The SNAP development team is pleased to announce a new SNAP release.
The different installers for the supported platforms can be downloaded.
In this release the development teams worked on general features like remote access of data, data processing, IO performance and the memory management. Also, sensor specific improvements and features have been implemented.
Noteworthy is also that we will drop the support for Python 2.7 for snappy with the next release. With version 9 of SNAP, a Python 3 environment will be required.
Here we just highlight the main improvements. Check out the full list of issues (>200) solved for SNAP 8 in our issue database: https://bit.ly/SNAP8_changelog.
In order to increase the performance of SNAP and improve the user experience some tasks have been performed.
The reading and writing performance of the BEAM-DIMAP format has been significantly improved. In some cases opening an image in SNAP took up to 5 minutes. Now, it only takes seconds. Similar magnitude of improvement has been achieved for the writing.
A first experimental step has also been made to lower the memory consumption. An experimental Tile Cache Operator has been implemented which gives more control of the memory usage. Please see the separate topic below for more details.
Reworked Product Library
The Product Library has been upgraded so that it can accommodate in a flexible way any type of sensor (radar, optical, atmospheric, etc.).
In addition, the display / interaction of / with product frames or quicklooks can be done on a 3D visualisation of Earth (NASA WorldWind).
The possibility to access (and search for) remote data was introduced by regarding remote data sources as independent SNAP plugins.
Besides, the parametrized search (with parameters in the form of values), it also handles the selection of the area of interest on the 3D visualisation of Earth.
New remote data repositories were added:
- Copernicus Scientific Data Hub (SciHub): for Sentinel-1, Sentinel-2 and Sentinel-3 data
- Amazon Web Services (AWS): for Sentinel-2 and Landsat-8 data
- Alaska Satellite Facility (ASF): for Sentinel-1 and ALOS data
- US Geological Survey (USGS): for Landsat-8 data
Windowed Reading of Products
Allows specifying a window (spatial subset) of either pixel coordinates or geographical coordinates, so that, instead of first opening a full product and then subsetting it to a region of interest, the reader will open directly the region of interest as a full-fledged product.
New NetCDF Library Version
We now use version 5.3 of the NetCDF library. If you use the NetCDF API directly exposed through SNAP you might be interested in the changes of the library.
You can check the documentation provided by unidata at https://docs.unidata.ucar.edu/netcdf-java/5.3/userguide/index.html.
A specific migration guide is provided (https://docs.unidata.ucar.edu/netcdf-java/5.3/userguide/upgrade_to_50.html).
We also removed the support for writing NetCDF3. They can still be read but not written anymore. If you still used the old format names ‘NetCDF-CF’ or ‘NetCDF-BEAM’ you should switch to NetCDF-CF or NetCDF-BEAM. You will benefit from smaller file sizes.
New Reimplemented GeoCoding
The pixel-based GeoCoding had some issues in the past. It was slow, and in some situations it just didn’t work and produced artifacts. These problems affect especially Sentinel-3 data. With SNAP 8 we now use a new implementation.
While it is now faster and more accurate you might notice slight differences to in geo-location to previous SNAP versions.
Experimental Tile Cache Operator
To further improve the memory management, especially during data processing, we introduced a special operator which can be used in processing graphs. This operator will cache only the data of its input. If this is used, and the general cache is disabled the amount of used memory can be reduced. The idea is that the cache memory is done automatically in the long-term, but to do so we need some experience on when data needs to be cached. Here we would need your feedback too.
More information can be found in the wiki.
Sensor specific improvements by Toolboxes
Capella and SAOCOM are now supported, and the support for RCM has been updated. There is a new Soil Moisture Toolkit for Radarsat-2/RCM.
Sentinel-3 SLSTR L1 oblique view is now correctly handled. The Sentinel-3 L2 FRP (Fire-Radiative-Products) are supported.
New Internal Default Data Format
With SNAP 8, we also want to introduce a new data format which will replace the BEAM-DIMAP. First we will just provide a BETA version as plugin shortly after the SNAP release. The version should not be used in a productive system or for operational services. But we would like to get your feedback on this new development, to further improve it.
More information will be provided with the release of the plugin.
SRTM 3 sec DEM
During testing we observed sometimes wrong values provided by the SRTM data. Instead of the expected elevation the DEM is rturning the elevation of an adjacent pixel. This is not always happening but depends on the download and on the operating system. It has been only observed on latest Unix systems. Deleting and redownloading the DEM file often helped to get the right values. For more information and updates on this issue please see https://senbox.atlassian.net/browse/SNAP-1344.
Enjoy the new release
The SNAP Team