Reduce ImageCollection to Image and export as Numpy

[Middlo]

I've tried to use the example you've provided on youtube where you extract an Image to a numpy array, however, when I try to reduce an ImageCollection consisting of a year of satellite images to a single Image, I'm no longer able to convert it to numpy, receiving an error "need at least one array to concatenate".

This is my code:

aoi = ee.Geometry.Polygon(
  [[[-110.8, 44.7],
    [-110.8, 44.6],
    [-110.6, 44.6],
    [-110.6, 44.7]]], None, False)

dataset = ee.ImageCollection('LANDSAT/LC08/C01/T1_SR').filterDate('2010-01-01', '2010-12-31')
dataset_bands = dataset.select(['B4', 'B3', 'B2'])

median = dataset_bands.reduce(ee.Reducer.median())

rgb_img = geemap.ee_to_numpy(median, region=aoi)
print(rgb_img.shape)

[Middlo]

Also is there any way to increase max pixel amount in image when converting to numpy? Maximum is 262144. Thanks!

[OfirMazor]

You are querying for Landsat 8 data from the year 2010 but this data is avialble since 2013-04-11 (to this day).
This is the reason there is no array (there is no image).
You can change the collection name to Landsat 7 ("LANDSAT/LE07/C01/T1_SR") or Landsat 5 if you need data from 2010.

Read more Landsat 8 specifications for GEE: https://developers.google.com/earth-engine/datasets/catalog/LANDSAT_LC08_C01_T1_SR

[Middlo]

I'm such an idiot, thank you so much. I was working with landsat 7 and didn't realised the section I copied was for landsat 8.
However, I have stumbled into an issue where now the numpy array is (1,1,3). Any solutions?

[giswqs]

See #287 (reply in thread)