The status of the GeoServer WPS

The status of GeoServer WPS

Ing. Andrea Aime, GeoSolutions

Ing. Simone Giannecchini, GeoSolutions

FOSS4G 2011, Denver 12th-16th September 2011

Overview

Who we are

WPS introduction

Vector processes

Raster processes

Conversion processes

Accessing local data sources

Chaining

Examples

Rendering transformations

Limitations and wish list

Question time


GeoSolutions

Founded in Italy in 2006

Expertise

• Image Processing, GeoSpatial Data Fusion

• Java, Java Enterprise, C++, Python

• JPEG2000, JPIP, Advanced 2D visualization

Supporting/Developing FOSS4G projects

GeoTools, GeoServer

GeoBatch, GeoNetwork

Clients Public Agencies

Private Companies

http://www.geo-solutions.it


http://www.geo-solutions.it/




WPS Quick introduction


WPS

Wikipedia introduces OGC WPS as:

[A service] designed to standardize the way that GIS calculations are made available to the Internet.

WPS can describe any calculation including all of its inputs and outputs, and trigger its execution

The specific processes served up by a WPS implementation are defined by the owner of that implementation.

Although WPS was designed to work with spatially referenced data, it can be used with any kind of data.


Capabilities document

Some metadata about the server and its owner

The list of available processes, with a description

Let’s say we’re interested in the JTS:buffer process


Process description

List of inputs and outputs

Descriptions

List of accepted formats (not shown in the xml)


Process execution

Buffer a L shaped geometry with distance “2”

Get the result back as GML


GeoServer Specifics


GeoServer WPS history

Started by Refractions in 2008, with limited capabilities (only single geometry and single feature support)

First overhaul attempt end of 2008 by the community, added testing, support for vector collections

New development since mid 2010, mostly new processes and input/output formats

Refractions GS community Current activity

2008 2009 2010 2011


WPS: demo builder

List processes

Describe

Set parameters and execute

All in one form


Inputs and outputs


IO setup

Processes parameters defined in terms of Java objects

Pluggable converters trade between the java object and the serialized representation

If a input is internal straight read from the source

Leveraging on all the available optimizations!


“Primitives” IO Parameters

Numbers: byte, short, int, long, float, double, any Number subclass, properly mapped in XML types

Strings and CharSequence in general

Date, Time, Timestamp

CoordinateReferenceSystem (EPSG:xxx and urn:… forms)

URLs

Range (min -> max)

Interpolation method


Complexes IO Parameters

Geometries

GML 2

GML 3

WKT

Rasters

ArcGrid

GeoTiff

Unreferenced PNG/JPEG

Vectors

WFS 1.0 collection

WFS 1.1 collection

GeoJSON

Zipped shapefile

Others

SLD 1.0

OGC Filter (1.0, 1.1)

CQL Filter


Open API

Easy to plug your own custom process parameter I/O


Available Vector processes


JTS processes


Example JTS process Intersection

POLYGON ((5 10, 10 10, 10 5, 5 5, 5 10))


Query oriented processes

AKA “All you wanted WFS to do for you but he never

wanted to”

gs:Aggregate: count/avg/max/median/min/stddev/sum on a feature collection

gs:Count: like WFS Hits, but on whatever source

gs:Bounds: bounds of whatever source, missing from WFS

gs:Query: query any source like WFS

gs:Unique: unique values of an attribute

gs:Nearest: find the nearest features


Aggregation example

Get the min, max and sum of the PERSONS attribute in the topp:states layer


Other Vector processes

gs:BufferFeatureCollection: buffer all features

gs:Clip: cookie cut featuresgs:Reproject: reproject any vector source

gs:Simplify: DouglasPeucker simplifier, retain attributes

gs:Snap: snap to grid

gs:InclusionFeatureCollection: overlay and get all features contained

gs:IntersectionFeatureCollection: overlay and intersect, retain attributes from both

gs:UnionFeatureCollection: merge two collections in one

gs:Import: save the features as a new GeoServer layer


Buffer example


Geometry <=> Feature

gs:collectGeometries: lump up all feature geometries into

a geometry collection

gs:feature: turn a single geometry into a feature collection

MULTIPOINT (

(-74.01046109936333 40.707587626256554),

(-74.0108375113659 40.70754683896324),

(-74.01053023879955 40.70938711687079),

(-74.00857344353275 40.711945649065406),

(-74.0118315772888 40.708529961953786),

(-74.00153046439813 40.719885123828675))


Raster processes


Raster processes

AKA “All you wanted WCS to do for you but it never wanted to”

Add/Multiply: add and multiply two rasters (waiting for full algebra to be implemented)

Crop: crop a coverage based on the specified cutting geometry

RangeLookup: classify raster image based on a set of ranges ([min,max] -> value)

ScaleCoverage: rescale and translate a given raster

StyleCoverage: apply a SLD style to a raster, getting back a styled (but still georeferenced) one

Georectify: turn a non geo-referenced coverage into one based on ground control points


Crop example




Bridging the raster and the vector world

Contour: extracts isolines given a set of levels or a interval

RasterAsPointCollection: extracts one point for each cell, with band contents as attributes

PolygonExtraction: extracts uniform polygons from raster, eventually given a set of value ranges

RasterZonalStatistics: given a raster and a polygonal compute min/max/sum/avg/stddev of the cells falling in each polygon, return an augmented polygonal

VectorToRaster: rasterizes vectors keeping a chosen attribute (a CQL expression eventually) as the band value

Contour example

Chaining


WPS Chaining

Feed the output of a process into another process

Allows for tree-like composition

Let’s see a typical “clip and ship” example, both raster and vector:

Extract the rivers into the “restricted areas” polygons

Extract from Blumarble any pixel in the USA


Raster clip and ship


Raster clip and ship


Vector clip and ship


GeoServer Integration 1: local data access


Direct data integration

WPS normally reads from remote WFS/WCS, parsing GML/GeoJSON or GeoTiff/ArcGrid

When the source is local we can dodge it though, read directly from the source (shapefile, DBMS, geotiff)

Hit the fictious http://geoserver/wfs url for local WFS

Hit the fictious http://geoserver/wcs url for local WCS

http://geoserver/wfs



Store back results (only vector)

gs:Import: saves the vector results into a store of choice, and publish as a layer

Use right away the new layer from WMS/WFS/WCS/WPS

Still missing the equivalent for rasters


GeoServer Integration 2: rendering transformations



On-the-fly data transformations inside rendering

chain

Calling WPS processes from SLD docs

Optimized for performance



Point feature extraction from two band raster data (e.g. Wind(u,v))

Computation of direction and module from SLD

Full SLD

Call gs:RasterAsPointCollection

Magnitude and direction of the arrow are computed on the fly by using filter functions

<WellKnownName>shape://carrow</WellKnownName>

Working at visual resolution

Use overviews and decimation

Fast with large datasets


http://demo.geo-solutions.it/share/garrows2.sld

Evolution and closing remarks


WPS*

Deficiencies

No support for asynchronous requests

Missing request limits enforcements (e.g. input/output maximum dimensions)

Wish list:

Scripting (Jython, GeoScript)

Sextante, IDL, JGrass (Grass?) integration

Improved robustness

Jiffle (jai-tools) based raster algebra

New layers as dynamic WPS processes (computing data on the fly as people do WMS/WCS/WFS requests)

FOSS4G 2011, Denver

12th-16th September 2011

The End

Questions? [email protected]

[email protected]


The status of the GeoServer WPS

Technology

vector processes gs

gml foss4g

outputs foss4g

features gs

geosolutions foss4g

geometry feature gs

attribute gs

attributes gs