Opened 3 years ago

Closed 3 years ago

#687 closed defect (fixed)

Wrong timestamp to time numeric conversion when vector is not time versor

Reported by: pcampalani Owned by: pcampalani
Priority: critical Milestone: 9.0.x
Component: petascope Version: development
Keywords: timestamp conversion vector Cc:
Complexity: Easy

Description

When the offset vector for time is not of norm 1, there is a bug in the conversion from ISO timestamp to time numeric coordinates:

subsetGridIndexes = dbMeta.getIndexesFromIrregularRectilinearAxis(
                      covMeta.getCoverageName(),
                      covMeta.getDomainIndexByName(axisName),
-->                   (new BigDecimal(numLo)).subtract(domMin),
-->                   (new BigDecimal(numHi)).subtract(domMin),
                      indexMin, indexMax);

numLo/numHi are time coordinates normalized by the offset vector in time, while domMin (the origin usually) is the absolute time coordinate within the specified CRS, not normalized by the offset vectors.

Change History (4)

comment:1 Changed 3 years ago by pcampalani

  • Status changed from new to accepted

comment:2 Changed 3 years ago by pcampalani

Patch upcoming (working and tested).
I want to add a new irr_cube_2 to our systemtest before applying the patch to have the following characteristics:

  • space-time CRS (like eobstest)
  • irregular time spacing
  • time origin different from CRS datum epoch
  • time offset vector different from unit vector of norm 1.

comment:3 Changed 3 years ago by dmisev

Can we close this?

comment:4 Changed 3 years ago by pcampalani

  • Resolution set to fixed
  • Status changed from accepted to closed

Bug fixed in changeset:dd33a59.

A new WC*S test datasets has been added too, called irr_cube_2: an irregular time series of 2-band georectified images, see this folder] and the description in the README.

Note: See TracTickets for help on using tickets.