loicdtx / ministac

Minimalist Spatio-Temporal Asset Catalog engine

Ministac

Ministac is a minimalist Spatio-Temporal Asset Catalog. It allows for instance an institution to have a searchable catalog of their in house satellite images

Features

• Manages metadata satellite images (items) and satellites images collections
• Stores information in Postgres database
• Search catalog using geojson geometry, temporal window, and/or cloud cover
• Complies with Spatio-Temporal Asset Catalog specifications
• Metadata validation against JSON schemas of stac-spec

• Built to work with:

  • ministac-admin (helper scripts to generate the catalog from multiple collections)
  • ministac-api (JSON REST API)
  • ministac-client (does not exist yet)

Installation

A complete installation requires the package, a postgres database with PostGIS extension, and a configuration file with database credentials.

Python package

Activate a python 3 virtual environment and run:

pip install ministac

Database

• Install postgresql and PostGIS
• Create the ministac database
• Enable PostGIS extension for the ministac database

createdb ministac
psql ministac -c "CREATE EXTENSION postgis;"
psql ministac -c "CREATE EXTENSION postgis_topology;"
# Optionally create dedicated user (probably better for remote access)
psql ministac -c "CREATE USER ministac_user WITH PASSWORD 'qwerty' CREATEDB;"

Configuration file

The configuration file must be named ~/.ministac; it contains the database access credentials.

[ministac]
db_database=name_of_the_database
# Optional, depending on database configuration (running on localhost or remote, require password, etc)
# db_hostname=
# db_username=
# db_password=
# db_port=

Usage

Although ministac has a functional API, most common use cases are probably covered by ministac-admin and ministac-api. Nethertheless here are some basic examples of direct use.

import json
from pprint import pprint
import datetime as dt

import ministac
from ministac.db import init_db, session_scope


# Create database tables
init_db()

# Open example collection from test directory of the ministac repository
with open('tests/data/collection_0.json') as src:
    landsat_8_collection = json.load(src)

# Register the Landsat 8 collection to the database
with session_scope() as session:
    ministac.add_collection(session, landsat_8_collection)

# Read some example items
with open('tests/data/item_list.json') as src:
    item_list = json.load(src)

# Ingest the items to the database
with session_scope() as session:
    ministac.add_items(session, item_list, 'landsat_sr_8')


# Query the entire landsat_sr_8 collection
with session_scope() as session:
        pprint(ministac.search(session, 'landsat_sr_8'))

# Add temporal filter
startDate = dt.datetime(2017, 12, 1)
with session_scope() as session:
    pprint(ministac.search(session, 'landsat_sr_8', startDate=startDate))

# Spatial filter
geom = {'coordinates': [[[-101.7, 19.59],
                         [-101.66, 19.54],
                         [-101.61, 19.56],
                         [-101.64, 19.58],
                         [-101.57, 19.63],
                         [-101.54, 19.66],
                         [-101.6, 19.68],
                         [-101.64, 19.64],
                         [-101.7, 19.59]]],
        'type': 'Polygon'}

with session_scope() as session:
    pprint(ministac.search(session, 'landsat_sr_8', geom=geom))

# Filter with cloud cover threshold
with session_scope() as session:
    pprint(ministac.search(session, 'landsat_sr_8', maxCloudCover=20))

Ackowledgements

This project received funding from CONABIO (Mexico's national commission for biodiversity research).

ministac's Issues

small error in usage

# Query the entire landsat_sr_8 collection
with session_scope() as session:
    pprint(ministac.search(session,collection='landsat_sr_8'))

instead of:

# Query the entire landsat_sr_8 collection
pprint(ministac.search('landsat_sr_8'))

in:

https://github.com/loicdtx/ministac#usage

Add function to clean a collection

A function that checks if every item in a collection is still where it's supposed to be, and optionally remove it from the database if it isn't