JTS has been upgraded to version 1.16. This marks a significant change in package naming, the library has switched from "com.vividsolutions" to "org.locationtech" packages and all GeoTools code has been updated to follow suit.
JTS classes are widely used in GeoTools, as a result the client code using GeoTools will also have to be modified to follow suite. Thankfully the changes amount to a search and replace and can be easily automated, please check the
upgrade instructions available in the GeoTools site.
Units library migrated to JSR-363
GeoTools was using an aging and un-maintained units library and this development round moved it to a modern and maintained set of packages compatible with Java 10. Like the JTS upgrade this might require the client code to be updated, both in terms of packages and in terms of classes and methods being used.
We have provided a
Units utility class, providing an easy way to match unit defined at runtime with the the correct constant.
Unit deg = Units.autoCorrect(SI.RADIAN.multiply(0.0174532925199433));
Thanks to César Martínez Izquierdo and the participants of the OSGeo Code sprint for this work.
Various other assorted dependencies upgrades
The release upgrades many dependencies to newer versions, in particular:
- Batik upgraded to version 1.10
- Commons-lang to version 3.7
- Commons-io to version 2.6
- Mysql JDBC driver to 5.1.46
- HSQLDB to 2.4.1
- SQLite-JDBC to version 3.23.1
- GeographicLib upgrades to version 1.49
These upgrades should not cause major incompatibilities, but look for for potential API changes.
Channel selection name allow expressions
GeoTools 20.x allows expressions to be used in SourceChannelName SLD elements, and in their code counterpart, thus allowing dynamic channel selection. This is welcomed news for anyone building applications that display multispectral or hyperspectral data, thus avoiding the need to build many different styles for the various interesting false color combinations.
Here is an SLD example:
<RasterSymbolizer>
<ChannelSelection>
<RedChannel>
<SourceChannelName>
<ogc:Function name="env">
<ogc:Literal>B1</ogc:Literal>
<ogc:Literal>2</ogc:Literal>
</ogc:Function>
</SourceChannelName>
</RedChannel>
<GreenChannel>
<SourceChannelName>
<ogc:Function name="env">
<ogc:Literal>B2</ogc:Literal>
<ogc:Literal>5</ogc:Literal>
</ogc:Function>
</SourceChannelName>
</GreenChannel>
<BlueChannel>
<SourceChannelName>
<ogc:Function name="env">
<ogc:Literal>B3</ogc:Literal>
<ogc:Literal>7</ogc:Literal>
</ogc:Function>
</SourceChannelName>
</BlueChannel>
</ChannelSelection>
<RasterSymbolizer>
Map algebra
This release adds support for an efficient map algebra package knows as Jiffle. Jiffle has been the work of a former GeoTools contributor, Michael Bedwards, that has been salvaged, upgraded to support Java 8, and integrated in jai-ext. From the there support has been added into the GeoTools gt-process-raster module, to be used either directly or as a rendering transformation.
The following SLD style calls onto Jiffle to perform a NDVI calculation on top of Sentinel 2 data:
<?xml version="1.0" encoding="UTF-8"?>
<StyledLayerDescriptor xmlns="http://www.opengis.net/sld" xmlns:ogc="http://www.opengis.net/ogc" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.opengis.net/sld
http://schemas.opengis.net/sld/1.0.0/StyledLayerDescriptor.xsd" version="1.0.0">
<NamedLayer>
<Name>Sentinel2 NDVI</Name>
<UserStyle>
<Title>NDVI</Title>
<FeatureTypeStyle>
<Transformation>
<ogc:Function name="ras:Jiffle">
<ogc:Function name="parameter">
<ogc:Literal>coverage</ogc:Literal>
</ogc:Function>
<ogc:Function name="parameter">
<ogc:Literal>script</ogc:Literal>
<ogc:Literal>
nir = src[7];
vir = src[3];
dest = (nir - vir) / (nir + vir);
</ogc:Literal>
</ogc:Function>
</ogc:Function>
</Transformation>
<Rule>
<RasterSymbolizer>
<Opacity>1.0</Opacity>
<ColorMap>
<ColorMapEntry color="#000000" quantity="-1"/>
<ColorMapEntry color="#0000ff" quantity="-0.75"/>
<ColorMapEntry color="#ff00ff" quantity="-0.25"/>
<ColorMapEntry color="#ff0000" quantity="0"/>
<ColorMapEntry color="#ffff00" quantity="0.5"/>
<ColorMapEntry color="#00ff00" quantity="1"/>
</ColorMap>
</RasterSymbolizer>
</Rule>
</FeatureTypeStyle>
</UserStyle>
</NamedLayer>
</StyledLayerDescriptor>
The performance is good enough for interactive display, and the result looks as follows:
PostGIS store improvements
The PostGIS datastore has been for years the only one that could encode a few filter functions used in filters down into native SQL, but it required a datastore creation flag to be enabled.
Starting with this release it will do so by default.
The functions supported for SQL encoding by the store are:
- String functions: strConcat, strEndsWith, strStartsWith, strEqualsIgnoreCase, strIndexOf, strLength, strToLowerCase, strToUpperCase, strReplace, strSubstring, strSubstringStart, strTrim, strTrim2
- Math functions: abs, abs_2, abs_3, abs_4, ceil, floor
- Date functions: dateDifference
Along with improvements in the handling of GroupByVisitor, this makes for efficient histogram computation directly in the database, while retaining the ability to share the same code with other stores that do not have the same optimizations.
This release adds support for "array" data type in the store, with full reading and writing support, as well as native filtering (with index support, where feasible).
Finally, it's now possible to read geometries with measures from PostGIS, also thanks to JTS improved CoordinateSequence API, and encode the results in GML (the encoding is subject to a configuration flag, as GML does not officially support M encoding). The work will continue in the next few month in order to cover more formats.
Image mosaic improvements
The image mosaic module never sleeps, in this release we see the following improvements:
- Support for remote images (e.g. S3 or Minio). In order to leverage this the mosaic index will have to be created up-front (manually, or with some home grown tools)
- A new "virtual native resolution" read parameter allows the mosaic to compose outputs respecting a native resolution other than its native one (useful in case you want to give selective resolution access to different users)
- Supporting multiple WKBs footprint for pixel precise overviews masking
- A new read mask parameter allows to cut the image to a given geometry (again, helps in providing different selective views to different users)
- Speed up NetCDF mosaics by allowing usage of stores coming from a repository, instead of re-creating them every time a NetCDF file is needed (to be setup in the auxiliary store config file, while the repository instance needs to be passed as a creation hint).
- The mosaic now works against images without a proper CRS, setting them into the "CRS not found" wildcard CRS, aka "EPSG:404000"
App-schema improvements
The app-schema module got significant performance and functionality improvements since 19.x series, in particular:
- Better delegation of spatial filters on nested properties to native database
- Improved support for multiple nested attribute matches in filters
- It is now possible to use Apache SOLR as a data source for app-schema
- The configuration files can be hosted on a remote HTTP server
Modules moving up and down
This release sees a number of modules moving down to unsupported due to lack of maitainers, unsupported module being removed, while other modules graduate to supported status. In particular:
- The unsupported image-collection, sfs, efeature, caching, feature-aggregate and geotiff-new module have been removed from the code base (you can still find them in older releases and version control history)
- The OGR and GTOPO30 modules have been downgraded to unsupported status due to lack of maintainership
- The MongoDB module moves up to supported status
Building on Windows
The GeoTools project builds again on Windows (with tests) and all new pull requests are verified using AppVeyor.
GeoTools is looking for Windows developers to join and help keep the build passing, as well as locating and squashing a few intermittent failures we're still experiencing.
Assorted improvements
