Issue with Emodnetbio__aca_spp_19582016_L1 data query

[eblondel]

Hi @annakrystalli

I'm looking at ows4R maintenance, and realized that some integration test doesn't pass anymore (it did before). The request is: https://geo.vliz.be/geoserver/wcs?service=WCS&version=2.0.1&coverageId=Emodnetbio__aca_spp_19582016_L1&subset=Lat(58.18,58.24)&subset=Long(8.37,8.41)&subset=time(%222000-02-16T01:00:00%22)&format=image/tiff&request=GetCoverage

I get the following exception: Requested SEASON subset does not intersect the available values [1, 2, 3, 4]

Sounds an extra dimension has been added? but we can't really query it.

[annakrystalli]

Thanks for the heads up @eblondel . Any ideas @salvafern ?

[salvafern]

We have recently updated the VLIZ geo stack. Might be related

GeoServer 2.22.0 (from 2.11) https://geoserver.org/announcements/vulnerability/2022/11/18/geoserver-2-22-0-released.html

We will have a look

[eblondel]

Thanks, are you sure the datasource didn't change? Sounds there is a extra dimension for this data cube...

[salvafern]

No it didn't. This file has a custom dimension "season", but the default value is 1, no need to explicitly query.

But we may have found the issue: it turns out geoserver guesses a few common causes if it can't return data, and in this case it hallucinates a wrong cause.

The real problem: the long subset is smaller than 1 data element in the geotiff.

Look what happens when you change the longitude 8.37 to 8.36:

http://geo.vliz.be/geoserver/wcs?service=WCS&version=2.0.1&coverageId=Emodnetbio__aca_spp_19582016_L1&subset=Lat(58.18,58.24)&subset=Long(8.36,8.41)&subset=time(%222000-02-16T01:00:00%22)&format=image/tiff&request=GetCoverage

This delivers a 1 pixel response

[eblondel]

Weird.... Indeed this request was aimed to deliver one pixel only. And in parallel I do the same query through a WMS GetFeatureInfo to compare values.

[eblondel]

I fixed the test at https://github.com/eblondel/ows4R/commit/80d0fa6338ac93329f299cafa03cc58c5c67f45a