Most operational Numerical Weather Prediction (NWP) centers have the capability to assimilate satellite radiance data directly using variational Data Assimilation Systems (DAS). The direct radiance assimilation method has led to significant improvements in forecast skill, with the NOAA AMSU-A instrument being the pathfinder in the radiance assimilation efforts. Necessary components of the direct assimilation method are the routine monitoring of the departures from the observed radiances (OB) and those computed using radiative transfer and a short term forecast from the NWP model, termed the background (BK), and bias correction schemes. For temperature sounding radiances, the measurement uncertainty requirements for forecast skill improvement are very demanding. Uncertainties of the bias corrected departures must be of the order 0.2 K (MW) to 0.4 K (IR) or less in order to improve the analysis and skill of the subsequent NWP forecast. This uncertainty requirement is often at or below the sensor calibration error and noise level (NEΔT), so that the routine monitoring of the patterns of the global departures can often detect problems with sensor data such as increased channel noise, channel failures, calibration drift, field of view intrusion and other calibration anomalies. For examples, the NWP DAS-based radiance monitoring capability employed for the DMSP SSMIS Cal/Val efforts was vital in understanding the spatial and temporal departure patterns, unveiling the underlying causes of the observed SSMIS calibration anomalies. Results of the success of the NWP DAS in detecting calibration anomalies with the SSMIS and future JPSS and DWSS sensors will be presented.