The earthengine
tool is a utility program that lets you manage Earth Engine
assets and tasks from the command line. It is installed automatically when you
install the Python API. To check whether the tool is installed
and functioning correctly, type the following on a command line:
earthengine
If the tool is properly installed, it prints out a short summary of available commands. To get help on a specific command, use:
earthengine command -h
When you first install the Python API you need to sign in using the
authenticate
command described below. The following sections describe the
available commands in more detail.
Service account credentials
To use the CLI with a service account's credentials, use the
service_account_file
flag to point to a JSON file containing the service
account's key.
earthengine --service_account_file=service_account_creds.json
Usage in Colab
The Earth Engine Command Line Tool is preinstalled and ready for use in Google Colab.
Authentication
Authenticate for each new Colab session or if the virtual machine expires from inactivity (credentials are not saved across sessions).
Import the Python client library and call ee.Authenticate()
to trigger the
authentication flow. Follow the prompts to complete authentication. The default
auth_mode
in Colab is colab
, see the
Authentication guide
for other options.
import ee
ee.Authenticate()
Command execution
To run command line utilities, like the Earth Engine CLI, you need to prepend command calls with an exclamation point.
!earthengine -h
Set a Cloud project
Use the --project
option to set a Cloud project for each individual
earthengine
command.
!earthengine --project my-project <command>
Alternatively, set a default project to be used by all earthengine
calls using the set_project
command. The project will be added
to a credentials file (~/.config/earthengine/credentials) and used for
subsequent commands, unless overridden by the --project
!earthengine set_project my-project
Command Reference
authenticate
Authenticates the command line tool and Python client library to Earth Engine. Example:
earthengine authenticate
Earth Engine uses the OAuth 2.0 protocol for authenticating clients. The earthengine authenticate command will prompt you through the authentication process using your web browser.
You will need to install gcloud if
using the default gcloud authentication mode. See other authentication modes
available through the auth_mode
parameter in the authentication
guide.
acl
Prints or updates the access control list (ACL) of an Earth Engine asset. The ACL controls who can read or write to an asset. Examples:
earthengine acl get projects/my-project/assets/asset_id
earthengine acl set public projects/my-project/assets/asset_id
earthengine acl ch -u username@gmail.com:R projects/my-project/assets/asset_id
The get
sub-command prints a JSON description of the ACL. The set
sub-command sets an ACL provided in a file with the same JSON format. You can
copy an ACL from one asset to others by saving the output from get
and
providing it to set
.
The set sub-command also accepts two special ACL names:
private
: Removes permissions from everyone except the owner.public
: Grants read permission to all users.
The ch
sub-command lets you make individual changes to an ACL. To grant read
permission specify -u username@gmail.com:R
, to grant write permission specify
-u username@gmail.com:W
, and to remove a user's permissions specify -d
username@gmail.com
. The special user identifier AllUsers
may be used to grant
or revoke read permission to or from all users. (Note that revoking AllUsers
permissions does not revoke any additional permissions you may have also
granted to individual users.)
asset
Prints or updates metadata associated with the an Earth Engine asset. Examples:
earthengine asset info projects/my-project/assets/asset_id
earthengine asset set -p name=value projects/my-project/assets/asset_id
The info
sub-command prints detailed information about the asset, including
its metadata, in JSON form. The set
sub-command sets individual metadata
properties on an asset.
The values of metadata properties that you set may be either numbers or strings.
When setting property names using the --property
or -p
flag, separate the
property name and value with an equals sign. The data type is detected
automatically, or you may specify it explicitly by prefixing the property name
with (string)
, (number)
, or (date)
. For example, this sets a string-valued
property with the value "42"
:
earthengine asset set -p '(string)name=42' projects/my-project/assets/asset_id
(The quotes in this example prevent the shell from interpreting the parentheses. They may or may not be necessary, depending on your shell and platform.)
Date properties are just numbers that represent a number of milliseconds since the Unix epoch (i.e. midnight on January 1st, 1970) and may be specified directly as a number or in one of the following formats:
YYYY-MM-DD
YYYY-MM-DDThh:mm:ss
YYYY-MM-DDThh:mm:ss.f
The time zone is assumed to be UTC. You may set the special start and end time
properties using the --time_start
and --time_end
flags:
earthengine asset set --time_start 1978-10-15T12:34:56 projects/my-project/assets/asset_id
cp
Copies an asset. Example:
earthengine cp projects/my-project/assets/asset_id projects/my-project/assets/new_asset_id
create
Creates new folders and image collections. Example:
earthengine create folder projects/my-project/assets/folder_id
earthengine create collection projects/my-project/assets/collection_id
Use the folder
sub-command to create folders and the collection
to create
image collections. You can specify the -p
option to recursively create parent
folders as required. Newly-created folders and images have private ACLs by
default.
ls
Lists the contents of one or more folders or collections. Example:
earthengine ls users/username
The -l
option requests a long format with more information about each asset
(currently just its type). You may specify --max_items number
(or -m
for
short) to limit the number of items from each folder or collection you list:
earthengine ls -m 10 projects/my-project/assets/my_large_collection
Running the ls
command with no arguments will list the top-level folders that
you own.
model
Tool with which to manipulate TensorFlow saved models.
model prepare
Prepare a saved model for serving in Earth Engine. Specifically, this transforms
your SavedModel
into a form suitable for processing requests from Earth
Engine. (Learn more about
SavedModel
.)
earthengine model prepare my_source_dir my_dest_dir '{"Conv2D:0":"my_input_band"}' '{"Sigmoid:0":"my_output_band"}'
Learn more about AI Platform models here. See a complete example here.
mv
Moves or renames an asset. Example:
earthengine mv projects/my-project/assets/asset_id projects/my-project/assets/new_asset_id
project_config
Sets project configuration values. Before using this command, be sure to set a
project using set_project
.
get
To view a project configuration, use the get
sub-command:
earthengine alpha project_config get
If you have permission to view the project's batch task settings, the output contains:
maxConcurrentExports
: a number indicating the maximum number of batch tasks that can run in parallel across all users for the given project. By default, this is set to the maximum allowed by the associated billing account's subscription plan.
Additionally, if you have permission to view the plan configuration, the output contains:
planMaxConcurrentExports
, a number indicating the maximum number of batch tasks that can run in parallel across all users and projects that use the billing account.
set
To update the project's configuration, use the set
sub-command. The following
settings can be configured:
max_concurrent_exports
, to control the project's maximum batch task parallelism- requires the correct permissions
For example, to configure a project to only ever allow 10 tasks to run in parallel for the given project:
earthengine alpha project_config set --max_concurrent_exports=10
The output displays the updated project configuration, identical to what
get
returns.
For more information about batch task parallelism, see the Earth Engine quotas page.
rm
Deletes one or more assets. Example:
earthengine rm projects/my-project/assets/asset_id
earthengine rm -r projects/my-project/assets/folder_id
You can use the -r
flag to delete the contents of a folder or collection
recursively. For safety when deleting multiple assets, you can use the
--dry_run
flag to verify exactly what will be deleted without actually
deleting anything.
set_project
Sets the Google Cloud project through which computation requests are routed.
earthengine set_project foo-project
This command is needed prior to running commands that require Cloud
functionality, for example model
.
task
Prints information about or manages long-running tasks. Examples:
earthengine task list
earthengine task list -l
earthengine task info TASK_ID
earthengine task cancel TASK_ID
The list
sub-command lists basic information about the tasks that you
submitted recently. The -l
option requests a long format with more information
about each task. The info
sub-command prints detailed information about
individual tasks. The cancel
sub-command cancels one or more running tasks.
upload
Uploads images or tables from Google Cloud Storage to Earth Engine, or creates assets backed by external images.
image
To upload an image asset using default settings:
earthengine upload image --asset_id=projects/my-project/assets/asset_id gs://bucket/image.tif
If you specify multiple input image files they will be interpreted as tiles of a single image asset. You can learn more about the options for uploading images to Earth Engine in Uploading image assets: Advanced options.
You can specify the pyramid reduction policy using the --pyramiding_policy
flag, which can be set to one of mean
(the default), sample
, mode
, min
,
or max
. This will control how Earth Engine generates the pyramid of
lower-resolution versions of your image:
earthengine upload image --asset_id=projects/my-project/assets/asset_id --pyramiding_policy=sample gs://bucket/image.tif
You can use the --last_band_alpha
to indicate that the mask for the image
should be taken from an alpha channel in the last band:
earthengine upload image --asset_id=projects/my-project/assets/asset_id --last_band_alpha gs://bucket/image.tif
You can specify a no-data value using the --nodata_value
flag. This will mask
any pixels in the image with that value:
earthengine upload image --asset_id=users/myuser/asset --nodata_value=255 gs://bucket/image.tif
You may also specify metadata properties to set on the asset using the same
flags that are accepted by the asset set
command described above.
The options are also described in the
Image Manifest
guide.
table
To upload a Shapefile, CSV, or TFRecord from Google Cloud Storage to an Earth Engine table asset, you can use any of:
earthengine upload table --asset_id=projects/my-project/assets/myUploadedShapefile gs://bucket/foo.shp
earthengine upload table --asset_id=projects/my-project/assets/myUploadedCSV gs://bucket/foo.csv
earthengine upload table --asset_id=projects/my-project/assets/myUploadedTFRecord gs://bucket/foo.tfrecord
There are many options pertaining to the way in which CSV and TFRecord files are interpreted. You can see a complete list of table upload options by visiting the Table Manifest guide or with:
earthengine upload table -h
external_image
To create an asset backed by an external image, run the upload_image
command
with a manifest:
earthengine alpha upload external_image --manifest /tmp/foo.json
An example manifest is:
{
"name": "projects/{project}/assets/cogdemo1",
"tilesets": [
{ "id": "0", "sources": [{ "uris": ["gs://ee-docs-demos/COG_demo.tif"] }] }
],
"properties": {
"source": "https://code.earthengine.google.com/d541cf8b268b2f9d8f834c255698201d"
},
"startTime": "2016-01-01T00:00:00.000000000Z",
"endTime": "2016-12-31T15:01:23.000000000Z"
}
See the Cloud GeoTIFF guide and the Image Manifest guide for more details on constructing the manifest.