Hello all
An OGC meeting is going on right now in Arlington. The meeting is not
yet over, but there is a summary of some points discussed so far:
Web Coverage Services (WCS)
-------------------------------------------------
A specification mapping GeoTIFF tags to standards elements (ISO) has
been adopted recently. Such GeoTIFF-ISO mapping is similar to the
NetCDF-ISO mapping defined by NOAA. Apache SIS already implements the
NetCDF-ISO mapping [1], we should also implement the GeoTIFF-ISO mapping
using that document. Note that this specification is not about the
GeoTIFF format itself, however there is some talk about whether OGC
should also write a specification for it. In the current state, the
community maintaining the GeoTIFF specification is somewhat informal.
Other data specification under work are NetCDF and JPEG2000. The
NetCDF-ISO mapping currently implemented by SIS is based on NOAA
documentation. If OGC defines formally the mapping in a specification,
we should probably update SIS accordingly.
Above paragraphs are about "data". On the "services" side, recently
adopted specifications extending the core are:
- CRS
- range subsetting
- scaling
- interpolation (often used together with scaling, but not necessarily)
The meteorological-oceanographic working group said that the CRS and
range subsetting extensions worked well for them. However an Earth
Observation and meteorological-oceanographic profiles of WCS may still
be needed for other reasons, to be discussed in the "MetOcean" section
below in this email.
A public wiki is at [2].
Transactional Web Coverage Services (WCS-T)
-------------------------------------------------
WCS-T 2.0 is a rewrite of WCS-T 1.4 without upward compatibility - the
changes are major. WCS-T purpose is to insert, update and delete WCS
coverage offerings. Specification for insert and delete operations are
done, update operation is in progress. The difficulty for the update
operation is to specify the criterion for determining if an update is
allowed or not (e.g. replacing a 3-banded image by a 1-banded image
should probably not be allowed).
Web Processing Service (WPS)
-------------------------------------------------
A WPS 2 specification is under way. WPS 2 improvements compared to WPS 1
are:
* Better support for process discovery
* Improved execution management (long running processes, ability to
delete a running job)
* Better support for synchronous / asynchronous execution
A specification draft for public comment is expected by the end of April.
Meteorological-oceanographic group (MetOcean)
-------------------------------------------------
MetOcean discussed about the extensions they wish to bring to WCS.
Rational for extensions are related to data shapes: GRIDS, time series,
cross sections, point collections (e.g. observations), vertical profiles
(e.g. ascents), trajectories. The group emphases on the needs for
sub-settings on the 4 axes, probably 5 axes in the future where the
fifth axis would be "ensemble" (e.g. probability, discussed below).
Keeping in mind that MetOcean uses two time axis (forecast time, model
run reference time), we have a potential for 6 dimensional coverages.
The group emphases also on slice, trim and sub-setting operations (e.g.
trim X, trim Y, slice Z = horizontal view).
Some extensions needed:
* Slice and trim requested in other CRS than the source data CRS.
* Want to advertise validity times for coverage data availability.
* Challenge: create 3D/4D grid from irregular grids with missing data.
* Needs to group coverages by collections (other amount of coverages
is unmanageable) with their own metadata. Note that the need to
define collection of coverages is probably a wider than the MetOcean
domain. Generalization will be investigated.
* Custom WCS operations: DescribeCoverageCollection, GetCorridorCoverage.
An OGC document is hoped for the end of this year.
The group provided some patterns of data extraction. For example
extracting vertical profile data for a trajectory (marine example:
sonar) = trim operation expressed in the CRS of the trajectory. The
group had a discussion about vertical CRS of the kind "pressure in hPa
relative to a changing datum (a pressure surface)". The action for now
is to take a closer look to ISO 19111-2 (parametric CRS). I think that
Apache SIS too should take a closer look to ISO 19111-2.
The MetOcean group presented "ensemble forecasts": producing more than
20 parallel forecasts based on original observations in order detect
"butterfly effect". Some statistical numbers of interest are quartile
values, decile values, 95% and 5% percentile values, mean and mode. Open
question is: can those needs be adressed by the NetCDF-Uncertainty
extension?
The MetOcean group has a projects on GitHub [3] providing the following
resources:
* World weather symbols, as SVG or PNG files.
* WMO Regional Associations GeoJSON.
* WMO core metadata profile 1.3. They provide an application
performing validation for a profile of ISO 19115 metadata.
The last point is related to Apache SIS, since we implement ISO 19115. I
will create a "Provides a meterological profile of ISO 19115" JIRA task.
Since I think there is an other Apache project related to climate, maybe
there is an opportunity to contact each other here?
Big data
-------------------------------------------------
Big data has been defined as the analysis on multi-dimensional arrays of
size several orders of magnitude above engine's main memory. A
discussion working group has been formed for investigating possible
issues with OGC specifications for achieving this goal. Chairs have been
nominated. As a side note, the European Space Agency (ESA) organizes a
conference on big data from space on November 12-14th, 2014 in Fracati,
Italy.
Moving features
-------------------------------------------------
A reference implementation named "Mobmap" has been shown. This
implementation run on Chrome browser and uses CSV files (points only) as
inputs. They were discussion about using NetCDF as a more efficient
binary storage format, but there is a difficulty in complying to the
NetCDF "trajectory" convention. An alternative could be to take some
liberties regarding the NetCDF trajectory convention, but this is
something that peoples would prefer to avoid.
The specification had only minor editions, and now includes the use case
provided by Nadeem Anjum on this mailing list. A final version of Moving
Feature specification is expected for December.
They were an interesting discussion about JSon being - as surprisingly
as it may seem - more verbose than XML for data like the Moving Feature
ones, because of massive repetition of keywords. I'm not a JSon expert
so I can not elaborate...
CITE Tests (Team engine):
-------------------------------------------------
The "Team engine" is the testing framework at OGC. It is different than
the "GeoAPI-conformance" module that we use for Apache SIS. OGC is
working on making Team Engine easier to use - currently it has been
reported that an expert needs 2.5 hours to configure and run tests. It
is too early for Apache SIS to use Team Engine, since we do not yet
implement web services. When we will start looking at CITE, ideally I
would like to see if some CITE-GeoAPI integration are possible.
Miscellaneous
-------------------------------------------------
Some related initiatives mentioned in the meeting are:
* Integrated Ocean Observing System (IOOS) [4]
* Integrated Justice Information System (IJIS)
* WMO Climate Data Management System Specification (CDMS)
They were also a mention about recent and coming discussion in other
organizations about whether we should keep leap second. The
International Atomic Time (TAI) currently differs from Coordinated
Universal Time (UTC) by 35 seconds due to leap seconds added in the last
decades. Some organizations want to suppress leap seconds. No decision
has been taken yet. Of course the decision does not depend on OGC, but
opinions are collected.
OGC is still experimenting ways to publish their specification as web
pages. A prototype is available for the GeoPackage specification.
However there is not yet a policy for allowing external projects - like
Apache SIS - to provide links to a stable anchor of OGC specifications.
For example if Apache SIS wants to link to a specific clause of an OGC
specification, whether this will be possible is still an open question.
Martin
[1]
http://sis.apache.org/apidocs/org/apache/sis/storage/netcdf/AttributeNames.html
[2] http://external.opengeospatial.org/twiki_public/CoveragesDWG/
[3] https://github.com/OGCMetOceanDWG
[4] http://www.ioos.noaa.gov/
