[
https://issues.apache.org/jira/browse/SDAP-50?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Joseph Jacob updated SDAP-50:
-----------------------------
Resolution: Fixed
Status: Done (was: In Progress)
This fixes several problems with the Spark Hovmoller Map algorithms:
(1) The Hovmoller calculation was averaging latitudes or longitudes within
individual tiles. However, depending on the search domain, unique lats/lons can
be in multiple tiles.
(2) A time stamp could appear multiple times in the JSON response
(3) Lack of cos(lat) weighting when averaging over latitude
Note that the non-Spark Hovmoller algorithms have these same problems, but are
considered deprecated and have not been fixed in this pull request.
> Fix incorrect Hovmoller Implementation
> --------------------------------------
>
> Key: SDAP-50
> URL: https://issues.apache.org/jira/browse/SDAP-50
> Project: Apache Science Data Analytics Platform
> Issue Type: Bug
> Components: nexus
> Reporter: Frank Greguska
> Assignee: Joseph Jacob
> Priority: Major
> Fix For: 0.1
>
>
> Currently the Hovmoller calculation is averaging latitudes or longitudes
> within individual tiles. However, depending on the search domain, unique
> lats/lons can be in multiple tiles.
> The algorithm needs to be updated to average all unique (time, lat) or (time,
> lon) values regardless of tiles.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
