Skip to content

Latest commit

 

History

History
154 lines (124 loc) · 5.61 KB

README.md

File metadata and controls

154 lines (124 loc) · 5.61 KB
Raw

Abomination ncwms

This plugin mimicks ncwms operation in geoserver by proxying requests to a thredds server using WFS as an index for time <-> url.

How does it work?

ncwms provides wms over netcdf files. In addition it also provides the following functions:

  • Get all dates with data
  • For a specific date - returns the times with data

ncwms knows that because it indexes the files periodically.

The abomination ncwms can know all of that because we have _url layers which map between time and filename. So we use this reference layer to find the information we need.

Mimicking date/time queries

To mimick the request of get all dates with data we simply ask for all features sorted by date for the given layer (_url) and then we just uniqify the dates.

To mimick the request of getting all times for a specific date, we craft a WFS CQL query and ask just for a specific day. The CQL query will look similar to:

"TIME >= that_day AND TIME < next_day",

Mimicking WMS

That's the easy part. Given we know what netcdf file we need to serve wms for, we can just proxy it to a thredds server that serves this file.

All incoming WMS requests should have a TIME= string. Otherwise we use the latest time for the layer. We can obtain it by sorting by time in a descending order and just take the first feature, that would be the latest TIME.

Thredds has full ncwms capabilities per file, for instance: http://thredds.aodn.org.au/thredds/wms/IMOS/ACORN/gridded_1h-avg-current-map_non-QC/CBG/2015/12/10/IMOS_ACORN_V_20151210T003000Z_CBG_FV00_1-hour-avg.nc?service=WMS&version=1.3.0&request=GetCapabilities

So the wms endpoint for the file IMOS/ACORN/gridded_1h-avg-current-map_non-QC/CBG/2015/12/10/IMOS_ACORN_V_20151210T003000Z_CBG_FV00_1-hour-avg.nc is at: http://thredds.aodn.org.au/thredds/wms/IMOS/ACORN/gridded_1h-avg-current-map_non-QC/CBG/2015/12/10/IMOS_ACORN_V_20151210T003000Z_CBG_FV00_1-hour-avg.nc

The abomination ncwms plugin will just proxy all requests to the wms endpoint of the corresponding file.

Caveats

You must pass SERVICE=ncwms instead of SERVICE=wms, otherwise the key/value pair parsing will be delegated to the original geoserver wms plugin and we don't want that because it'll parse them in a wrong way.

Metadata configuration

Usually the metadata configuration for ncwms would look like:

<gmd:onLine>
    <gmd:CI_OnlineResource>
        <gmd:linkage>
            <gmd:URL>http://ncwms.aodn.org.au/ncwms/wms</gmd:URL>
        </gmd:linkage>
        <gmd:protocol>
            <gco:CharacterString>OGC:WMS-1.1.1-http-get-map</gco:CharacterString>
        </gmd:protocol>
        <gmd:name>
            <gco:CharacterString>ACORN_ROT_QT/sea_water_velocity</gco:CharacterString>
        </gmd:name>
        <gmd:description>
            <gco:CharacterString>Rottnest shelf - sea water velocity (QC data)</gco:CharacterString>
        </gmd:description>
    </gmd:CI_OnlineResource>
</gmd:onLine>

We will need to configure both the linkage and name parameters. For linkage we will need to point to the abominationncwms endpoint on geoserver. For example: http://geoserver-123.aodn.org.au/geoserver/ncwms

The name element is built as follows:

WORKSPACE:URL_LAYER_NAME#TIME_FIELD,URL_FIELD/NETCDF_VARIABLE

A more concrete example for name is:

imos:acorn_hourly_avg_rot_qc_timeseries_url#time,file_url/sea_water_velocity

We can omit WORKSPACE if it's the default one. We can also omit TIME_FIELD/URL_FIELD if they are the default. The default for TIME_FIELD is time and the default for URL_FIELD is file_url. Then it becomes:

acorn_hourly_avg_rot_qc_timeseries_url/sea_water_velocity

And last, a real example of an online resource:

<gmd:onLine>
    <gmd:CI_OnlineResource>
        <gmd:linkage>
            <gmd:URL>http://geoserver-123.aodn.org.au/geoserver/ncwms</gmd:URL>
        </gmd:linkage>
        <gmd:protocol>
            <gco:CharacterString>IMOS:NCWMS--proto</gco:CharacterString>
        </gmd:protocol>
        <gmd:name>
            <gco:CharacterString>acorn_hourly_avg_rot_qc_timeseries_url/sea_water_velocity</gco:CharacterString>
        </gmd:name>
        <gmd:description>
            <gco:CharacterString>Rottnest shelf - sea water velocity (QC data)</gco:CharacterString>
        </gmd:description>
    </gmd:CI_OnlineResource>
</gmd:onLine>

A helpful note for developers

The JDBCFeatureReader class in GeoTools sends the query to the datastore to retrieve the Thredds netcdf URL. The result is stored in this.rs.getString(1). The query used is in runner.arg$2.

void runQuery(QueryRunner runner, Statement st) throws SQLException {
    this.callback.beforeQuery(st);
    try {
        this.rs = runner.run();
        this.callback.afterQuery(st);
    } catch (Exception var6) {
        this.callback.queryError(var6);

        try {
            this.close();
        } catch (IOException var5) {
            LOGGER.log(Level.FINE, "Failed to close the reader, moving on", var5);
        }

        throw new SQLException(var6);
    }
}

Geoserver configuration

Include an ncwms.xml file in the Geoserver data directory. This contains the location of the wfs service and the url to substitute for the Thredds server. NCWMS will not follow redirects between http:\\ and https:\\ so make sure you use the correct protocol for your Thredds server.

<ncwms>
  <wfsServer>http://localhost:8080/geoserver/ows</wfsServer>
  <urlSubstitution key="^">https://thredds.aodn.org.au/thredds/wms/</urlSubstitution>
</ncwms>