Get Numpy arrays from Google Earth Engine

Question

I would like to get the numpy arrays of some Google Earth Engine collection.

For example here my intent would be to obtain a 4d numpy array with dimensions [timestamp, height, width, bands] (or whatever order) with the Sigma0 data of a given region, defined by its bounds (here in UTM projection).

e.g.

import ee
import geopandas as gpd
from shapely.geometry import box
ee.Initialize()
crs = {'init': 'epsg:32628'}
poly = gpd.GeoSeries(box(578150.0, 1572350.0, 578350.0, 1572550.0), crs=crs)
poly_json = eval(poly.to_json())
gj_geom = poly_json['features'][0]['geometry'] 
geom = ee.Geometry(gj_geom, poly.crs['init'][5:])
collection = (ee.ImageCollection('COPERNICUS/S1_GRD')
 .filter(ee.Filter.date('2019-01-01', '2019-02-1'))
 .filter(ee.Filter.geometry(geom))
 )

How do I obtain the raw data arrays from the filtered collection?

• Vincent: Is this the duplicated question: gis.stackexchange.com/questions/309195/…?
• dzang: @Vincent indeed that answers my question, thanks for pointing to it

Answer 1

You can use ee.Image.sampleRectangle() (see this post) for converting GEE images to 2d array. However, there is a limit of 262144 pixels that can be transferred to protect your system from becoming unresponsive.

So for your collection, you can map over a function that applies this sampleRectangle and retrieves the respective array.

Answer 2

It's possible to export google earth engine image to local files using geemap library.

import geemap
import numpy as np
band_names = image.bandNames().getInfo() # List with band names
selected_band = image.select(band_name) # selected band to trnasform
image_array = geemap.ee_to_numpy(selected_band, region=geometry) # Client side
ndvi_array = np.squeeze(ndvi_array) # Deleting single-dimensional axis