MetOfficeRelativeHumidityCorrection¶
There are differences in values of relative humidity H(x) produced by the UM interface and the Met Office OPS system. The differences are caused by the order in which (1) RH is computed from specific humidity (2) temporal and spatial interpolation are performed. The computation of relative humidity from specific humidity is nonlinear, which can lead to differences in H(x) of up to 5%.
This ObsFunction computes two values of relative humidity H(x). The first reproduces what occurs in the UM interface, i.e. it vertically interpolates the relative humidity GeoVaL at each location. The second reproduces what occurs in OPS, i.e. it computes relative humidity from GeoVaLs of specific humidity, temperature, and pressure and then performs vertical interpolation. The output of the ObsFunction is the difference between the two interpolated H(x) values.
The H(x) difference can be added to observed relative humidity (using an Arithmetic
ObsFunction) prior to running QC filters that rely on relative humidity O-B such as the
Background Check. After those filters have run, the H(x) difference can be subtracted back off.
Parameters¶
model_pressure
: Name of model pressure.model_specific_humidity
: Name of model specific humidity.model_relative_humidity
: Name of model relative humidity.model_temperature
: Name of model temperature.observed_pressure
: Name of observed pressure.capsupersat
: [optional, defaultfalse
] Cap relative humidity at 100%. Defaultfalse
.
Example yaml¶
The following yaml block shows how the ObsFunction can be used.
- filter: Variable Assignment
assignments:
- name: HofXCorrection/relative_humidity
type: float
function:
name: ObsFunction/MetOfficeRelativeHumidityCorrection
options:
model pressure: air_pressure
model specific humidity: specific_humidity
model relative humidity: relative_humidity
model temperature: air_temperature
observed pressure: MetaData/pressure
The correction variable, HofXCorrection/relative_humidity
, can be added to the RH observation
values prior to using other filters that use RH O-B.
(Adding the correction to the observation values is equivalent to subtracting it from the background values.)
The correction should be subtracted off afterwards.