VRT support

[BoehnkeC]

GDAL offers building of virtual raster datasets, so does rasterio and this gives us the possibility to have it implemented with an interface easy to use.

Consider the following:

import os
import rasterio.shutil

from ukis_pysat import raster as pysat_raster

vrt_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), 'tests', 'testfiles', 'vrt')

for sample in os.listdir(vrt_dir):
    src = pysat_raster.Image(os.path.join(vrt_dir, sample))
    with rasterio.vrt.WarpedVRT(src.dataset, crs=src.dataset.crs, resampling=rasterio.enums.Resampling.nearest) as vrt:
        rasterio.shutil.copy(vrt, os.path.join(vrt_dir, 'test.vrt'), driver='VRT')

vrt = pysat_raster.Image(os.path.join(vrt_dir, 'test.vrt'))

[BoehnkeC]

This topic is a bit more complicated. A first draft is provided with this commit.

Having an intial source raster and a bunch of auxiliary rasters, overlapping the source raster, we want to create a mosaic from the auxiliary rasters. This can be clipped to the extent of the source raster later on.

src = Image(source_raster_file)
array, transform = src.mosaic_from_vrt([Image(item) for item in list_of_aux_raster_files])

We call a new method from ukis_pysat.raster, namely mosaic_from_vrt. It is referenced to the source raster object and accepts a list of auxiliary raster objects.

Within mosaic_from_vrt we perform the following steps:

with rasterio.vrt.WarpedVRT(aux_raster_object.dataset, **vrt_options) as vrt:
    target = aux_raster_object.__update_dataset(crs=vrt.crs, transform=vrt.transform)

This produces a VRT dataset, which is basically an XML styled file with transformation information and reference to its original raster object. The VRT dataset can be appended to a list of VRT datasets.

rasterio.merge.merge() accepts a list of datasets and merges them to a resulting array with transformation information.

A few thoughts

• rasterio.vrt.WarpedVRT keeps a reference of the original dataset in memory: <SourceFilename relativeToVRT="0">/vsimem/5751c324-e285-49ea-ac5a-32b8be93a3fe.</SourceFilename>. Maybe we find a way of referencing files on disk as they are usually located there.
• Also this comment mentions performance issues of VRT compared to directly warping rasters to a given extent. We have to check that.

[BoehnkeC]

I compared exploiting the VRT method with the direct use of rasterio.merge.merge, i.e. merging from VRT datasets or building the mosaic from the auxiliary rasters directly. The test revealed that using VRT support is as twice as much slower compared to not using VRT support and building the mosaic right away. Also the memory consumption is a bit higher when using VRT support. The changes were provided with this commit.

[BoehnkeC]

As they is no need for VRT implementation we decided to drop this issue. Further development will be conducted in #65.