An Adaptive Calibration Window for Noise Reduction of Satellite Microwave Radiometers

Over the years, a fixed window for smoothing radiometer cold-space and warm-load counts and processing brightness temperature in calibration has been used for all microwave sounders at European Organization for the Exploitation of Meteorological Satellites (EUMETSAT) and NOAA. Although this practice is based on ground tests and legacy satellites, it remains unclear if this empirical parameter is optimal for in-orbit radiometers, as the space environment is different from the ground and radiometers may drift. We found that the fixed window is not optimal and leads to large noise. We have developed an adaptive window that accommodates channel differences and temporal changes in hardware. Our method has reduced noise by as much as 50% for 183 GHz channels of MetOp-C microwave humidity sounder (MHS). We observed temporal jumps and shifts in counts, gain, and noise of 89 and 190 GHz, and accordingly, the adaptive window can adjust to reduce such an impact. Further analyses reveal that
$1/f$
noise plays an important role for determining the adaptive window.
$1/f$
noise is nonstationary and gives rise to the fluctuation of counts and gain. As a result, for channels with large
$1/f$
noise a short window should be used to mitigate the fluctuation. Our study suggests that an adaptive method has advantages over the fixed method for considering channel differences and time-varying noise.