Wednesday: Experimenting with Observations

Today we will explore making changes to the yaml files that control the observations that are assimilated.

For this practical, you can choose your favorite data assimilation algorithm from yesterday’s practical, and build on top of its configuration file. You can also experiment with using different assimilation algorithms with different observations.

For all of the exercises 1–4 plot and compare increments.

Exercise 1: Run assimilation including radiosonde and aircraft wind observations. Tip: the variables that are to be simulated in H(x) are specified in ObsSpace.simulate section of config file. Variable names for winds: eastward_wind, northward_wind.

Exercise 2: Run assimilation including AMSU-A radiance observations, using CRTM to calculate H(x). Tip: you’ll have to create a new section for this ObsType in yaml. CRTM ObsOperator also has some additional parameters:

n_Absorbers: 3             # 2: use H2O and O3, 3: use H20, O3 and CO2
n_Clouds: 2                # 0: no clouds, 1: water; 2: water and ice clouds
n_Aerosols: 0
Sensor_ID: amsua_n19
EndianType: little_endian
CoefficientPath: Data/

Tip: to specify which channels to simulate, use the following syntax in ObsSpace.simulate:

ObsSpace:
  simulate:
    variables: [brightness_temperature]
    channels: 1-5

Note: channels can use syntax like: “1, 4, 6-8, 10, 22-67”

Exercise 3: Run assimilation adding background QC to AMSU-A. Compare how many observations pass the QC with various background check parameters.

Exercise 4: Experiment with using different QCs and run assimilation.

Exercise 5 (very optional): Improve CRTM TL/AD by adding moisture variable to the TL/AD. Note: the changes would be in file ufo/src/ufo/crtm/ufo_radiancecrtm_tlad_mod.F90. The variable name for moisture variable is “humidity_mixing_ratio”, constant with name var_mixr is defined in the ufo/src/ufo/ufo_variables_mod.F90.