modis.io
HDF5Base
Parent class for interaction with HDF5 files
This class serves as a parent class for ModisRawH5 and ModisSmoothH5, enabling uniform chunked read and write of datasets and attributes to and from HDF5 files.
__init__(self, filename)
special
Initialize HDF5Base instance.
Creates an instance of the HDF5Base class. This is not stricly intended to be called
by itself (although it can be), but rather through a super call of the child class.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
filename |
str |
Full path to the HDF5 file. |
required |
read_chunked(self, dataset, xoffset=0, xchunk=None, arr_out=None)
Read data from dataset in a chunked manner.
The chunks are iterated in a row by column pattern, where
each chunk along row axis is yielded once the full
column size is read into the array. The chunking of
the colums (xchunk) can me modified, while the row chunking
(ychunk) is strictly defined by the dataset.
To enable the nsmooth functionality, an xoffset can be provided
to skip datapoints along the time dimension.
If no arr_out is provided, a new array will be created with the
necessary dimensions.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
dataset |
str |
Name of the dataset (expect 2d array). |
required |
xoffset |
int |
Offset for time-dimension (xaxis) in file. |
0 |
xchunk |
int |
Chunking along time-dimension. If not specified, it'll be read from the dataset. |
None |
arr_out |
ndarray |
Output array. |
None |
Returns:
| Type | Description |
|---|---|
ndarray |
Yields the output chunk as np.ndarray |
Exceptions:
| Type | Description |
|---|---|
AssertionError |
Raised when |
write_chunk(self, dataset, arr_in, xoffset=0, xchunk=None, yoffset=0)
Write chunk back to HDF5 file.
Writes complete chunk back to HDF5 file, iterating
over the time-dimension (x). The chunksize for x
can be adjusted manually using xchunk.
To implement nupdate behaviour, xoffset can be used to skip
prior datapoints in the time-dimension.
To write successive spatial chunks, yoffset has to be provided (the default is 0,
as it starts at the top left).
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
dataset |
str |
Name of the dataset (expect 2d array). |
required |
arr_in |
ndarray |
Input data to be written to file. |
required |
xchunk |
int |
Chunking along row (x) axis. If not specified, it'll be read from the dataset. |
None |
xoffset |
int |
Offset for colums (xaxis) in file. |
0 |
yoffset |
int |
Offset for rows (yaxis) in file. |
0 |
Returns:
| Type | Description |
|---|---|
bool |
Returns |
Exceptions:
| Type | Description |
|---|---|
AssertionError |
Raised when |
HDFHandler
Class to handle reading from MODIS HDF files.
This class enables reading specific subdatasets and attributes from the raw MODIS HDF files.
iter_handles(self)
Iterates over all open dataset handles
coming from open_datasets and returns a Tuple with index and a
gdal.Dataset for each.
Returns:
| Type | Description |
|---|---|
Tuple[int, gdal.Dataset] |
Tuple with index and corresponding |
open_datasets(self)
Opens the selected subdataset from all files within a context manager and stores them in a class variable. When the context manager closes, the refereces are removed, closing all datasets.
read_chunk(x, **kwargs)
staticmethod
Reads a chunk of an opened subdataset.
The size of the chunk being read is defined by the
values passed to gdal.Dataset.ReadAsArray with kwargs and can
be as big as the entire dataset.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
x |
gdal.Dataset |
GDAL Dataset. |
required |
**kwargs |
dict |
kwargs passed on to |
{} |
Returns:
| Type | Description |
|---|---|
ndarray |
Requested chunk as |