More convenient than uploading your geospatial files one by one.
This is a small tutorial that explains how to upload your own geospatial data to the Google Earth Engine. I assume you have installed the python API as explained here. The upload tool comes with the GEE asset manager which you can download from github.
Step 1: Make sure Git is install on your machine.
sudo apt-get install git
Step 2: Clone the GEE asset manager to your machine
git clone https://github.com/tracek/gee_asset_manager.git
Step 3: Go into the gee_asset_manager folder and install the application
cd gee_asset_manager sudo python setup.py install
The requirements.txt shows the follow dependencies:
earthengine_api >= 0.1.87
requests >= 2.10.0
retrying >= 1.3.3
bs4 >= 4.5.1
requests_toolbelt >= 0.7.0
Step 4: Install the dependencies.
sudo pip install bs4
Step 5: Test the application by running
I got this error (Linux mint):
File "build/bdist.linux-x86_64/egg/retrying.py", line 47, in wrap AttributeError: 'module' object has no attribute 'wraps'
So downgraded retrying to make the tool work
sudo -H pip install retrying==1.2.3
Step 6: The line below gives you all options to upload all files from a directory:
geebam upload -h usage: geebam upload [-h] -u USER -d DIRECTORY [-m METADATA] [-c COLLECTION] [-p PATH] [--large] [--nodata NODATA] optional arguments: -h, --help show this help message and exit Required named arguments.: -u USER, --user USER Google account name (gmail address). -d DIRECTORY, --directory DIRECTORY Path to the directory with images. Optional named arguments: -m METADATA, --metadata METADATA Path to CSV with metadata. -c COLLECTION, --collection COLLECTION Name with path of the collection to create. If not provided, directory name will be used. It assumes the upload goes to the user folder. Need upload to a shared directory? Use --path instead. They are mutuall exclusive -p PATH, --path PATH Absolute upload path. It does not take any assumptions about user folder, so it can be used to upload to a shared folder. Mutually exclusive with --collection. --large (Advanced) Use multipart upload. Might help if upload of large files is failing on some systems. Might cause other issues. --nodata NODATA The value to burn into the raster as NoData (missing data)
geebam upload -u firstname.lastname@example.org -d /path/to/my/files/ -c mycollectionName --nodata -9999 2016-11-23 09:23:08,175 :: oauth2client.client :: INFO :: Attempting refresh to obtain initial access_token 2016-11-23 09:23:08,175 :: oauth2client.client :: INFO :: Refreshing access_token Password: 2016-11-23 09:23:16,430 :: urllib3.connectionpool :: INFO :: Starting new HTTPS connection (1): accounts.google.com 2016-11-23 09:23:17,488 :: urllib3.connectionpool :: INFO :: Starting new HTTPS connection (1): accounts.google.com.vn 2016-11-23 09:23:17,912 :: urllib3.connectionpool :: INFO :: Starting new HTTPS connection (1): myaccount.google.com 2016-11-23 09:23:23,871 :: root :: INFO :: New collection users/myname/mycollectionName created 2016-11-23 09:23:27,535 :: root :: INFO :: Processing image 1 out of 132: /path/to/my/files/file1.tif 2016-11-23 09:23:27,536 :: urllib3.connectionpool :: INFO :: Starting new HTTPS connection (1): ee-api.appspot.com 2016-11-23 09:23:28,063 :: urllib3.connectionpool :: INFO :: Starting new HTTPS connection (1): www.google.com 2016-11-23 09:23:28,419 :: urllib3.connectionpool :: INFO :: Starting new HTTPS connection (1): appengine.google.com 2016-11-23 09:23:49,524 :: root :: INFO :: Processing image 2 out of 132: /path/to/my/files/file2.tif ...
Note on parsing metadata:
### Parsing metadata
By metadata we understand here the properties associated with each image.
Thanks to these, GEE user can easily filter collection based on specified criteria. The file with metadata should be organised as follows:
| filename (without extension) | property1 header | property2 header |
| file1 | value1 | value2 |
| file2 | value3 | value4 |