modis.window

ModisMosaic

Warning

only implemented for MOD/MYD 11/13 products!

Class for subsetting or mosaicing HDF5 files.

This class can be used to mosaic multiple raw or smooth HDF5 files or to subset a single HDF5 file (both tiled and global products). The generated data will exported as GeoTIFF using the Python GDAL bindings.

`init(self, input_files)` `special`

Initialize instance of ModisMosaic class.

If multiple HDF5 files are specified as input_files, a mosaic of the files will be created (need to be of same product and tile/global). It's required that all of the files share the same temporal axis. If only one file is specified, either the full file is exported or a subset is created.

Parameters:

Name	Type	Description	Default
`input_files`	`List[str]`	List of paths to HDF5 files for mosaicing.	required

Exceptions:

Type	Description
`HDF5MosaicError`	If temporal axis is not conistent with multiple files (i.e. contained dates don't line up)

`generate_mosaics(self, dataset, targetdir, target_srs=None, aoi=None, overwrite=False, force_doy=False, prefix=None, start=None, stop=None, clip_valid=False, round_int=None, **kwargs)`

Generate GeoTIFF mosaics/subsets.

This method is creating a GeoTiff mosaic/subsets from the HDF5 files passed to the class instance. Internally, a virtual raster for each timestep is created, warped to the desired SRS (target_srs) and then optionally clipped to the area of interest (aoi as list of corner coordinates). If the dataset is already in the desired SRS, the warping is skipped. To transform the resolution of meters to degrees in EPSG:4326, the original resolution of the dataset in meters is divided by 112000.

The exported dates can be limited with start and stop. By default, existing GeoTIFFs are not overwritten. This can be forced with setting overwrite to True.

The data to be exported can be clipped to the valid data range of the MODIS product being processed (if applicable) to the setting clip_valid to True. The data can also be rounded by passing an integer to round_int. This integer will be passed on to np.round.

To modify the output naming, a prefix can be specified and if force_doy is set to True, the timestamp will be forced to the format YYYYjDDD independent of temporal interpolation.

More fine scaled adjustment of the output datasets can be achieved with passing keyword arguments to gdal.WarpOptions and gdal.TranslateOptions.