ValueError: operands could not be broadcast together with shapes (0,1)

EO Products / Scripts
1

I use the following code and example provided in Sentinel-Hub to calculate the area of lakes:

Here

For me this code worked for one region but the other 7 got this error:

ValueError: operands could not be broadcast together with shapes (0,1) (818,1003,1)

2

orginal error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Input In [15], in <cell line: 4>()
      1 # The download task requires additional arguments at execution. These are linked to the node the task is in.
      2 download_node = workflow_nodes[0]
----> 4 result = workflow.execute(
      5     {
      6         download_node: {"bbox": dam_bbox, "time_interval": time_interval},
      7     }
      8 )

File c:\users\basin\appdata\local\programs\python\python38\lib\site-packages\eolearn\core\eoworkflow.py:152, in EOWorkflow.execute(self, input_kwargs, raise_errors)
    149 self.validate_input_kwargs(input_kwargs)
    150 uid_input_kwargs = {node.uid: args for node, args in input_kwargs.items()}
--> 152 output_results, stats_dict = self._execute_nodes(
    153     uid_input_kwargs=uid_input_kwargs, out_degrees=out_degrees, raise_errors=raise_errors
    154 )
    156 results = WorkflowResults(
    157     outputs=output_results, start_time=start_time, end_time=dt.datetime.now(), stats=stats_dict
    158 )
    160 LOGGER.debug("EOWorkflow ended with results %s", repr(results))

File c:\users\basin\appdata\local\programs\python\python38\lib\site-packages\eolearn\core\eoworkflow.py:205, in EOWorkflow._execute_nodes(self, uid_input_kwargs, out_degrees, raise_errors)
    202 stats_dict = {}
    204 for node in self._nodes:
--> 205     result, stats = self._execute_node(
    206         node=node,
    207         node_input_values=[intermediate_results[input_node.uid] for input_node in node.inputs],
    208         node_input_kwargs=uid_input_kwargs.get(node.uid, {}),
    209         raise_errors=raise_errors,
    210     )
    212     stats_dict[node.uid] = stats
    213     if stats.exception is not None:

File c:\users\basin\appdata\local\programs\python\python38\lib\site-packages\eolearn\core\eoworkflow.py:239, in EOWorkflow._execute_node(self, node, node_input_values, node_input_kwargs, raise_errors)
    237 LOGGER.debug("Computing %s(*%s, **%s)", node.task.__class__.__name__, str(task_args), str(node_input_kwargs))
    238 start_time = dt.datetime.now()
--> 239 result, is_success = self._execute_task(node.task, task_args, node_input_kwargs, raise_errors=raise_errors)
    240 end_time = dt.datetime.now()
    242 if is_success:

File c:\users\basin\appdata\local\programs\python\python38\lib\site-packages\eolearn\core\eoworkflow.py:267, in EOWorkflow._execute_task(task, task_args, task_kwargs, raise_errors)
    265 """Executes an EOTask and handles any potential exceptions."""
    266 if raise_errors:
--> 267     return task.execute(*task_args, **task_kwargs), True
    269 try:
    270     return task.execute(*task_args, **task_kwargs), True

Input In [12], in WaterDetectionTask.execute(self, eopatch)
     13 water_masks = np.asarray([self.detect_water(ndwi[..., 0]) for ndwi in eopatch.data["NDWI"]])
     15 # we're only interested in the water within the dam borders
---> 16 water_masks = water_masks[..., np.newaxis] * eopatch.mask_timeless["NOMINAL_WATER"]
     18 water_levels = np.asarray(
     19     [np.count_nonzero(mask) / np.count_nonzero(eopatch.mask_timeless["NOMINAL_WATER"]) for mask in water_masks]
     20 )
     22 eopatch[FeatureType.MASK, "WATER_MASK"] = water_masks

ValueError: operands could not be broadcast together with shapes (0,1) (818,1003,1)
3

Hi @www.wrm.ir ,

The error message indicated that the dimensions of the two arrays were not compatible for broadcasting in line 16:

16 water_masks = water_masks[..., np.newaxis] * eopatch.mask_timeless["NOMINAL_WATER"]

The issue could be that there is no data in your EOPatch.data["NDWI"]. Could you check if the data is downloaded and processed correctly as EOPatch before the water_detection step via removing water_detection from the workflow_nodes in cell 13 and run it again?

4

Thank you very much for your reply

After removing “water_detection” from line 13 of the code, I ran it again. This time, no errors were received during the processing of lines 15 and 16. But by checking the Sentinel 2 images in EO browser I know there is water in the selected area for the selected time period.

Is it possible that there is a problem in the creation of the WKT file? I didn’t have this problem for another scope.

5

Would you mind sharing one of the AOIs that ran into an error? I can take a closer look for you.

6

Thank you for your attention

I have attached 5 WKT files (AOIs) among them only “Maharlou.wkt” works correctly for the period November 13, 2022 to February 20, 2023.

Thankful.

7

Hi @www.wrm.ir ,

Those problematic geometries that led to the value error cover a huge area and there are no tile that covers the area more than 95%, which is the threshold we set for the filter in the example notebook. As a result, there is no data after filtering.

This can be solved by adding the following argument to the SentinelHubInputTask to create a simple mosaic for each available acquisition date which should cover the entire area.

time_difference=datetime.timedelta(days=1)
1 Like
8

excellent. This works well.

9

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.