metadata: sla calculation details as attribute (or new variable)

[andreustaylor]

rads2nc: record SLA details

'sla' construction details are only in input xml, not output netcdf

It would be beneficial to include a summary of SLA calculation choices within each netcdf file.

• make outputs more self contained
• allow quality cross-checks

complication

1) When using a hierarchy of xml inputs, how can we be confident exactly which fields were used?

• default conf/rads.xml
• local_overrides.xml 2) When using aliases in the xml, how can we be confident where the alias was pointing? eg
• alias name="tide_ocean" field="12"
• did we actually use 1222 or 1224? etc

ideas

(A) 'data' xml string as an an attribute

alt range SUB dry_tropo SUB wet_tropo SUB iono SUB dac SUB tide_solid SUB tide_ocean SUB tide_load SUB
        tide_pole SUB ssb SUB mss SUB ref_frame_offset SUB

• form like that above doesnt help with clarification of aliases
• better if each tag is translated into long name as applied:

  alt_gdrf range_ka SUB dry_tropo_ecmwf SUB ...

  (B) field code summation string as an attribute

  Add a summary of low level field codes used to netcdf attribute 'source' or a new attribute like 'rads_source_fields'.

• "field - field - field ..."
• or reverse polish " field field SUB field feld SUB ...etc"
• "0404-0601-0701- ..."

(C) field codes and factors in two new variables

This is a solution we had in place using rads3 low level functions

dimensions:
    options = 99 ;
variables:
    int options(options) ;
        options:long_name = "RADS options set by input XML" ;
        options:units = "None" ;
    short factors(options) ;
        factors:long_name = "RADS factors set by input XML" ;
        factors:units = "None" ;
 options = 0, 1, 1, 1, 25, -1, 1, 1, 1, 3, 6, 1, 24, 24, 1, 2, 18, 1, 1, -1, 
    2, 1, -5, -2, 0, -16, 1, 0, 2, 2, -2, 0, 0, -1, 0, 0, -1, -1, 1, -1, -1, 
    -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ;

 factors = 1, 0, 0, 0, 1, 0, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, 
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, -1, 0, 0, 0, 
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 
    0, 0, 0, 0, 0, 0, 0, 0, 0 ;

[andreustaylor]

working solution

https://github.com/andreustaylor/rads/commit/9b8556392d9a18ba8d8c412e93eef4c483c367f5

[andreustaylor]

update solution to use attribute 'method'

https://github.com/andreustaylor/rads/commit/aed44a5de398884c4c37c13b580d582c4f979504

[andreustaylor]

method attribute ambiguity with alias corrections

Flaw in the method attribute implementation on my fork. When the correction is an alias (eg dac) the method string cant identify exactly which flavor/s of the correction were actually accessed. This is shame as the intention was to encode the exact corrections used in the output file.

In the process, created a variant attribute that records the corrections by rads3 field codes rather than long names. No difference for the ambiguity issue though.

Suspect this cant be resolved for the general case without going beyond simple netcdf attributes.

Latest fork version: https://github.com/andreustaylor/rads/commit/c6bca5c839186253d3420d8dcfa47100b51da3a6

more details for a specific correction

• dac is an alias for list of possible correction sources
• resulting method attribute string is a comma-separated list that reflects the content of the rads.xml file
• <alias name="dac" field="10">inv_bar_mog2d_era inv_bar_mog2d inv_bar_mok2d</alias>
• MOG2D dynamic atmospheric correction (ERA Interim forcing), MOG2D dynamic atmospheric correction, MOK2D inverse barometer correction
• doesn't distinguish between which of these 3 sources was actually applied for a particular output file

sla:method = "RPN=|CNES GDR-F, CNES GDR-E orbital altitude|Ku-band range corrected for instr. effects|SUB|ECMWF dry tropospheric correction|SUB|radiometer, ECMWF wet tropospheric correction|SUB|smoothed dual-frequency, JPL GIM, NIC09 ionospheric correction|SUB|solid earth tide|SUB|FES2014b ocean tide|SUB|FES2014a load tide|SUB|pole tide|SUB|Tran et al. 2012 non-parametric sea state bias|SUB|DTU15 mean sea surface height|SUB|reference frame offset|SUB|MOG2D dynamic atmospheric correction (ERA Interim forcing), MOG2D dynamic atmospheric correction, MOK2D inverse barometer correction|SUB" ;

sla:method_fields = "|   4:  -1|   6:  -1|SUB|   7:  -1|SUB|   8:  -1|SUB|   9:  -1|SUB|1101:  11|SUB|  12:  -1|SUB|  13:  -1|SUB|1401:  14|SUB|  15:  -1|SUB|  16:  -1|SUB|3801:  38|SUB|  10:  -1|SUB" ;