We are submitting collection metadata to CMR in DIF10 form. In the Data Resolution field there is an entry for Temporal Resolution with a definition that it is the frequency of data sampled. For Level 3 and Level 4 data this seems pretty straightforward but may present some challenges for Level 1 and Level 2 data, particularly if we apply the definition in terms of linking it with ground spatial resolution or repeat cycle for satellite data.  For example, for OCO-2 Level 2 records we are preparing the temporal resolution is listed as 16 days (repeat cycle) and for Aqua AIRS Level 2 records this is listed as Twice per day; day and night; Orbital repeat cycle 16 days.

Are we mis-characterizing or over complicating this attribute? Is this as simple as the temporal extent of the individual data granules in the collection (e.g., data granularity)?

Our goal here is to submit accurate metadata records consistent with other holdings in CMR.

Any feedback is appreciated.

  • No labels

10 Comments

  1. user-7b92a

    In general we're for more accurate and complete metadata even if the CMR isn't able to take advantage of it. End users downloading a DIF10 version of the data would see the data that you sent. I'll forward this along to GCMD to see if they have any recommendations regarding this field.

    Scott Ritz or Thomas Cherry, are you able to answer Bruce's question about DIF10 temporal data?

  2. Bruce Vollmer

    Thanks Jason. I look forward to GCMD responses regarding how other records are populated. However, I don't want to restrict this discussion to only DIF10 records as we will be using MMT at some point in the future where the native form is UMM-C. Upon checking UMM-C documents I do not see temporal resolution explicitly defined. In terms of long term curation I would like today's metadata records to be consistent with records submitted in the future from MMT.

    1. user-7b92a

      user-b9cd4 and Erich Reiter, do you have some insight into the Temporal Resolution field in the UMM to answer Bruce's question or know who the best person is? Is this something Jon Pals may be able to answer?

  3. Bruce Vollmer

    FWIW, I extracted a couple of records from CMR via the API (umm/json) that were submitted as DIF10 that contained  temporal resolution. The contents of the temporal resolution field in the DIF10 were not in the umm/json extraction.

  4. Scott Ritz

    Bruce, I don't think all field are translated to umm/Json.

  5. Scott Ritz

    Bruce, I am going to extract a dump of all the Temporal Resolution values currently in the CMR from all providers/formats and analyze how DAACs are using the field.  I will post the results in this discussion.    

    BV: Are we mis-characterizing or over complicating this attribute?

    SR: I don't think you are mis-characterizing or over-complicating this attribute. You are following the DIF-10 guidelines and ensuring the value is accurate and complete.

    BV: Is this as simple as the temporal extent of the individual data granules in the collection (e.g., data granularity)?

    SR: Yes

  6. Christopher Lynnes

    Just spitballing here, but maybe temporal resolution is not useful / meaningful for L1/L2 data and should just be left blank in those cases?

  7. Bruce Vollmer

    I was hoping to get feedback from metadata experts.

    It looks as though temporal resolution is unique to the docBuilder/DIF form of collection metadata. Other forms (ECHO10, ISO) have the all-encompassing "keyword" to represent this field from the DIF record. When loading a DIF record into the Metadata Manage Tool (MMT) this field is non-existent.

    While there are many GES DISC records in CMR currently that have this field populated we are likely to de-emphasize this field as collection level records are curated in CMR.

  8. Scott Ritz

    Bruce, I apologize for not giving you feedback sooner.  Temporal Keywords including "DIF-9/Temporal_Resolution" are indexed for searching in the CMR.  Since as you mentioned ECHO-10 and ISO capture Temporal_Resolution in an all encompassing "keyword" field it is difficult to know if resolution is indicated or it is just a general temporal related keyword.  Although these values may not be that useful from a search point of view I feel that they are important from an informational perspective.  Example: "Twice per day; day and night; Orbital repeat cycle 16 day" in AIRIBRAD_NRT_005 is important information for determining if this is the right data set for my research.  There are currently 3,013 unique values for Temporal_Resolution specifically in the CMR.  I will attach a dump of the values. 

  9. Scott Ritz

    Temporal_Resolution_Dump.xmlDump of all Temporal_Resolution Values in CMR.