NCAR / pyDARTdiags

Python observation space diagnostics for the Data Assimilation Research Testbed (DART)
Apache License 2.0
2 stars 1 forks source link

Feat request: multi component observations #10

Open hkershaw-brown opened 1 month ago

hkershaw-brown commented 1 month ago

Observations types that are components of a measurement need treating differently to regular obs types.

Example is U_WIND, V_WIND combined into horizontal wind

https://github.com/NCAR/DART/blob/c8ddfac035a39bbb24451ced3b0aa5410ac62494/assimilation_code/programs/obs_diag/threed_sphere/obs_diag.f90#L2940-L2942

https://github.com/NCAR/DART/blob/c8ddfac035a39bbb24451ced3b0aa5410ac62494/assimilation_code/programs/obs_diag/threed_sphere/obs_diag.f90#L2993-L3021

hkershaw-brown commented 1 month ago

defined in yaml?

acars_horizontal_wind:
  description: ACARS-derived Horizontal wind speed
  components:
    - acars_u_wind_component
    - acars_v_wind_component

sat_horizontal_wind:
  description: Satellite-derived horizontal wind speed
  components:
    - sat_u_wind_component
    - sat_v_wind_component
hkershaw-brown commented 1 month ago

note on duplicate observations. Will need to deal with these when finding matching component types.

number of each type of obs

one

duplicates where lat, lon, vertical, time & observation value matches. I feel like these are actual duplicates.

two

This is matches but not for observation value. Obs in the same place and time, same type, different observation value. 12544 out of 107212 sat_u_wind (10% ish). Different measurements of the same thing?

three

This is looking at the SAT_U_WIND_COMPENT dups. Sorted by latitude so you can see the dups a bit easier. four