Raster2dggs
DGGS indexer for raster data
Install / Use
/learn @manaakiwhenua/Raster2dggsREADME
raster2dggs
Python-based CLI tool to index raster files to DGGS in parallel, writing out to Parquet.
This is the raster equivalent of vector2dggs.
Currently this supports the following DGGSs:
And these geocode systems:
Contributions (particularly for additional DGGSs), suggestions, bug reports and strongly worded letters are all welcome.
Installation
This tool makes use of optional extras to allow you to install a limited subset of DGGSs.
If you want all possible:
pip install raster2dggs[all]
If you want only a subset, use the pattern pip install raster2dggs[a5] (for one) or pip install raster2dggs[h3,s2,isea4r] (for multiple).
A bare pip install raster2dggs will not install any DGGS backends.
Usage
raster2dggs --help
Usage: raster2dggs [OPTIONS] COMMAND [ARGS]...
Options:
--version Show the version and exit.
--help Show this message and exit.
Commands:
a5 Index raster data into the A5 DGGS
geohash Index raster data into the Geohash DGGS
h3 Index raster data into the H3 DGGS
healpix Index raster data into the HEALPix DGGS
isea4r Index raster data into the ISEA4R DGGS
isea7h Index raster data into the ISEA7H DGGS
isea9r Index raster data into the ISEA9R DGGS
ivea4r Index raster data into the IVEA4R DGGS
ivea7h Index raster data into the IVEA7H DGGS
ivea9r Index raster data into the IVEA9R DGGS
maidenhead Index raster data into the Maidenhead DGGS
rhp Index raster data into the rHEALPix DGGS
rtea4r Index raster data into the RTEA9R DGGS
rtea7h Index raster data into the RTEA7H DGGS
rtea9r Index raster data into the RTEA9R DGGS
s2 Index raster data into the S2 DGGS
raster2dggs h3 --help
Usage: raster2dggs h3 [OPTIONS] RASTER_INPUT OUTPUT_DIRECTORY
Ingest a raster image and index it to the H3 DGGS.
RASTER_INPUT is the path to input raster data; prepend with protocol like
s3:// or hdfs:// for remote data. OUTPUT_DIRECTORY should be a directory,
not a file, as it will be the write location for an Apache Parquet data
store, with partitions equivalent to parent cells of target cells at a fixed
offset. However, this can also be remote (use the appropriate prefix, e.g.
s3://).
Options:
-v, --verbosity LVL Either CRITICAL, ERROR, WARNING, INFO or
DEBUG [default: INFO]
-r, --resolution INTEGER RANGE H3 resolution to index [0<=x<=15; required]
-pr, --parent_res INTEGER RANGE
H3 parent resolution to index and aggregate
to. Defaults to max(0, resolution - 6)
[0<=x<=15]
-b, --band TEXT Band(s) to include in the output. Can
specify multiple, e.g. `-b 1 -b 2 -b 4` for
bands 1, 2, and 4 (all unspecified bands are
ignored). If unused, all bands are included
in the output (this is the default
behaviour). Bands can be specified as
numeric indices (1-based indexing) or string
band labels (if present in the input), e.g.
-b B02 -b B07 -b B12.
-u, --upscale INTEGER Upscaling factor, used to upsample input
data on the fly; useful when the raster
resolution is lower than the target DGGS
resolution. Default (1) applies no
upscaling. The resampling method controls
interpolation. [default: 1]
-c, --compression TEXT Compression method to use for the output
Parquet files. Options include 'snappy',
'gzip', 'brotli', 'lz4', 'zstd', etc. Use
'none' for no compression. [default:
snappy]
-t, --threads INTEGER Number of threads to use when running in
parallel. The default is determined based
dynamically as the total number of available
cores, minus one. [default: 19]
-a, --aggfunc [count|mean|sum|prod|std|var|min|max|median|mode]
Numpy aggregate function to apply when
aggregating cell values after DGGS indexing,
in case of multiple pixels mapping to the
same DGGS cell. [default: mean]
-d, --decimals INTEGER Number of decimal places to round values
when aggregating. Use 0 for integer output.
[default: 1]
-o, --overwrite
--warp_mem_limit INTEGER Input raster may be warped to EPSG:4326 if
it is not already in this CRS. This setting
specifies the warp operation's memory limit
in MB. [default: 12000]
--resampling [nearest|bilinear|cubic|cubic_spline|lanczos|average|mode|gauss|max|min|med|q1|q3|sum|rms]
Input raster may be warped to EPSG:4326 if
it is not already in this CRS. Or, if the
upscale parameter is greater than 1, there
is a need to resample. This setting
specifies this resampling algorithm.
[default: average]
-co, --compact Compact the cells up to the parent
resolution. Compaction is not applied for
cells without identical values across all
bands.
-g, --geo [point|polygon|none] Write output as a GeoParquet (v1.1.0) with
either point or polygon geometry. [default:
none]
--tempdir PATH Temporary data is created during the
execution of this program. This parameter
allows you to control where this data will
be written.
--version Show the version and exit.
--help Show this message and exit.
Visualising output
Output is in the Apache Parquet format, hive partitioned with the parent resolution as partition key. The example below is with -pr 3 with the H3 DGGS.
tree /home/user/example.pq
/home/user/example.pq
├── h3_03=83bb09fffffffff
│ └── part.0.parquet
└── h3_03=83bb0dfffffffff
└── part.0.parquet
Output can also be written to GeoParquet (v1.1.0) by including the -g/--geo parameter, which accepts:
polygonfor cells represented as boundary polygonspointfor cells represented as centre pointsnonefor standard Parquet output (not GeoParquet) ← this is the default if-g/--geois not used
GeoParquet output is useful if you want to use the spatial representations of the DGGS cells in traditional spatial analysis, or if you merely want to visualise the output.
Below are some ways to read and visualise it.
DuckDB
$ duckdb
DuckDB v1.4.1 (Andium) b390a7c376
Enter ".help" for usage hints.
Connected to a transient in-memory database.
Use ".open FILENAME" to reopen on a persistent database.
D INSTALL spatial;
D LOAD spatial;
D SELECT * FROM read_parquet('se_island.pq') LIMIT 7;
┌┌────────┬────────┬────────┬────────────────────────────────────────────────────────────────────────────────┬─────────────┬─────────┐
│ band_1 │ band_2 │ band_3 │ geometry │ s2_19 │ s2_08 │
│ float │ float │ float │ geometry │ varchar │ varchar │
├────────┼────────┼────────┼────────────────────────────────────────────────────────────────────────────────┼─────────────┼─────────┤
│ 0.0 │ 0.0 │ 0.0 │ POLYGON ((-176.17946725380486 -44.33542073938414, -176.17946725380486 -44.33… │ 72b47e01e24 │ 72b47 │
│ 0.0 │ 0.0 │ 0.0 │ POLYGON ((-176.18439390505398 -44.33543749229784, -176.18439390505398 -44.33… │ 72b47e02a14 │ 72b47 │
│ 0.0 │ 0.1 │ 0.1 │ POLYGON ((-176.18550630891403 -44.33547457195554, -176.18550630891403 -44.33… │ 72b47e1d54c │ 72b47 │
│ 0.0 │ 0.0 │ 0.0 │ POLYGON ((-176.17819578278952 -44.33537828938332, -176.17819578278952 -44.33… │ 72b47e01d64 │ 72b47 │
│ 0.1 │ 0.1 │ 0.3 │ POLYGON ((-176.18344039674218 -44.335553297533835, -176.18344039674218 -44.3… │ 72b47e0282c │ 72b47 │
│ 0.0 │ 0.0 │ 0.0 │ POLYGON ((-176.1789904558
