Analysis of Nonstationary Radiometer Gain Using Ensemble Detection

Area of Science:

• Radiometry
• Signal Processing
• Statistical Modeling

Background:

• Radiometer gain is often assumed stationary but is typically a nonstationary random process.
• Direct observation of radiometer gain is not feasible, making analysis of its nonstationary properties difficult.
• Radiometer calibration is essential for characterizing gain, especially its dynamic behavior.

Purpose of the Study:

• To develop an ensemble detection algorithm for analytically characterizing unknown radiometer gain.
• To model both stationary (Gaussian, AR(1)) and nonstationary radiometer gain processes.
• To retrieve parameters of equivalent Gaussian or AR(1) models for nonstationary gain.

Main Methods:

• Utilizing time series of postgain voltages to form an ensemble set.
• Applying the ensemble detection algorithm to analytically characterize gain.
• Modeling gain as Gaussian (strictly stationary) or 1st order autoregressive (AR(1), weakly stationary) processes.

Main Results:

• The ensemble detection algorithm successfully characterizes Gaussian and AR(1) gain models.
• Nonstationary gain can be represented by equivalent Gaussian or AR(1) models.
• Retrieved parameters for nonstationary gain models are highly dependent on calibration setup and timing.

Conclusions:

• The ensemble detection algorithm provides a robust method for radiometer gain characterization.
• This approach enables the modeling of both stationary and nonstationary radiometer gain.
• Understanding the influence of calibration parameters is crucial for accurate nonstationary gain modeling.