Evaluation of Statistical Retrieval Errors for Ground Based Microwave Radiometer Measurements

Dual frequency ground based microwave radiometers (MWR) are widely used to measure precipitable water vapor (PWV) and integrated cloud liquid water (ICL). In the statistical retrieval method, where the relationship between measured brightness temperature (Tb) and opacity ( ) is used, an accurate statistical retrieval algorithm is required. These algorithms include a set of input parameters whose values vary according to location, season, and weather conditions, [1]. One of these input parameters, the mean radiating temperature of the atmosphere (Tmr), is assumed to be known, [2, 3]. The importance of the reduction of the MWR measurement error has increased because are often used as references for other instruments, such as the Global Position Systems (GPS). Consequently different methods of specifying Tmr have been established [2]. The quality of the radiometer retrievals is usually established by comparing it with radiosonde measurements, while the most frequently used estimation of the Tmr value is a monthly mean value calculated from the climatological data. For different MWR sites and seasons the contribution of the Tmr errors in the statistical retrievals will be different. The main objective of this study is to exam the PWV and ICL MWR retrieval errors due to Tmr uncertainty across Canada during the year. The results should give us awareness on when it is necessary to consider an alternative way of estimating Tmr besides the common climatological approach. This is especially the case when the MWR site is not close enough to one of the climatological radiosonde locations. Also the results can be used to estimate how much the MWR measurements could be improved with a different substitution of Tmr.