Introducing OGC Web Services at FMI

EGOWS 2008 Ljubljana

Ilkka Rinne, Finnish Meteorological Institute

Presentation outline
Overview of the OGC Web Services (WMS, WFS, WCS and CS-W)

Meteorology and OGC: GML in observations and forecasts

Technology: Experiences from OGC

Real world uses for the OGC Web
Web Service implementations used at Services: Experiences from the FMI
FMI (UMN MapServer & Geoserver) pilot projects

What's in the horizon: Plans and expectations in the near future

Open Geospatial Consortium (OGC) Web Service

Web Map Service

(WMS)

Catalogue Service
for Web (CS-W)

Superimposed "flat" maplike images.

Images: png, svg, KML,..

Web Feature
Service (WFS)

Web Coverage
Service (WCS)

Geospatial objects:
types or instances

Gridded data with GML

metadata

GML documents

Binary files: GeoTiff, HDF-EOS,

CF-NetCDF, JPEG2000,...

Registries for provided

services, GML Application
Schemas, visualization
descriptions, standardized
terms, ontologies,...
XML documents

Web Map Service (WMS)

Map-like views of pre-rendered
information: HTTP queries to a WMS
server return pre-rendered images
(layers) from different sources to be
layered on top of each other in the client
application.
Available layers, projections, and other
parameters can be queried with
GetCapabilities-request.

Map
Roads
Satellite
NWM Precipitation
Radar Precipitation

Web Feature Service (WFS)

Web Feature Service defines protocols for

requesting as well as (optionally, WFS-T)
inserting and modifying GML Features
(Geospatial objects) through HTTP
requests.
GML Features are geospatial objects
semantically describing real-world entities,
like Meteorological objects (fronts,
precipitation areas), weather observation
readings, observation stations or roads.
WFS can be used as a client-server data
retrieval, server-to-server data
synchronization or as a common access
interface on top of heterogenous
geospatial data stores.

Optionalcall
Typical
Transaction
sequencerequests
in WFS client
(WFS-T)
requesting
enable
inserting,
a GMLupdating,
Feature (or
deleting
set ofand
Features):
locking
GML Features:
WFS
Client
What is
available?
What can I
do with it?
What can I
expect to
receive?
Select and
receive the
features

WFS-T
WFS
Server

GetCapabilities()
Capabilities document
DescribeFeature(X)
XML Schema for the feature

GetFeature()
GML FeatureCollection

edit
feature y
Transaction:Update(y)
Update(y)
Transaction:

Request: GetCoverage

Web Coverage Service (WCS)

Similar to WMS and WFS, but instead of

images or features deals with coverages:
collection of mappings from
geometry (grid points) into parameter
(temperature, pressure,..) values = grid
data.
Analogically to WFS, the client requests
the WCS server for available data
(GetCapabilities), asks it to describe the
interesting coverages (DescribeCoverage)
and create and return the requested subset
of the available coverage data
(GetCoverage). Optionally only the HTTP
URLs for the response data sets can be
returned as an XML document.

Response: a data grid

Catalogue Service for Web (CS-W)

"Catalogue services support the ability to

publish and search collections of descriptive
information (metadata) for data, services, and
related information objects."
"Metadata in catalogues represent
resource characteristics that can be queried
and presented for evaluation and further
processing by both humans and software."
"Catalogue services are required to support
the discovery and binding to registered
information resources within an information
community."

A very likely INSPIRE requirement

OGC: ebRIM is the preferred

meta-model for future App. Profiles

CS-W = OpenGIS Catalog Service + HTTP Protocol Binding

(OpenGIS Catalogue Services Specification)

Web Registry Service = CS-W with ebRIM AP

Feature Portrayal
WMS can be used as presentation frontend for feature and coverage services: A
WMS instance retrieves the data and
renders it into an image using predefined
rendering rules.
Presentation rules and symbols are
described in separate documents having
their own language, like Styled Layer
Descriptor (SLD) or XSLT (to create SVG
or KML).

A Cold Front GML Feature

These presentation rules can be

maintained in a CS-W registry and shared
with all interested parties.
Dedicated Feature Portrayal Service (FPS)
specification is under discussion in OGC

A Styled Layer Descriptor (SLD) for Cold Fronts

Weather Observations & OGC

Two models for observation events:

GML Observation in the GML Core
schema and the O&M Observation in
the OGC Observations &
Measurements (O&M) schema.
Meteorological observations may be
measurements with numeric values,
category observations (generic type of
weather, type of cloud), truth-valued (is
it raining?).

gml:AbstractFeature

gml:Observation
validTime
using
target
resultOf

Models a general act

of observing something
like taking a picture of a
flower
Simple

om:Observation
samplingTime
resultTime
resultQuality
parameter
procedure
featureOfInterest
observedProperty
result

"Scientific" observation from

measurement collection point
of view
More complex, more flexible

Results for observations over spatial

areas and/or non-singular time instance
are coverages.

O&M model suits meteorological observations

well, but its properties need to be more specific
to enable interoperability in
climate-environmental community

Weather Forecasts & OGC

No obvious GML Application schema for
modeling forecasting events exists (as far as I
know).
Results are coverages in most cases (grids or
collections of point-value pairs); the same data
structures as with observations are useful, but
the overlapping forecast time periods are
problematic to express in queries.
Completely open: Expressing forecast
uncertainty? Spatial and temporal interpolation
of values?
We, as part of the climate-environmental
community, should be more active in the
standardization process to enable global
interoperability.

gml:AbstractFeature

WeatherForecast
validTimePerioid
analysisTime
creationTime
featureOfInterest

result
uncertainty
gml:Coverage

NumericalWeatherModel
*
*
ForecastingProcess?

gml:Coverage

Severe weather warnings: Meteoalarm could

be a very good case for defining a GML
Application Schema for weather warnings and
distributing them through WFS.

Experiences & Technologies

We currently have experience of two server products supporting OGC Web Services at
FMI: MapServer and Geoserver
UMN MapServer

Implements WMS, WFS 1.0, WCS 1.0, GML 2.1.2 & 3.1.0 Level 0, SLD and Filter
Encoding. Supports many raster formats and vector formats through GDAL and OGC
libraries: ArcInfo, DODS/OPeNDAP, GRIB1&2, NetCDF, TIFF/ GeoTIFF, JPEG2000,
Netpbm, ESRI ArcSDE & Shapefile,...
Well suited for WMS server: easy to configure, stable, direct Oracle SQL queries
possible (even without Spatial), PostgreSQL, SQLite. On-the-fly (re)projections of the
data using Proj4. Configuration is file-based (like it or not).
Biggest problems with meteorological grid data formats: reading does not always
succeed even if it should be supported, probably caused by slight incompatibilities in
data formats.
Recommendation: for easy-to-setup WMS

Experiences & Technologies

Geoserver

Implements WMS, WFS 1.0. & 1.1, WCS, SLD, Google Earth KML as WMS output format.
WFS-T
Supports PostGIS, DB2, Oracle, MySQL and ESRI Shapefiles as input data sources.
Possibility to build more with GeoTools DataStore API. Streaming data: no memory limits for
serving large datasets.
Deployed as a standard WAR package inside Java Servlet containers. Integrated Web GUI
for configuration, good or bad depending of if you like a click-through or write-through setup.
Currently outperforms MapServer in WFS support.
Directly supports only simple features: feature properties must match columns in the DB
tables. Support for complex features is on the roadmap.
Recommendation: For building a Java EE based, extensible geospatial data server,
especially for WFS use

OGC Web Service -Related Projects at FMI

Ubicasting

Cooperating EO
Sensors (COPS)

Smartmet II

WMS: radar data,

WFS: point observations and forecasts,
WFS-T: chemical release features, triggering the dispersion model calculations

WMS: Weather information and air quality maps (temperature, wind,..)

WFS: Observation data from Helsinki Testbed area
WCS: MODIS satellite data received in Sodankyl station

WMS: Client for background maps, or any other geospatial layers for the workstation
WFS-T: Meteorological objects created by the meteorologists (fronts, symbols,...)

Ubicasting
Goal: Develop technologies for enabling small-scale highly targeted, near-realtime
combined weather & air quality information for anybody, anywhere (ubiquitous computing).
The use cases of the pilot customers: Provide quick response weather and dispersion
information for the rescue activities in the case of chemical or radioactive substance
releases.
Continuum of the Helsinki Testbed project: The focus at FMI side is to try to build OGC
compatible interfaces for providing point observation, point forecast, lightning and radar data
for the external partners to use in their applications.
A lot of OGC related discussion and on-going work with Vaisala (a recent OGC member)
Status (June 2008): radar images and simple point forecast features available through
Mapserver
During summer-autumn 2008: More elaborate forecast and observation features

WFS GetFeature reply:

Ubicasting Simple GML Features

gml:FeatureCollection
gml:AbstractFeatureType
description
name
boundedBy

ubi:ObservationFeature
gml:PointPropertyType

ubi:ObservationFeature
ubi:ObservationFeature

geometry_data
ubi:ObservationFeatureType
station_name
altitude
height
value
status
data_type
creation_time

Good: Can be served by Geoserver with standard setup

and easily mapped with PostgreSQL database (one
table/view per feature, one column per property)

ubi:ObservationFeature
ubi:ObservationFeature
ubi:ObservationFeature
ubi:ObservationFeature
ubi:ObservationFeature
ubi:ObservationFeature
.
.
.

ubi:ObservationFeature

Bad: Verbose and not flexible

Ubicasting GML Features:

Observations & Measurements schema
gml:AbstractFeatureType
description
name
boundedBy

WFS GetFeature reply:

gml:FeatureCollection
om:Observation
metadata
samplingTime
resultTime

ObservationStation
featureOfInterest
om:Observation
samplingTime
resultTime
resultQuality
parameter
procedure
observedProperty
result

Very flexible (even too flexible?) structure of the

observation events. References to the other GML
Features, possibility for expressing many kinds of
observations with the same schema (good & bad)
Needs complex DB-object mapping

procedure
resultQuality
observedProperty
featureOfInterest
parameter
result

.
.
.

om:Observation

Cooperating EO Sensors (COPS)

An ESA technology research project for building applications which combine and share insitu sensor observation data and remote EO (satellite) data using OGC Web Service
interfaces.
In the Finnish part, the sensor data from the Helsinki Testbed area is provided by using
WFS, and MODIS satellite data coverages by using WCS.
Mixing height and Surface layer model results are provided both as data (WCS) and previsualized images (WMS).
OGC Sensor Observation Service (SOS) is used in the Spanish subsystem, but from the
FMI side we want to use the same WFS planned in the Ubicasting project to distribute
weather observations => WFS-SOS adapter. SOS observations are expressed in O&M, so
this should not be a big problem.
WCS implementation to use is still a big open question.
Status (June 2008): Detailed technical design phase starting

Smartmet II
The new Java-based Meteorological Workstation in development at FMI
Completely modular (component-based internal architecture), the selected plugin set
defines both the GUI features and the operation implementations) => both standalone and lightweight, heavily server-dependent configurations possible.
Possibility to include geospatial WMS layers in the display stack (WMS client plugin)
Meteorological Objects editor plugin: visualization and editing of fronts, symbols,..
Serialized as GML Features and stored at the server (could be implemented using
WFS-T, but current early version of the server is very simple servlet storing any
XML).
Satellite and radar data currently fetched as pre-rendered images through existing
HTTP servers. Will move into using WMS and possibly WCS when they become
available.
Status (June 2008): early version with MetObject editor plugin released for testing by
the meteorologists

OGC Web Services Outlook at FMI for 2008-09

Long-term goal: Simplify the maintenance of the data transfer both in-house and with
external partners by using open standard interfaces in the information exchange
whenever possible. At the moment OGC Web Services are considered to be a very
important part of implementing this scenario.
Gather more information and experience both about the standards and the available
implementations though the on-going pilot and research projects.
Take part in community efforts for finding the best practices and improving the
standards to better suit the meteorological needs.
Setup a WFS for distributing point observations and forecasts using Geoserver
Find a suitable implementation for delivering satellite data using WCS (Geoserver,
THREDDS?)
Continue the planning of an internal Catalog Service, possibly using CS-W