Multiple fixes for minimizing the impact of the "Long Paths" issue on Windows OS. As long as you managed to uncompress a Sentinel-2 product on your HDD from the Windows Explorer, it will open correctly inside SNAP.
The specific 10m, 20m and 60m readers have been removed. They were doing an automatic nearest neighbor resampling of any band to the target resolution, and this work is much better done with the Resampling operator, available via the Raster->Geometric Operations->Resampling menu. The Resampling operator allows for more choices of the target resolution, and allows one to choose the up and downsampling algorithms.
This also has the advantage that products where all granules are in the same UTM zone are read automatically by a single reader, so does not trigger the combo box for choosing the actual reader to use anymore. Products containing granules in several distincts UTM zones still require the user to choose which UTM zone to read.
The Level-2A reader, reading products produced by Sen2Cor, has been updated to provide multi-size products, in line with what is done for L1C products. It chooses the most precise resolution available for each band : L1C products processed by sen2cor up to 20m will have B1 and B9 at 60m, and all other bands at 20m. L1C products processed by sen2cor up to 10m will have B1 and B9 at 60m, B5/B6/B7/B8a/B11/B12 at 20m, and B2/B3/B4/B8 at 10m.
It is now possible to drag-and-drop the directory containing a S2 product into the Product Explorer window to open them. No need to go find the xml file anymore.
The scale factor of 10000 used to convert reflectance from their raw integer value to a meaningful reflectance value in [0, 1] is applied during product reading.
In the L2A product reader, appropriate scaling factors are applied to AOT and WVP bands to produce physical values.
The granule-specific metadata were not easily distinguishable. They now contain the MGRS tile identifier.
A small issue prevented L2A products from being used with the GPT. This is now fixed.
The angles are now read at their original resolution, as provided in the metadata (5000 m). This will reduce a lot the HDD space requirements when converting S2 products to other format (BEAM-DIMAP, GeoTIFF).
The angles grids of adjacent tiles are automatically mosaicked. Since their resolution (5000m) is not compatible with the distance between two adjacent tiles (99960m), adjacent angles grid are interpolated to produce a mosaic at 5000m resolution. This particular issue has been reported to the S2-MPC.
Viewing angles retrieval from the metadata was buggy in previous SNAP versions. Now the Sentinel-2 reader produces one viewing angle grid for each band independently, merging the different viewing angle grids of each detector into one single image. For ease of use, a mean viewing angle grid is also provided, which averages the data provided by each band-specific grid. Their actual use in processors is still problematic since S2 products do not provide precise enough information about the detector from which a reflectance band pixel has been interpolated from. In the neighborhood of a detector switch, the viewing angles thus cannot be trusted.
S2 products now include proper time information (Sensing start/stop time).
Masks are now grouped by category, to avoid the masks folder to be too populated.
##### Level-2A Scene Classification colors
The different indices of the Scene Classification now follow the Sen2Cor documentation. For consistency, the masks generated for each class in the Scene Classification band also follow the same colors.
Added support for the Level-3 products as generated by Sen2Three, the multitemporal synthesis processor, developed by Telespazio Vega Gmbh
OpenJPEG, the underlying tool which decompress the JP2 images has been updated to include important performance optimizations.
The reading of cloud masks (opaque_clouds and cirrus_clouds) available in L1C products, and copied as-is in L2A products by Sen2Cor, can be disabled in the preferences (Tools->Options->S2TBX). For cloudy S2 products, the loading time can be very high because of the reading of the cloud masks. In addition, L2A products provide a far more optimized opaque clouds/cirrus masks in raster format, as part of the Scene Classification band, making the cloud masks available in L1C products useless.
The interpretation of GML masks has been optimized
RGB rendering of Spot 6/7 products appeared overly dark. These has been fixed, by not using the histogram provided in the metadata.
The order of the bands has been fixed for the default profile.
The Sentinel-2 Toolbox caches uncompressed regions of the JPEG2000 files read by the Sentinel-2 reader, the generic JP2 reader and the Spot 6/7 reader. Those files are stored in the user home directory and can grow big.
The cache can now be emptied regularly at startup. The frequency of the cache clean up can be tuned in the preferences pane. The default behavior for new installation is to remove from the cache the files that are more-than-a-week old.
The Sentinel-2 Toolbox now provides a biophysical products processor, developped by CS and INRA.
The processor computes LAI, FAPAR, FCOVER, CWC, CCC from top-of-canopy reflectances.
The Sentinel-2 Toolbox now includes the Idepix Processor for Sentinel-2, developped by Brockmann Consult Gmbh.
The Idepix Processor provides a pixel classification into properties such as clear/cloudy, land/water, snow, ice etc.
The Sentinel-3 Toolbox provides similar functionnality for MODIS and Landsat-8 data.
The Sentinel-2 Toolbox now includes an important number of well-known radiometric indices processors, developped by CS-Romania.
The following indices are made available :