An on-demand geospatial compute engine.

Dynamic Compute

Dynamic-Compute is a map computation engine. It enables users to dynamically generate maps from a composable set of Python operations. Together, these properties enable data scientists in the building of complex GIS applications.

The Dynamic Compute API can be found as Python package earthdaily-earthone-dynamic-compute, and can be imported as earthdaily.earthone.dynamic_compute.

class Mosaic(graft, bands=None, product_id=None, start_datetime=None, end_datetime=None, image_ids=None)

Class wrapper around mosaic operations

argmax(axis)

Apply np.ma.argmax to the ComputeMap

Parameters:: axis – (str) Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”]. If called on an ImageStack, must be in [“bands”, “images”]
Returns:: Mosaic

argmin(axis)

Apply np.ma.argmin to the ComputeMap

Parameters:: axis – (str) Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”]. If called on an ImageStack, must be in [“bands”, “images”]
Returns:: Mosaic

aspect(resolution_x=None, resolution_y=None)

Compute the aspect of the mosaic

Returns:: Mosaic object with the aspect
Return type:: Mosaic

clip(lo, hi)

Generate a new Mosaic that is bounded by low and hi.

Parameters:

lo (Number) – Lower bound
hi (Number) – Upper bound

Returns:

bounded – New Mosaic object that is bounded

Return type:

Mosaic

compute(aoi, **kwargs)

Evaluate this ComputeMap for a particular AOI

Parameters:: aoi (earthdaily.earthone.geo.GeoContext) – GeoContext for which to compute evaluate this ComputeMap
Returns:: results – Evaluation of self for this AOI. The return will be either an array, properties as a list, properties as dict, or both as a DotDict
Return type:: Union[Array, List, Dict, DotDict]

concat_bands(other)

Create a new Mosaic that stacks bands. This call does not mutate this

Parameters:: other (Union[Mosaic, str]) – concat the bands of this mosiac with those of other, if other is a Mosaic. otherwise assume other is a list of bands and concat this with those bands.
Returns:: stack – Mosaic object with stacked bands
Return type:: Mosaic

filled(fill_val)

Generate a new Mosaic that has masked values filled with specified value.

Parameters:: fill_val (Number) – value to fill masked pixels
Returns:: bounded – New Mosaic object that is bounded
Return type:: Mosaic

classmethod from_image_ids(image_ids, bands, **kwargs)

Create a new Mosaic object from a list of image IDs

Parameters:

image_ids (list) – List of image IDs for the images that will be mosaicked
bands (Union[str, List[str]]) – A space-separated list of bands within the product, or a list of strings.

Returns:

m – New mosaic object.

Return type:

Mosaic

classmethod from_product_bands(product_id, bands, start_datetime=None, end_datetime=None, **kwargs)

Create a new Mosaic object

Parameters:

product_id (str) – ID of the product from which we want to access data
bands (Union[str, List[str]]) – A space-separated list of bands within the product, or a list of strings.
start_datetime (Optional[Union[str, datetime.date, datetime.datetime]) – Start date for mosaic
end_datetime (Optional[Union[str, datetime.date, datetime.datetime]) – End date for mosaic

Returns:

m – New mosaic object.

Return type:

Mosaic

gradient_x(resolution=None)

Compute the gradient in the x (E-W) direction

Returns:: Mosaic object with the E-W gradient
Return type:: Mosaic

gradient_y(resolution=None)

Compute the gradient in the y (N-S) direction

Returns:: Mosaic object with the N-S gradient
Return type:: Mosaic

mask(mask)

Apply a mask as a delayed object. This call does not mutate this

Parameters:: mask (ComputeMap) – Delayed object to use as a mask.
Returns:: masked – Masked mosaic.
Return type:: Mosaic

max(axis)

Apply np.ma.max to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

mean(axis)

Apply np.ma.mean to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

median(axis)

Apply np.ma.median to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

min(axis)

Apply np.ma.min to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

pick_bands(bands)

Create a new Mosaic object with the specified bands and the product-id of this Mosaic object

Parameters:: bands (str) – A space-separated list of bands within the product, or a list of bands as strings
Returns:: m – New mosaic object.
Return type:: Mosaic

reduce(reducer, axis='bands')

Call a reduction function on this Mosaic

Parameters:

reducer (Callable) – function to reduce Mosaic
axis (str) – Axis over which to call the reducer, must be in [“bands”].

Raises:

NotImplementedError – axis must be bands

Returns:

Mosaic

rename_bands(bands): Rename the bands of an array.

slope(resolution_x=None, resolution_y=None)

Compute the slope of the mosaic

Returns:: Mosaic object with the slope
Return type:: Mosaic

std(axis)

Apply np.ma.std to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

sum(axis)

Apply np.ma.sum to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

tile_layer(name=None, scales=None, colormap=None, checkerboard=True, classes=None, val_range=None, alpha=None, log_level=10, **parameter_overrides)

A .DynamicComputeLayer for this Mosaic.

Generally, use Mosaic.visualize for displaying on map. Only use this method if you’re managing your own ipyleaflet Map instances, and creating more custom visualizations.

An empty Mosaic will be rendered as a checkerboard (default) or blank tile.

Parameters:

name (str) – The name of the layer.
scales (list of lists, default None) –
The scaling to apply to each band in the Mosaic.

If Mosaic contains 3 bands, scales must be a list like [(0, 1), (0, 1), (-1, 1)].

If Mosaic contains 1 band, scales must be a list like [(0, 1)], or just (0, 1) for convenience

If None, each 256x256 tile will be scaled independently. based on the min and max values of its data.
colormap (str, default None) –

The name of the colormap to apply to the Mosaic. Only valid if the
Mosaic has a single band.
classes (list, default None) – Whether or not there are classes in the layer, indicating it is classified
checkerboard (bool, default True) – Whether to display a checkerboarded background for missing or masked data.
log_level (int, default logging.DEBUG) –
Only listen for log records at or above this log level during tile computation. See https://docs.python.org/3/library/logging.html#logging-levels

for valid log levels.
**parameter_overrides (JSON-serializable value, Proxytype, or ipywidgets.Widget) –
Values—or ipywidgets—for any parameters that this Mosaic depends on.

If this Mosaic depends on dc.widgets, you don’t have to pass anything for those—any widgets it depends on are automatically linked to the layer. However, you can override their current values (or widgets) by passing new values (or ipywidget instances) here.

Values can be given as Proxytypes, or as Python objects like numbers, lists, and dicts that can be promoted to them. These arguments cannot depend on any parameters.

If an ipywidgets.Widget is given, it’s automatically linked, so updating the widget causes the argument value to change, and the layer to update.

Once these initial argument values are set, they can be modified by assigning to ~.DynamicComputeLayer.parameters on the returned DynamicComputeLayer.

For more information, see the docstring to ParameterSet.

Returns:

layer

Return type:

.DynamicComputeLayer

unpack_bands(bands)

Create a tuple of new Mosaic objects with one for each bands.

Parameters:: bands (str) – A space-separated list of bands within the product.
Returns:: m – New mosaic object per band passed in.
Return type:: Tuple[Mosaic, …]

update_resampler(resampler)

Create a new mosaic object with an updated resampler

Parameters:: resampler (eo.catalog.ResampleAlgorithm) – New resampler algorithm
Returns:: updated_mosaic – Mosaic with updated resampling
Return type:: Mosaic

visualize(name, map, colormap=None, scales=None, checkerboard=True, classes=None, val_range=False, alpha=None, **parameter_overrides)

Visualize this Mosaic instance on a map. This call does not mutate this :param name: Name of this layer on the map :type name: str :param map: IPyleaflet map on which to add this mosaic as a layer :type map: ipyleaflet.leaflet.Map :param colormap: Optional colormap to use :type colormap: str :param scales: List of lists where each sub-list is a lower and upper bound. There must be

as many sub-lists as bands in the mosaic

Parameters:: classes (list, default None) – Whether or not there are classes in the layer, indicating it is classified
Returns:: layer – IPyleaflet tile layer on the map.
Return type:: lyr

class ImageStack(full_graft, bands=None, product_id=None, start_datetime=None, end_datetime=None, pad=0, resampler=ResampleAlgorithm.NEAR, predicate_filter=None, sort_by=None, ascending=None)

argmax(axis)

Apply np.ma.argmax to the ComputeMap

Parameters:: axis – (str) Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”]. If called on an ImageStack, must be in [“bands”, “images”]
Returns:: Mosaic

argmin(axis)

Apply np.ma.argmin to the ComputeMap

Parameters:: axis – (str) Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”]. If called on an ImageStack, must be in [“bands”, “images”]
Returns:: Mosaic

clip(lo, hi)

Generate a new ImageStack that is bounded by low and hi.

Parameters:

lo (Number) – Lower bound
hi (Number) – Upper bound

Returns:

bounded – New ImageStack object that is bounded

Return type:

ImageStack

compute(aoi, **kwargs)

Evaluate this ComputeMap for a particular AOI

Parameters:: aoi (earthdaily.earthone.geo.GeoContext) – GeoContext for which to compute evaluate this ComputeMap
Returns:: results – Evaluation of self for this AOI. The return will be either an array, properties as a list, properties as dict, or both as a DotDict
Return type:: Union[Array, List, Dict, DotDict]

concat_bands(other)

Create a new ImageStack that stacks bands. This call does not mutate this

Note that per-image metadata for the returned ImageStack instance is taken from self not other.

Parameters:: other (ImageStack) – concat the bands of this ImageStack with those of other
Returns:: stack – ImageStack object with stacked bands
Return type:: ImageStack

filled(fill_val)

Generate a new ImageStack that has masked values filled with specified value.

Parameters:: fill_val (Number) – value to fill masked pixels
Returns:: filled – New ImageStack object that is filled
Return type:: ImageStack

filter(pred)

Filter an image stack to based on image properties.

Parameters:: pred (eo.catalog.properties.OpExpression) – Predicate for image selection
Returns:: New ImageStack object.
Return type:: ImageStack

filter_by_id(id_list)

Filter an image stack to based on image properties.

Parameters:: id_list (List[str]) – List of image ids.
Returns:: New ImageStack object.
Return type:: ImageStack

classmethod from_product_bands(product_id, bands, start_datetime, end_datetime, predicate_filter=None, sort_by=None, ascending=True, **kwargs)

Create a new ImageStack object

Parameters:

product_id (str) – ID of the product from which we want to access data
bands (Union[str, List[str]]) – A space-separated list of bands within the product, or a list of strings.
start_datetime (Union[str, datetime.date, datetime.datetime]) – Start date for image stack
end_datetime (Union[str, datetime.date, datetime.datetime]) – End date for image stack
predicate_filter (eo.catalog.properties.OpExpression) – Optional kwarg that allows filtering of scenes prior to creating the ImageStack. The benefit of using this over the .filter() method is that it allows filtering of metadata before an ImageCollection is created, which allows less data to be requested from platform. The downside is this filter cannot be changed without changing the ImageStack object.
sort_by (str) – Optional kwarg that allows for sorting of images. Follows the same logic as sorting for catalog. See for more details: https://docs.earthdaily.earthone.com/earthdaily.earthone/catalog/docs/image.html
ascending (bool) – Optional kwarg that is used in sorting. Defaults to True. If sort_by is not provided, this kwarg will be ignored.

Returns:

m – New ImageStack object.

Return type:

ImageStack

groupby(grouping_func)

Perform a grouping function over either images or bands and return an ImageStackGroupBy object.

Parameters:: grouping_func (Callable[[dict], Hashable]) – Function to pick out the values to group by. The function should take as input a dotted-dict representing metadata for an image and should return a hashable value upon which groups will be build.
Return type:: ImageStackGroupBy object.

Example

>>> import earthdaily.earthone.dynamic_compute as dc 
>>> m = dc.map 
>>> m 
>>> sigma0_vv = dc.ImageStack.from_product_bands( 
        "esa:sentinel-1:sigma0v:v1", "vv", "20230101", "20230401" 
    ) 
>>> # group by acquired month
>>> grouped_sigma = sigma0_vv.groupby(lambda x: x.acquired.month) 
>>> # loop through each grouping, applying a max reducer and visualize it on the map
>>> for group_name, image_stack in grouped_sigma.groups.compute(m.geocontext()): 
        image_stack.max(axis="images").visualize(str(group_name), m, colormap="turbo") 

length()

Proxy object for the length of this image stack

Returns:: compute_map – Proxy object for the length
Return type:: ComputeMap

mask(mask)

Apply a mask as a delayed object. This call does not mutate this

Parameters:: mask (ComputeMap) – Delayed object to use as a mask.
Returns:: masked – Masked mosaic.
Return type:: Mosaic

max(axis)

Apply np.ma.max to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

mean(axis)

Apply np.ma.mean to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

median(axis)

Apply np.ma.median to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

min(axis)

Apply np.ma.min to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

pick_bands(bands)

Create a new ImageStack object with the specified bands and the product-id of this ImageStack object

Parameters:: bands (str) – A space-separated list of bands within the product, or a list of bands as strings
Returns:: m – New mosaic object.
Return type:: ImageStack

reduce(reducer, axis='images')

Perform a reduction over either images or bands. Note that this does not mutate self.

Parameters:

reducer (Callable[[np.ndarray], np.ndarray]) – Function to perform the reduction
axis (str) – Axis over which to reduce, either “bands” or “images”

Returns:

new_obj – Reduced object, either a Mosaic if axis is “images” or an ImageStack if axis is “bands”

Return type:

Union[Mosaic, ImageStack]

rename_bands(bands): Rename the bands of an array.

std(axis)

Apply np.ma.std to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

sum(axis)

Apply np.ma.sum to the ComputeMap

Parameters:: axis (str) – Axis over which to call the reducer. If called on a Mosaic, must be in [“bands”].
Returns:: ComputeMap

unpack_bands(bands)

Create a tuple of new ImageStack objects with one for each bands.

Parameters:: bands (str) – A space-separated list of bands within the product.
Returns:: m – New mosaic object per band passed in.
Return type:: Tuple[ImageStack, …]

update_resampler(resampler)

Create a new mosaic object with an updated resampler

Parameters:: resampler (eo.catalog.ResampleAlgorithm) – New resampler algorithm
Returns:: updated_mosaic – Mosaic with updated resampling
Return type:: ImageStack