Evaluation of SWIR Crop Residue Bands for the Landsat Next Mission

Drought impact on cropland use monitored with AVIRIS imagery in Central Valley, California

During 2012-2016 California experienced the longest and most severe drought in the last centuries. This water scarcity led to an increase in non-cultivated croplands during this period. The objective of this study was to quantify agricultural trends in the Central Valley (California) at peak growth from 2013 to 2016 during the drought and in 2017-2018 post-drought. For this purpose, we analysed yearly official harvested area reported at county level for the main crops and compared them to visible-shortwave infrared (VSWIR) spectra acquired by the Airborne Visible/Infrared Imaging-Spectrometer (AVIRIS-classic) over 2334 km² of the Central Valley each year. Multiple Endmember Spectral Mixture Analysis (MESMA) was applied to AVIRIS data to estimate green vegetation (GV), non-photosynthetic vegetation (NPV) and soil fractions in crop fields each year. MESMA and crop reports (R² = 0.9) showed that soil (i.e.; non-cultivated areas) increased during the summers of the drought; with the smallest GV area in 2015, the second year classified with exceptional drought in this period. According to MESMA, 34 % of the cropland was covered by GV in 2015, and 69.5·10⁴ ha according to the crop reports. MESMA also registered the highest value of soil area in 2015 (48 %). The year with most cultivated area was 2017, the wettest year in the studied period, with 54 % of the croplands covered by GV and 75.2·10⁴ ha. This study verified that the non-cultivated areas increased in the Central Valley during the exceptional drought period and validated the use of AVIRIS imagery to monitor broad-scale cropland use changes in future climatic extreme events.

Multi-Temporal LiDAR and Hyperspectral Data Fusion for Classification of Semi-Arid Woody Cover Species

Mapping the spatial distribution of woody vegetation is important for monitoring, managing, and studying woody encroachment in grasslands. However, in semi-arid regions, remotely sensed discrimination of tree species is difficult primarily due to the tree similarities, small and sparse canopy cover, but may also be due to overlapping woody canopies as well as seasonal leaf retention (deciduous versus evergreen) characteristics. Similar studies in different biomes have achieved low accuracies using coarse spatial resolution image data. The objective of this study was to investigate the use of multi-temporal, airborne hyperspectral imagery and light detection and ranging (LiDAR) derived data for tree species classification in a semi-arid desert region. This study produces highly accurate classifications by combining multi-temporal fine spatial resolution hyperspectral and LiDAR data (~1 m) through a reproducible scripting and machine learning approach that can be applied to larger areas and similar datasets. Combining multi-temporal vegetation indices and canopy height models led to an overall accuracy of 95.28% and kappa of 94.17%. Five woody species were discriminated resulting in producer accuracies ranging from 86.12% to 98.38%. The influence of fusing spectral and structural information in a random forest classifier for tree identification is evident. Additionally, a multi-temporal dataset slightly increases classification accuracies over a single data collection. Our results show a promising methodology for tree species classification in a semi-arid region using multi-temporal hyperspectral and LiDAR remote sensing data.

Assessment of the Usefulness of Spectral Bands for the Next Generation of Sentinel-2 Satellites by Reconstruction of Missing Bands

The Sentinel-2 constellation has been providing high spatial, spectral and temporal resolution optical imagery of the continental surfaces since 2015. The spatial and temporal resolution improvements that Sentinel-2 brings with respect to previous systems have been demonstrated in both the literature and operational applications. On the other hand, the spectral capabilities of Sentinel-2 appear to have been exploited to a limited extent only. At the moment of definition of the new generation of Sentinel-2 satellites, an assessment of the usefulness of the current available spectral bands seems appropriate. In this work, we investigate the unique information contained by each 20 m resolution Sentinel-2 band. A statistical quantitative approach is adopted in order to yield conclusions that are application agnostic: multivariate regression is used to reconstruct some bands, using the others as predictors. We conclude that, for most observed surfaces, it is possible to reconstruct the reflectances of most red edge or NIR bands from the rest of the observed bands with an accuracy within the radiometric requirements of Sentinel-2. Removing two of those bands could be possible at the cost of slightly higher reconstruction errors. We also identify mission scenarios for which several of the current Sentinel-2 bands could be removed for the next generation of sensors.