geospatial-hdf5

A geospatial interface for HDF5 using h5py. Support for the KEA image format, + others to come.

Requirements

• NumPy
• h5py
• affine
• enum34
• pandas (when working with the raster attribute table)

Installing

[sudo] python setup.py install

KEA format

See http://kealib.org/ for more information

KEA write example

import numpy
from geoh5 import kea

data = numpy.random.randint(0, 256, (6, 100, 100)).astype('uint8')

count, height, width = data.shape
kwargs = {'width': width,
          'height': height,
          'count': count,
          'dtype': data.dtype.name,
          'compression': 2,
          'no_data': 0,
          'chunks': (25, 25),
          'blocksize': 25}

with kea.open('file1.kea', 'w', **kwargs) as src:
    src.write(data, bands=range(1, count+1))

KEA read/write example

from geoh5 import kea

with kea.open('file1.kea', 'r') as src:
    # Read the first band
    src.read(1)

    # Read bands [4, 3, 2] and return in that order
    data = src.read([4,3,2])

    kwargs = {'width': src.width,
              'height': src.height,
              'count': 3,
              'transform': src.transform,
              'crs': src.crs,
              'compression': 4,
              'no_data': src.no_data[1],
              'chunks': (50, 50),
              'blocksize': 50,
              'dtype': src.dtype}

    with kea.open('file2.kea', 'w', **kwargs) as out_src:
        # Write the first band of data into band 3 on disk, etc..
        out_src.write(data, bands=[3,2,1])

KEA read raster attribute table example

from geoh5 import kea

ds = kea.open('file3.kea', 'r')

df = ds.read_rat(row_end=10)

print df

	Histogram	Red	Green	Blue	Alpha	NIR	RED	GREEN
0	0.0	0	0	0	255	0.000000	0.000000	0.000000
1	27056378.0	98	22	91	255	-998.799112	-998.631636	-998.348469
2	527.0	80	82	82	255	-999.000000	-999.000000	1154.679317
3	522.0	173	39	234	255	-980.936782	1603.070881	1203.337165
4	197.0	157	138	35	255	1797.959391	936.984772	781.756345
5	167.0	74	1	164	255	2829.137725	1424.215569	1256.730539
6	230.0	160	179	8	255	205.569565	287.030435	487.886957
7	259.0	136	119	126	255	2915.200772	1609.903475	1290.042471
8	114.0	130	213	252	255	2578.789474	1265.815789	1033.131579
9	102.0	165	206	207	255	2748.068627	1307.705882	1069.568627

Parallel (MPI) interface

If HDF5 & h5py have been built with the parallel option, then files can be created with the MPIO driver and you workflow can be run using MPI. See examples/parallel_kea_example.py for an example of an MPI workflow. See http://docs.h5py.org/en/latest/mpi.html for information on building parallel HDF5 and h5py.