aisdb.network_graph module

collect vessel transits between domain zones (graph nodes), and aggregate trajectory statistics within the overall domain

aisdb.network_graph.graph(qry, *, outputfile, domain, dbconn: ~aisdb.database.dbconn.ConnectionType, data_dir: str, trafficDBpath: str, maxdelta: ~datetime.timedelta = datetime.timedelta(days=7), speed_threshold: float = 50, distance_threshold: float = 200000, interp_delta: float = datetime.timedelta(seconds=600), minscore: float = 0, qryfcn=<function crawl_dynamic_static>, decimate: float = 0.0001, verbose: bool = False)

Compute network graph of vessel movements within domain zones. Zone polygons will be used as network nodes, with graph edges represented by movements between zones.

Parameters:

• qry (aisdb.database.dbqry.DBQuery) – database query generator

• domain (aisdb.gis.Domain) – collection of zones defined as polygons, these will be used as nodes in the network graph

• dbconn (ConnectionType) – Either a aisdb.database.dbconn.SQLiteDBConn or aisdb.database.dbconn.PostgresDBConn database connection objects

• data_dir (string) – location of raster data

• trafficDBpath (string) – path to marinetraffic database file

• outpufile (string) – filepath for resulting CSV output

• maxdelta (datetime.timedelta) – maximum time between vessel messages before considering it to be part of a new trajectory. See aisdb.track_gen.split_timedelta() for more info

• speed_threshold (int, float) – maximum speed in knots for encoder segmentation. See aisdb.denoising_encoder.encode_greatcircledistance() for more info

• distance_threshold (int, float) – maximum distance in meters for encoder segmentation. See aisdb.denoising_encoder.encode_greatcircledistance() for more info

• interp_delta (timedelta) – track positions will be interpolated to the given sample rate

• minscore (float) – minimum score for segments to be considered sequential. See aisdb.denoising_encoder.encode_greatcircledistance() for more info

Network graph activity is computed following these steps:

• Create database query with aisdb.database.dbqry.DBQuery.gen_qry(), and supply resulting generator as rowgen arg. Define a domain (aisdb.gis.Domain) in which to compute movements

• Vectorize tracks using aisdb.track_gen.TrackGen()

• Append vessel metadata to track vessels with aisdb.webdata.marinetraffic.vessel_info()

• Segment track vectors where time between messages exceeds maxdelta using aisdb.track_gen.split_timedelta()

• Segment track vectors as encoded by aisdb.denoising_encoder.encode_greatcircledistance()

• Perform geofencing on track segments using aisdb.track_gen.fence_tracks() to determine zone containment

• Check where zone boundaries are transited and serialize results to outputfile. Additional metrics per zone activity is also aggregated at this step.

Example usage:

>>> import os
>>> import shapely
>>> from datetime import datetime
>>> from aisdb import SQLiteDBConn, DBQuery, Domain, graph, decode_msgs
>>> from aisdb.database.sqlfcn_callbacks import in_bbox_time

>>> # create example database file
>>> dbpath = './example.sqlitedb'
>>> filepaths = ['./aisdb/tests/testdata/test_data_20210701.csv',
...              './aisdb/tests/testdata/test_data_20211101.nm4']
>>> with SQLiteDBConn(dbpath) as dbconn:
...     decode_msgs(filepaths=filepaths, dbconn=dbconn,
...     source='TESTING', verbose=False)

Next, configure query area using Domain to compute region boundary

>>> zones = [{
...     'name': 'Zone1',
...     'geometry': shapely.geometry.Polygon(zip(
...             [-170.24, -170.24, -38.5, -38.5, -170.24],
...             [29.0, 75.2, 75.2, 29.0, 29.0],
...         ))
...     }]
>>> domain = Domain(name='new_domain', zones=zones)
>>> trafficDBpath = './testdata/marinetraffic_test.db'
>>> data_dir = os.environ.get('AISDBDATADIR', '/tmp/ais/')

Then, query db for points in domain

>>> with SQLiteDBConn(dbpath) as dbconn:
...     qry = DBQuery(
...             dbconn=dbconn,
...             callback=in_bbox_time,
...             start=datetime(2021, 7, 1),
...             end=datetime(2021, 7, 3),
...             **domain.boundary,
...         )
...     graph(qry,
...           outputfile=os.path.join('testdata', 'test_graph.csv'),
...           dbconn=dbconn,
...           domain=domain,
...           data_dir=data_dir,
...           trafficDBpath=trafficDBpath)

Afterwards, delete the example database file

>>> os.remove(dbpath)
>>> os.remove(os.path.join('testdata', 'test_graph.csv'))

process the vessel movement graph edges. caution: this may consume a large amount of memory