Usage examples for spatiotemporal trajectories-Lindorm(Lindorm)-阿里云帮助中心

Lindorm GanosBase provides a Java API compatible with GeoMesa for storing and querying spatiotemporal trajectory data. This tutorial walks through three tasks using a trajectory line segmentation model:

Create a trajectory table — configure a point model, build a line model, add a secondary index, and call createSchema.
Write trajectory data — parse raw points, apply a segmentation policy, and batch-write trajectory segments.
Query trajectories — build Common Query Language (CQL) filters and retrieve either matching segments or complete trajectories.

Prerequisites

Before you begin, ensure that you have:

A Lindorm GanosBase instance
A DataStore connection — see Getting started for setup instructions

Key concepts

Trajectory point model — Each trajectory point is a spatiotemporal object with attributes. Define the point model first; the line model is built on top of it. In GeoTools terms, SimpleFeatureType defines the schema (similar to a table definition in a relational database), and SimpleFeature is a single data record (equivalent to a row).

Trajectory line segmentation model — A derived model that groups trajectory points into segments. TrajectoryFeatureType includes default attribute fields that cannot be removed; add extra fields with addTrajProperties.

Create a table

The steps below create a trajectory table with a primary index on id and a secondary spatiotemporal index on xz2t.

Step 1: Define the trajectory point model.

String pointSpec = "id:String,dtg:Date,*geom:Point:srid=4326";
SimpleFeatureType pointSFT = SimpleFeatureTypes.createType("point", pointSpec);

Each trajectory point has an ID, a timestamp (dtg), and a geometry (geom). The asterisk (*) marks geom as the default geometry column.

Step 2: Build the trajectory line model.

TrajectoryBuilder trajBuilder = new TrajectoryBuilder("trajectory", pointSFT, "dtg", "id", true);
TrajectoryFeatureType trajFeatureType = trajBuilder.getTrajFeatureType();

TrajectoryBuilder takes the point model (pointSFT), the timestamp field (dtg), and the ID field (id). To add custom attribute fields to each trajectory segment, call addTrajProperties before calling getTrajFeatureType:

TrajectoryBuilder trajBuilder = new TrajectoryBuilder("trajectory", pointSFT, "dtg", "id", true);
trajBuilder.addTrajProperties("speed:Double,traj_end_time:Date");
TrajectoryFeatureType trajFeatureType = trajBuilder.getTrajFeatureType();

For the full list of default fields, see Terms.

Step 3: Configure the secondary index.

trajFeatureType.getUserData().put("geomesa.secondary.enabled", "true");
trajFeatureType.getUserData().put("geomesa.primary.index", "id");
trajFeatureType.getUserData().put("geomesa.indices.enabled", "id,xz2t");

The following table describes the index configuration keys:

Key	Value in this example	Effect
`geomesa.secondary.enabled`	`true`	Enables secondary index creation
`geomesa.primary.index`	`id`	Sets `id` as the index key column of the primary table
`geomesa.indices.enabled`	`id,xz2t`	Creates an `id` index and an `xz2t` spatiotemporal index

The following table describes the supported index types:

Index	Algorithm	Best for
`id`	Feature ID as primary key	Point-in-time lookups by object ID
`xz2t`	Three-dimensional space-filling curve (latitude, longitude, time) for non-point geometries	Spatiotemporal range queries on trajectory segments

Step 4: Create the table.

ds.createSchema(trajFeatureType);

This creates both the primary table and the secondary index table. The secondary index table is managed internally by Lindorm and is not visible to users.

Write trajectory data

Step 1: Parse trajectory points from raw data.

public SimpleFeature createFeature(SimpleFeatureType sft, String id, Object... values) {
    SimpleFeatureBuilder sfBuilder = new SimpleFeatureBuilder(sft);
    SimpleFeature simpleFeature = sfBuilder.buildFeature(id, values);
    return simpleFeature;
}

List<SimpleFeature> points = new ArrayList<>();
for (String pointData : rawData) {
    // Parse id and point values from pointData
    String pointID = ...;
    Object[] values = ...;

    points.add(createFeature(pointSFT, pointID, values));
}

Step 2: Define a segmentation policy.

SplitConfigs splitConfigs = new SplitConfigs(SplitConfigs.BY_COUNT(), new Object[]{50});

This policy groups every 50 trajectory points into one trajectory segment.

Step 3: Create trajectory segments.

String objectID = "001";
List<SimpleFeature> trajFeatures = trajBuilder.createTrajSegments(objectID, features, splitConfigs);

Important

If you called addTrajProperties when building the model, pass an extraTrajProperties array to createTrajSegments with values for every added field.

Object[] extraTrajProperties = new Object[]{speed, trajEndTime};
List<SimpleFeature> trajFeatures = trajBuilder.createTrajSegments(objectID, features, splitConfigs, extraTrajProperties);

Step 4: Batch-write trajectory segments.

SimpleFeatureStore featureStore = (SimpleFeatureStore) ds.getFeatureSource(trajFeatureType.getTypeName());
featureStore.addFeatures(new ListFeatureCollection(trajFeatureType, trajFeatures));

Query trajectories

Lindorm GanosBase accepts CQL predicates compatible with GeoMesa. Build a filter with ECQL.toFilter, wrap it in a Query, and call the appropriate method on LindormDataStore. For the full CQL syntax reference, see GanosQuery spatiotemporal objects.

The following table summarizes the supported query types:

Query type	Example predicate
ID and point in time	`object_id in ('001') and seg_start_time > 1549522511`
Attribute query	`speed > 2.0` — for a trajectory point model, queries point attributes; for a trajectory line model, queries trajectory-level and segment-level attributes only
Spatial range query	`INTERSECTS(geom, POLYGON((-78 26,-73 26,-73 31,-78 31,-78 26)))`
Spatiotemporal range query	`INTERSECTS(geom, POLYGON((-78 26,-73 26,-73 31,-78 31,-78 26))) AND traj_start_time >= 1549510271 AND traj_start_time <= 1549582040`

ID and point-in-time query

Use getFeatures with an ID and timestamp predicate to retrieve matching segments:

String ecqlPredicate = "object_id in ('001') and seg_start_time > 1549522511";
Query query = new Query("trajectory", ECQL.toFilter(ecqlPredicate));

SimpleFeatureCollection featureCollection =
    ((LindormDataStore) ds).getFeatureSource("trajectory").getFeatures(query);

SimpleFeatureIterator iterator = featureCollection.features();
while (iterator.hasNext()) {
    SimpleFeature feature = iterator.next();
    // Process each matching segment
}

Attribute query

Use getFeatures with an attribute predicate. For a trajectory point model, this queries point attributes. For a trajectory line model, this queries trajectory-level and segment-level attributes only.

String ecqlPredicate = "speed > 2.0";
Query query = new Query("trajectory", ECQL.toFilter(ecqlPredicate));

SimpleFeatureCollection featureCollection =
    ((LindormDataStore) ds).getFeatureSource("trajectory").getFeatures(query);

SimpleFeatureIterator iterator = featureCollection.features();
while (iterator.hasNext()) {
    SimpleFeature feature = iterator.next();
    // Process each matching segment
}

Spatial range query

Use getFeatures with an INTERSECTS predicate to retrieve segments whose geometry intersects the specified polygon:

String ecqlPredicate = "INTERSECTS(geom, POLYGON((-78 26,-73 26,-73 31,-78 31,-78 26)))";
Query query = new Query("trajectory", ECQL.toFilter(ecqlPredicate));

SimpleFeatureCollection featureCollection =
    ((LindormDataStore) ds).getFeatureSource("trajectory").getFeatures(query);

SimpleFeatureIterator iterator = featureCollection.features();
while (iterator.hasNext()) {
    SimpleFeature feature = iterator.next();
    // Process each matching segment
}

Spatiotemporal range query

Combine a spatial predicate with a time range to filter by both location and time window.

Return matching segments only — use getFeatures to retrieve only the segments that match the filter:

String ecqlPredicate = "INTERSECTS(geom, POLYGON((-78 26,-73 26,-73 31,-78 31,-78 26)))" +
    " AND traj_start_time >= 1549510271 AND traj_start_time <= 1549582040";
Query query = new Query("trajectory", ECQL.toFilter(ecqlPredicate));

SimpleFeatureCollection featureCollection =
    ((LindormDataStore) ds).getFeatureSource("trajectory").getFeatures(query);

SimpleFeatureIterator iterator = featureCollection.features();
while (iterator.hasNext()) {
    SimpleFeature feature = iterator.next();
    // Process each matching segment
}

Return complete trajectories — use getWholeTrajectoies to reconstruct and return entire trajectories that contain matching segments:

String ecqlPredicate = "INTERSECTS(geom, POLYGON((-78 26,-73 26,-73 31,-78 31,-78 26)))" +
    " AND traj_start_time >= 1549510271 AND traj_start_time <= 1549582040";
Query query = new Query("trajectory", ECQL.toFilter(ecqlPredicate));

List<TrajectoryFeature> trajectories =
    ((LindormDataStore) ds).getFeatureSource("trajectory").getWholeTrajectoies(query);

What's next

Terms — Definitions of trajectory storage model fields and supported index types
Query spatiotemporal objects — Full CQL syntax and advanced query options