A python client for sat-api
Clone or download
Latest commit bb41641 Aug 28, 2018
Failed to load latest commit information.
.circleci ci: update config to deploy on master Aug 21, 2018
test fix failing test Aug 28, 2018
setup.py set python as setup interpreter Aug 25, 2017




Sat-search is a Python 2/3 library and a command line tool for discovering and downloading publicly available satellite imagery using a conformant API such as sat-api.

The legacy version of sat-search (<1.0.0) can be used with the legacy version of sat-api (<1.0.0), currently deployed at https://api.developmentseed.org/satellites.


It is recommended to use pyenv and virtualenv to to control Python versions and installed dependencies. sat-search can be conveniently installed from PyPi:

# install the latest release version
$ pip install satsearch

Sat-search is a very lightweight application, with the only dependency being requests.

Using sat-search

Sat-search has several features:

  • search catalog
  • STAC compliant interface
  • save results of a search
  • load results of a search
  • download assets (e.g. thumbnails, data files) of the results


The sat-search CLI has an extensive online help that can be printed with the -h switch.

$ sat-search -h
usage: sat-search [-h] {search,load} ...

sat-search (v1.0.0b8)

positional arguments:
    search       Perform new search of scenes
    load         Load scenes from previous search

optional arguments:
  -h, --help     show this help message and exit

As can be seen there are two subcommands, each of which has it's own online help (i.e. "sat-search search -h" and "sat-search load -h") and will be discussed in detail below.


$ sat-search search -h
usage: sat-search search [-h] [--version] [-v VERBOSITY]
                         [--print_md [PRINT_MD [PRINT_MD ...]]] [--print_cal]
                         [--save SAVE] [--append] [-c [C:ID [C:ID ...]]]
                         [--intersects INTERSECTS] [--datetime DATETIME]
                         [--eo:cloud_cover EO:CLOUD_COVER]
                         [-p [PARAM [PARAM ...]]] [--url URL]

optional arguments:
  -h, --help            show this help message and exit
  --version             Print version and exit
  -v VERBOSITY, --verbosity VERBOSITY
                        0:quiet, 1:error, 2:warning, 3:info, 4:debug (default:

output options:
  --print_md [PRINT_MD [PRINT_MD ...]]
                        Print specified metadata for matched scenes (default:
  --print_cal           Print calendar showing dates (default: False)
  --save SAVE           Save results as GeoJSON (default: None)
  --append              Append scenes to GeoJSON file (specified by save)
                        (default: False)

search options:
  -c [C:ID [C:ID ...]], --c:id [C:ID [C:ID ...]]
                        Name(s) of collection (default: None)
  --intersects INTERSECTS
                        GeoJSON Feature (file or string) (default: None)
  --datetime DATETIME   Single date/time or begin and end date/time (e.g.,
                        2017-01-01/2017-02-15 (default: None)
  --eo:cloud_cover EO:CLOUD_COVER
                        Range of acceptable cloud cover (e.g., 0/20) (default:
  -p [PARAM [PARAM ...]], --param [PARAM [PARAM ...]]
                        Additional parameters of form KEY=VALUE (default:
  --url URL             URL of the API (default: https://sat-

Search options

  • c:id - A list of names of collections (i.e. sensors). The collections supported depend on the API, and for sat-api can be seen at the collections endpoint. If one or more collections are not defined, all collections are searched.
  • intersects - Provide a GeoJSON Feature string or the name of a GeoJSON file containing a single Feature that is a Polygon of an AOI to be searched.
  • datetime - Provide a single partial or full datetime (e.g., 2017, 2017-10, 2017-10-11, 2017-10-11T12:00), or two seperated by a slash that defines a range. e.g., 2017-01-01/2017-06-30 will search for scenes acquired in the first 6 months of 2017.
  • eo:cloud_cover - Provide a single percent cloud cover to match (e.g., 0) or two numbers separated by a slash indicating the range of acceptable cloud cover (e.g., 0/20 searches for scenes with 0% - 20% cloud cover).
  • param - Allows searching for any other scene properties by providing the pair as KEY=VALUE (e.g. -p landsat:row=42)
  • url - The URL endpoint of a STAC compliant API, this can also be set with the environment variable SATUTILS_API_URL

Output options These options control what to do with the search results, multiple switches can be provided.

  • print_md - Prints a list of specific metadata fields for all the scenes. If given without any arguments it will print a list of the dates and scene IDs. Otherwise it will print a list of fields that are provided. (e.g., --print_md date eo:cloud_cover eo:platform will print a list of date, cloud cover, and the satellite platform such as WORLDVIEW03)
  • print_cal - Prints a text calendar with specific days colored depending on the platform of the scene (e.g. landsat-8), along with a legend.
  • save - Saves results as a FeatureCollection. The FeatureCollection 'properties' contains all of the arguments used in the search and the 'features' contain all of the individual scenes, with individual scene metadata merged with collection level metadata (metadata fields that are the same across all one collection, such as eo:platform)
  • append - The save option will always create a new file, even overwriting an existing one. If append is provided then the scenes will be appended to the FeatureCollection given by the save filename.


Scenes that were previously saved with sat-search search --save ... can be loaded with the load subcommand.

$ sat-search load -h
usage: sat-search load [-h] [--version] [-v VERBOSITY]
                       [--print_md [PRINT_MD [PRINT_MD ...]]] [--print_cal]
                       [--save SAVE] [--append] [--datadir DATADIR]
                       [--filename FILENAME]
                       [--download [DOWNLOAD [DOWNLOAD ...]]]

positional arguments:
  scenes                GeoJSON file of scenes

optional arguments:
  -h, --help            show this help message and exit
  --version             Print version and exit
  -v VERBOSITY, --verbosity VERBOSITY
                        0:quiet, 1:error, 2:warning, 3:info, 4:debug (default:

output options:
  --print_md [PRINT_MD [PRINT_MD ...]]
                        Print specified metadata for matched scenes (default:
  --print_cal           Print calendar showing dates (default: False)
  --save SAVE           Save results as GeoJSON (default: None)
  --append              Append scenes to GeoJSON file (specified by save)
                        (default: False)

download options:
  --datadir DATADIR     Directory pattern to save assets (default:
  --filename FILENAME   Save assets with this filename pattern based on
                        metadata keys (default: ${id})
  --download [DOWNLOAD [DOWNLOAD ...]]
                        Download assets (default: None)

Note that while the search options are gone, output options are still available and can be used with the search results loaded from the file. There is also a new series of options now, for downloading data.

Downloading assets

When loading results from a file, the user now has the option to download assets from the scenes.

Download options These control the downloading of assets. Both datadir and filename can include metadata patterns that will be substituted per scene.

  • datadir - This specifies where downloaded assets will be saved to. It can also be specified by setting the environment variable SATUTILS_DATADIR.
  • filename - The name of the file to save. It can also be set by setting the environment variable SATUTILS_FILENAME
  • download - Provide a list of keys to download these assets. For DG currently only thumbnail and full are supported. More information on downloading data is provided below.

Metadata patterns Metadata patterns can be within datadir and filename in order to have custom path and filenames based on the scene metadata. For instance specifying datadir as "./${eo:platform}/${date}" will save assets for each scene under directories of the platform and the date. So a WorldView-3 scene from June 20, 2018 will have it's assets saved in a directory './WORLDVIEW03/2017-06-20'. For filenames these work exactly the same way, except the appropriate extension will be used at the end of the filename, depending on the asset.

Assets The thumbnail for each scene in a scenes.json file can be downloaded with

    sat-search load scenes.json --download assetname1 assetname2 ...

The thumbnails will be saved using a directory and filename according to the datadir and filename options, and will also have a '_thumbnail` suffix. When thumbnails are downloaded an ESRI Worldfile (.wld) file is created, which is a sidecar file that describes the coordinates and resolution of the images. This enables the thumbnails to be viewed in a GIS program like QGIS in their proper geographical location. The world file does not set the spatial reference system used (lat/lon, or WGS-84, or EPSG:4326), so when opened in QGIS it will need to be selected (EPSG:4326).


The sat-search library is made up of several Python classes. The Scene class represents a single set of images for an indentical date (or daterange) and footprint. The Scenes class is a collection of Scene objects that makes it easier to iterate through them and perform common tasks over all the scenes, such as downloading data.

The Query class is a single set of arguments for searching scenes, functions for querying the API with those arguments (and handling of multiple pages if needed) as well storing the results. The higher level Search class which is more often used, can deal with multiple Query objects, such as individual Scene ids or disparate date ranges that must be issued to the API with different arguments.


sat-search was created by Development Seed and is part of a collection of tools called sat-utils.