12. GeoMesa Processes

The following analytic processes are available and optimized on GeoMesa data stores, found in the geomesa-process module:

• ArrowConversionProcess - encodes simple features in the Apache Arrow format
• BinConversionProcess - encodes simple features in a minimized 16-byte format
• DensityProcess - computes a density heatmap for a CQL query
• DateOffsetProcess - modifies the specified date field in a feature collection by an input time period.
• HashAttributeProcess/HashAttributeColorProcess - computes an additional ‘hash’ attribute which is useful for styling.
• JoinProcess - merges features from two different schemas using a common attribute field
• KNearestNeighborSearchProcess - performs a KNN search
• Point2PointProcess - aggregates a collection of points into a collection of line segments
• ProximitySearchProcess - searches near a set input features
• QueryProcess - performs a Geomesa query, useful as input for nested requests
• RouteSearchProcess - matches features traveling along a given route
• SamplingProcess - uses statistical sampling to reduces the features returned by a query
• StatsProcess - returns various stats for a CQL query
• TrackLabelProcess - selects the last feature in a track based on a common attribute, useful for styling
• TubeSelectProcess - performs a correlated search across time and space
• UniqueProcess - identifies unique values for an attribute

Where possible, the calculations are pushed out to a distributed system for faster performance. Currently this has been implemented in the Accumulo data store and partially in the HBase data store. Other back-ends can still be used, but local processing will be used.

12.1. Installation

While they can be used independently, the common use case is to use them with GeoServer. To deploy them in GeoServer requires:

1. a GeoMesa datastore plugin
2. the GeoServer WPS extension
3. the geomesa-process-wps_2.11-<version>.jar deployed in ${GEOSERVER_HOME}/WEB-INF/lib

Note

Some processes also require custom output formats, available separately in the GPL licensed GeoMesa GeoServer WFS module

The GeoMesa datastore plugin and GeoMesa process jars are both available in the binary distribution in the gs-plugins directory.

Documentation about the GeoServer WPS Extension (including download instructions) is available here.

To verify the install, start GeoServer, and you should see a line like INFO [geoserver.wps] - Found 15 bindable processes in GeoMesa Process Factory.

In the GeoServer web UI, click ‘Demos’ and then ‘WPS request builder’. From the request builder, under ‘Choose Process’, click on any of the ‘geomesa:’ options to build up example requests and in some cases see results.

12.2. Processors

12.2.1. ArrowConversionProcess

The ArrowConversionProcess converts an input feature collection to arrow format.

Parameters	Description
features	Input feature collection to encode.
includeFids	Include feature IDs in arrow file.
dictionaryFields	Attributes to dictionary encode.
useCachedDictionaries	Use cached top-k stats (if available), or run a dynamic stats query to build dictionaries.
sortField	Attribute to sort by.
sortReverse	Reverse the default sort order.
batchSize	Number of features to include in each record batch.
doublePass	Build dictionaries first, then query results in a separate scan.

12.2.2. BinConversionProcess

The BinConversionProcess converts an input feature collection to BIN format.

Parameters	Description
features	Input feature collection to query.
track	Track field to use for BIN records.
geom	Geometry field to use for BIN records.
dtg	Use cached top-k stats (if available), or run a dynamic stats query to build dictionaries.
label	Attribute to sort by.
axisOrder	Reverse the default sort order.

12.2.3. DensityProcess

The DensityProcess computes a density map over a set of features stored in GeoMesa. A raster image is returned.

Parameters	Description
data	Input Simple Feature Collection to run the density process over.
radiusPixels	Radius of the density kernel in pixels. Controls the “fuzziness” of the density map.
weightAttr	Name of the attribute to use for data point weights.
outputBBOX	Bounding box and CRS of the output raster.
outputWidth	Width of the output raster in pixels.
outputHeight	Height of the output raster in pixels.

12.2.4. DateOffsetProcess

The DateOffsetProcess modifies the specified date field in a feature collection by an input time period.

Parameters	Description
data	Input features.
dateField	The date attribute to modify.
timeOffset	Time offset (e.g. P1D).

12.2.5. HashAttributeProcess

The HashAttributeProcess adds an attribute to each SimpleFeature that hashes the configured attribute modulo the configured param.

Parameters	Description
data	Input Simple Feature Collection to run the hash process over.
attribute	The attribute to hash on.
modulo	The divisor.

12.2.5.1. Hash example (XML)

HashAttributeProcess_wps.xml is a geoserver WPS call to the GeoMesa HashAttributeProcess. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@HashAttributeProcess_wps.xml localhost:8080/geoserver/wps

The query should generate results that look like this:

{
    "id" : "d0971735-f8fe-47ed-a7cd-2e12280e8ac1",
    "geometry" : {
        "coordinates" : [
            151.1554,
            18.2014
        ],
        "type" : "Point"
    },
    "type" : "Feature",
    "properties" : {
        "Vitesse" : 614,
        "Heading" : 244,
        "Date" : "2016-05-02T18:00:44.030+0000",
        "hash" : 237,
        "CabId" : 150002,
     }
 }

12.2.6. HashAttributeColorProcess

The HashAttributeColorProcess adds an attribute to each SimpleFeature that hashes the configured attribute modulo the configured param and emit a color.

Parameters	Description
data	Input Simple Feature Collection to run the hash process over.
attribute	The attribute to hash on.
modulo	The divisor.

12.2.7. JoinProcess

The JoinProcess queries a feature type based on attributes from a second feature type.

Parameters	Description
primary	Primary feature collection being queried.
secondary	Secondary feature collection to be joined.
joinAttribute	Attribute field to join on.
joinFilter	Additional filter to apply to joined features.
attributes	Attributes to return. Attribute names should be qualified with the schema name, e.g. foo.bar.

12.2.8. KNearestNeighborSearchProcess

The KNearestNeighborSearchProcess performs a K-nearest-neighbor search on a feature collection using a second feature collection as input. It returns k neighbors for each point in the input data set. Note that if a feature is the nearest neighbor of multiple points in the input data set, it is returned only once.

Parameter	Description
inputFeatures	Input feature collection. The geometries of the features defines the KNN search.
dataFeatures	The data set to query for neighbors.
numDesired	k, number of nearest neighbors to return.
estimatedDistance	Estimate of the distance in meters for the k-th nearest neighbor, used for the initial query window.
maxSearchDistance	Maximum search distance in meters, used to prevent runaway queries of the entire data set.

12.2.8.1. K-Nearest-Neighbor Example (XML)

KNNProcess_wps.xml is a geoserver WPS call to the GeoMesa KNearestNeighborSearchProcess. In this example, it is chained with a Query process (see Chaining Processes), in order to avoid returning the query features as data. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@KNNProcess_wps.xml localhost:8080/geoserver/wps

12.2.9. Point2PointProcess

The Point2PointProcess aggregates a collection of points into a collection of line segments.

Parameters	Description
data	Input feature collection.
groupingField	Field on which to group.
sortField	Field on which to sort (must be Date type).
minimumNumberOfPoints	Minimum number of points.
breakOnDay	Break connections on day marks.
filterSingularPoints	Filter out segments that fall on the same point.

12.2.9.1. Point2Point example (XML)

Point2PointProcess_wps.xml is a geoserver WPS call to the GeoMesa Point2PointProcess. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@Point2PointProcess_wps.xml localhost:8080/geoserver/wps

The query should generate results that look like this:

{
    "id" : "367152240-4",
    "geometry" : {
        "coordinates" : [
            [
                -13.4041,
                37.8067
            ],
            [
                -13.4041,
                37.8068
            ]
        ],
        "type" : "LineString"
    },
    "type" : "Feature",
    "properties" : {
        "Date_end" : "2018-02-05T14:54:36.598+0000",
        "CabId" : 367152240,
        "Date_start" : "2018-02-05T14:53:58.078+0000"
    }
}

12.2.10. ProximitySearchProcess

The ProximitySearchProcess performs a proximity search on a Geomesa feature collection using another feature collection as input.

Parameters	Description
inputFeatures	Input feature collection that defines the proximity search.
dataFeatures	The data set to query for matching features.
bufferDistance	Buffer size in meters.

12.2.10.1. Proximity search example (XML)

ProximitySearchProcess_wps.xml is a geoserver WPS call to the GeoMesa ProximitySearchProcess. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@ProximitySearchProcess_wps.xml localhost:8080/geoserver/wps

12.2.11. RouteSearchProcess

The RouteSearchProcess finds features around a route that are heading along the route and not just crossing over it.

Parameters	Description
features	Input feature collection to query.
routes	Routes to search along. Features must have a geometry of LineString.
bufferSize	Buffer size (in meters) to search around the route.
headingThreshold	Threshold for comparing headings, in degrees.
routeGeomField	Attribute that will be examined for routes to match. Must be a LineString.
geomField	Attribute that will be examined for route matching.
bidirectional	Consider the direction of the route or just the path of the route.
headingField	Attribute that will be examined for heading in the input features. If not provided, input features geometries must be LineStrings.

12.2.11.1. Route search example (XML)

RouteSearchProcess_wps.xml is a geoserver WPS call to the GeoMesa RouteSearchProcess. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@RouteSearchProcess_wps.xml localhost:8080/geoserver/wps

12.2.12. SamplingProcess

The SamplingProcess uses statistical sampling to reduces the features returned by a query.

Parameters	Description
data	Input features.
samplePercent	Percent of features to return, between 0 and 1.
threadBy	Attribute field to link associated features for sampling.

12.2.12.1. Sampling example (XML)

SamplingProcess_wps.xml is a geoserver WPS call to the GeoMesa SamplingProcess. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@SamplingProcess_wps.xml localhost:8080/geoserver/wps

12.2.13. StatsProcess

The StatsProcess allows the running of statistics on a given feature set.

Parameters	Description
features	The feature set on which to query. Can be a raw text input, reference to a remote URL, a subquery or a vector layer.
statString	Stat string indicating which stats to instantiate. More info here Stat Strings.
encode	Return the values encoded as json. Must be true or false; empty values will not work.
properties	The properties / transforms to apply before gathering stats.

12.2.13.1. Stat Strings

Stat strings are a GeoMesa domain specific language (DSL) that allows the specification of stats for the iterators to collect. The available stat function are listed below:

Note

Items marked with * are the name of an attribute, either in your sft or as the result of a transformation or projection.

Note

A TimePeriod is defined as one of the following strings: “day”, “week”, “month”, “year”

Syntax	Parameters	Description
Count
Count()		Counts the number of features.
MinMax
MinMax(attribute)	attribute*: String	Finds the min and max values of the given attribute.
GroupBy
GroupBy(attribute,stat)	attribute*: String stat: Stat String	Groups stats by the given attribute and then runs the given stat on each group. Any stat can be provided.
Descriptive Stats
DescriptiveStats(attribute)	attribute*: String	Runs single pass stats on the given attribute calculating stats describing the attribute such as: count; min; max; mean; and population and sample versions of variance, standard deviation, kurtosis, excess kurtosis, covariance, and correlation.
Enumeration
Enumeration(attribute)	attribute*: String	Enumerates the values in the give attribute and the number of occurrences.
TopK
TopK(attribute)	attribute*: String	TopK of the given attribute
Histogram
Histogram(attribute,numBins,lower,upper)	attribute*: String numBins: Int lower: Int upper: Int	Provides a histogram of the given attribute, binning the results into a binned array using the numBins as the number of bins and lower and upper as the bounds of the binned array.
Frequency
Frequency(attribute,dtg,period,precision)	attribute*: String dtg*: String period: TimePeriod precision: Int	Estimates frequency counts at scale.
z3Histogram
Z3Histogram(geom,dtg,period,length)	geom*: String dtg*: String period: TimePeriod length: Int	Provides a histogram similar to Histogram but treats the geometry and date attributes as a single value.
z3Frequency
Z3Frequency(geom,dtg,period,precision)	geom*: String dtg*: String period: TimePeriod precision: Int	Provides a freqency estimate similar to Frequency but treats the geometry and date attributes as a single value.
Iterator Stack
IteratorStackCount()		IteratorStackCount keeps track of the number of times Accumulo sets up an iterator stack as a result of a query.

12.2.14. TrackLabelProcess

The TrackLabelProcess returns a single feature that is the head of a track of related simple features.

Parameters	Description
data	Input features.
track	Track attribute to use for grouping features.
dtg	Date attribute to use for ordering tracks.

12.2.14.1. TrackLabel example (XML)

TrackLabelProcess_wps.xml is a geoserver WPS call to the GeoMesa TrackLabelProcess. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@TrackLabelProcess_wps.xml localhost:8080/geoserver/wps

12.2.15. TubeSelectProcess

The TubeSelectProcess performs a tube select on a Geomesa feature collection based on another feature collection. To get more informations on TubeSelectProcess and how to use it, you can read this tutorial.

Parameters	Description
tubeFeatures	Input feature collection (must have geometry and datetime).
featureCollection	The data set to query for matching features.
filter	The filter to apply to the featureCollection.
maxSpeed	Max speed of the object in m/s for nofill & line gapfill methods.
maxTime	Time as seconds for nofill & line gapfill methods.
bufferSize	Buffer size in meters to use instead of maxSpeed/maxTime calculation.
maxBins	Number of bins to use for breaking up query into individual queries.
gapFill	Method of filling gap (nofill, line).

12.2.15.1. TubeSelect example (XML)

TubeSelectProcess_wps.xml is a geoserver WPS call to the GeoMesa TubeSelectProcess. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@TubeSelectProcess_wps.xml localhost:8080/geoserver/wps

12.2.16. QueryProcess

The QueryProcess takes an (E)CQL query/filter for a given feature set as a text object and returns the result as a json object.

Parameters	Description
features	The data source feature collection to query. Reference as store:layername. For an XML file enter <wfs:Query typeName=store:layername /> For interactive WPS request builder select VECTOR_LAYER & choose store:layername
filter	The filter to apply to the feature collection. For an XML file enter: <wps:ComplexData mimeType="text/plain; subtype=cql"> <![CDATA[some-query-text]] </wps:ComplexData> For interactive WPS request builder select TEXT & choose "text/plain; subtype=cql" enter the query text in the text box
output	Specify how the output feature collection will be presented. For an XML file enter: <wps:ResponseForm> <wps:RawDataOutput mimeType="application/json"> <ows:Identifier>result</ows:Identifier> </wps:RawDataOutput> </wps:ResponseForm> For interactive WPS request builder check the Generate box and choose “application/json”
properties	The properties / transforms to apply before gathering stats.

12.2.16.1. Query example (XML)

QueryProcess_wps.xml is a geoserver WPS call to the GeoMesa QueryProcess that performs the same query shown in the Accumulo-quickstart. It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@QueryProcess_wps.xml localhost:8080/geoserver/wps

The query should generate results that look like this:

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": {
        "type": "Point",
        "coordinates": [
          -76.513,
          -37.4941
        ]
      },
      "properties": {
        "Who": "Bierce",
        "What": 931,
        "When": "2014-07-04T22:25:38.000+0000"
      },
      "id": "Observation.931"
    }
  ]
}

12.2.17. UniqueProcess

The UniqueProcess class is optimized for GeoMesa to find unique attributes values for a feature collection, which are returned as a json object.

Parameters	Description
features	The data source feature collection to query. Reference as store:layername. For an XML file enter <wfs:Query typeName=store:layername /> For interactive WPS request builder select VECTOR_LAYER & choose store:layername
attribute	The attribute for which unique values will be extracted. Attributes are expressed as a string. For an XML file enter <wps:LiteralData>attribute-name</wps:LiteralData>
filter	The filter to apply to the feature collection. For an XML file enter: <wps:ComplexData mimeType="text/plain; subtype=cql"> <![CDATA[some-query-text]] </wps:ComplexData> For interactive WPS request builder select TEXT & choose "text/plain; subtype=cql" enter the query text in the text box.
histogram	Create a histogram of attribute values. Expressed as a boolean (true/false). For an XML file enter <wps:LiteralData>true/false</wps:LiteralData>
sort	Sort the results. Expressed as a string; allowed values are ASC or DESC. For an XML file enter <wps:LiteralData>ASC/DESC</wps:LiteralData>
sortByCount	Sort by histogram counts instead of attribute values. Expressed as a boolean (true/false). For an XML file enter <wps:LiteralData>true/false</wps:LiteralData>
output	Specify how the output feature collection will be presented. For an XML file enter: <wps:ResponseForm> <wps:RawDataOutput mimeType="application/json"> <ows:Identifier>result</ows:Identifier> </wps:RawDataOutput> </wps:ResponseForm> For interactive WPS request builder check the Generate box and choose “application/json”

12.2.17.1. Unique example (XML)

UniqueProcess_wps.xml is a geoserver WPS call to the GeoMesa UniqueProcess that reports the unique names in in the ‘Who’ field of the Accumulo quickstart data for a restricted bounding box (-77.5, -37.5, -76.5, -36.5)). It can be run with the following curl call:

curl -v -u admin:geoserver -H "Content-Type: text/xml" -d@UniqueProcess_wps.xml localhost:8080/geoserver/wps

The query should generate results that look like this:

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "properties": {
        "value": "Addams",
        "count": 37
      },
      "id": "fid--21d4eb0_15b68e0e8ca_-7fd6"
    },
    {
      "type": "Feature",
      "properties": {
        "value": "Bierce",
        "count": 43
      },
      "id": "fid--21d4eb0_15b68e0e8ca_-7fd5"
    },
    {
      "type": "Feature",
      "properties": {
        "value": "Clemens",
        "count": 48
      },
      "id": "fid--21d4eb0_15b68e0e8ca_-7fd4"
    }
  ]
}

12.2.18. Chaining Processes

WPS processes can be chained, using the result of one process as the input for another. For example, a bounding box in a GeoMesa QueryProcess can be used to restrict data sent to StatsProcess. GeoMesa_WPS_chain_example.xml will get all points from the AccumuloQuickStart table that are within a specified bounding box (-77.5, -37.5, -76.5, -36.5), and calculate descriptive statistics on the ‘What’ attribute of the results.

The query should generate results that look like this:

{
  "type": "FeatureCollection",
  "features": [
    {
      "type": "Feature",
      "geometry": {
        "type": "Point",
        "coordinates": [
          0,
          0
        ]
      },
      "properties": {
        "stats": "{\"count\":128,\"minimum\":[29.0],\"maximum\":[991.0],\"mean\":[508.5781249999999],\"population_variance\":[85116.25952148438],\"population_standard_deviation\":[291.74691004616375],\"population_skewness\":[-0.11170819256679464],\"population_kurtosis\":[1.7823482287566166],\"population_excess_kurtosis\":[-1.2176517712433834],\"sample_variance\":[85786.46628937007],\"sample_standard_deviation\":[292.893267743337],\"sample_skewness\":[-0.11303718280959842],\"sample_kurtosis\":[1.8519712064424219],\"sample_excess_kurtosis\":[-1.1480287935575781],\"population_covariance\":[85116.25952148438],\"population_correlation\":[1.0],\"sample_covariance\":[85786.46628937007],\"sample_correlation\":[1.0]}"
      },
      "id": "stat"
    }
  ]
}