapping and highlighted a major need for the Sahelian and Sudanian agrosystems. This paper proposes a knowledge-based approach to map cropland in the Sahelian and Sudanian agrosystems that benefits from the 100-m spatial resolution of the recent PROBA-V sensor. The methodology uses five temporal features characterizing crop development throughout the vegetative season to optimize cropland discrimination. A feature importance analysis validates the efficiency of using a diversity of temporal features. The fully-automated method offers the first cropland map at 100-m using the PROBA-V sensor with an overall accuracy of 84% and an F-score for the cropland class of 74%. The improvements observed compared to existing cropland products are related to the hectometric resolution, to the methodology and to the quality of the labeling layer from which reliable training samples were automatically extracted. Classification errors are mainly explained by data availability and landscape fragmentation. Further improvements are expected with the upcoming enhanced cloud screening of the PROBA-V sensor.

Abstract:With the ever-increasing number of satellites and the availability of data free of charge, the integration of multi-sensor images in coherent time series offers new opportunities for land cover and crop type classification. This article investigates the potential of structural biophysical variables as common parameters to consistently combine multi-sensor time series and to exploit them for land/crop cover classification. Artificial neural networks were trained based on a radiative transfer model in order to retrieve high resolution LAI, FAPAR and FCOVER from Landsat-8 and SPOT-4. The correlation coefficients between field measurements and the retrieved biophysical variables were 0.83, 0.85 and 0.79 for LAI, FAPAR and FCOVER, respectively. The retrieved biophysical variables' time series displayed consistent average temporal trajectories, even though the class variability and signal-to-noise ratio increased compared to NDVI. Six random forest classifiers were trained and applied along the season with different inputs: spectral bands, NDVI, as well as FAPAR, LAI and FCOVER, separately Remote Sens. 2015, 7 10401 and jointly. Classifications with structural biophysical variables reached end-of-season overall accuracies ranging from 73%-76% when used alone and 77% when used jointly. This corresponds to 90% and 95% of the accuracy level achieved with the spectral bands and NDVI. FCOVER appears to be the most promising biophysical variable for classification. When assuming that the cropland extent is known, crop type classification reaches 89% with spectral information, 87% with the NDVI and 81%-84% with biophysical variables.

No abstract