SymmCldImpact

This obsfunction computes the Symmetric Cloud Impact (SCI) parameter for cloud-affected infrared (IR) satellite radiances (can be potentially used for microwave channels as well). The SCI provides a continuous, physically consistent measure of scene cloudiness derived from both the observed and model background brightness temperatures (BTs), with values near zero indicating clear-sky scenes and increasing values reflecting progressively cloudier conditions.

The obsfunction supports two formulations, selected automatically based on the options provided:

  • Harnish et al. (2016): used when btlim is provided.

  • Okamoto et al. (2014): used when btlim is not provided.

This obsfunction is used as input to ObsErrorModelRamp to dynamically scale observation errors according to scene cloudiness, assigning larger errors to more heavily clouded scenes where representativeness errors are largest.

Harnisch et al. (2016) formulation

For each channel, the SCI is computed as:

\[ \begin{align}\begin{aligned}C_{\text{mod}} = \max(0, BT_{\text{lim}} - BT_{\text{background}})\\C_{\text{obs}} = \max(0, BT_{\text{lim}} - BT_{\text{observed,bc}})\\\text{SCI} = \frac{1}{2}(C_{\text{mod}} + C_{\text{obs}})\end{aligned}\end{align} \]

where \(BT_{\text{lim}}\) is a prescribed channel-specific clear-sky reference BT limit, \(BT_{\text{background}}\) is the model-simulated BT including the effects of clouds and precipitation as computed by the forward operator, and \(BT_{\text{observed,bc}}\) is the bias-corrected observed BT. The \(\max(0, \cdot)\) operator enforces the physical constraint that IR cloud impacts must be negative (cooling).

Okamoto et al. (2014) formulation

For each channel, the SCI is computed as:

\[ \begin{align}\begin{aligned}C_{\text{mod}} = |BT_{\text{clr}} - BT_{\text{background}}|\\C_{\text{obs}} = |BT_{\text{clr}} - BT_{\text{observed,bc}}|\\\text{SCI} = \frac{1}{2}(C_{\text{mod}} + C_{\text{obs}})\end{aligned}\end{align} \]

where \(BT_{\text{clr}}\) is the clear-sky brightness temperature from the forward operator diagnostics, \(BT_{\text{background}}\) is the all-sky model-simulated BT, and \(BT_{\text{observed,bc}}\) is the bias-corrected observed BT. Optionally, the SCI can be scaled by a sigmoid function of the observation-minus-background (O-B) departure to reduce the weight of heavily clouded scenes.

Options

channels (required) specifies the list of channels for which SCI will be computed.

btlim (optional) specifies the per-channel clear-sky reference BT limits \(BT_{\text{lim}}\) in Kelvin, provided as a list in the same order as channels. When provided, the Harnish et al. (2016) formulation is used; otherwise the Okamoto et al. (2014) formulation is used. These values are instrument- and channel-specific and should be derived empirically following Harnisch et al. (2016).

scale by omb (optional, default: false) when true, scales the SCI by a sigmoid function of the O-B departure. Only applies to the Okamoto et al. (2014) formulation.

sigmoid constant 1 (optional, default: 10.0) controls the slope of the sigmoid function. Only applies to the Okamoto et al. (2014) formulation.

sigmoid constant 2 (optional, default: 10.0) controls the 50-percent value of the sigmoid function. Only applies to the Okamoto et al. (2014) formulation.

Example: Situation-dependent observation error model (Harnish et al. 2016)

In this application, SymmCldImpact is used as the input variable to ObsErrorModelRamp, which maps the continuous SCI value to a piecewise linear observation error using channel-specific minimum and maximum error bounds (err0, err1) and SCI thresholds (x0, x1). The following example shows the configuration for the Himawari-9 AHI sensor.

- filter: Perform Action
  <<: *multiIterationFilter
  filter variables:
  - name: brightnessTemperature
    channels: &channels 8-10,13
  action:
    name: assign error
    _symmetric cld: &ahi_himawari9
      x0: [0.5, 0.5, 0.5, 0.5]
      x1: [14.5, 19.0, 22.5, 36.50]
      err0: [2.8, 2.9, 2.5, 0.95]
      err1: [13.1, 16.7, 19.07, 32.70]
    error function:
      name: ObsFunction/ObsErrorModelRamp
      channels: *channels
      options:
        <<: *ahi_himawari9
        channels: *channels
        xvar:
          name: ObsFunction/SymmCldImpact
          channels: *channels
          options:
            channels: *channels
            btlim: [234.27, 244.75, 255.98, 292.99]

References

Harnisch, F., M. Weissmann, and Á. Periáñez, 2016: Error model for the assimilation of cloud-affected infrared satellite observations in an ensemble data assimilation system. Quart. J. Roy. Meteor. Soc., 142, 1797–1808, https://doi.org/10.1002/qj.2776.

Okamoto, K., McNally, A.P. and Bell, W., 2014: Progress towards the assimilation of all-sky infrared radiances: an evaluation of cloud effects. Quart. J. Roy. Meteor. Soc., 140, 1603–1614, https://doi.org/10.1002/qj.2242.