The multi-channel microwave radiometer （RAD） on-aboard SZ-4 spacecraft

was launched at the end of 2002

the first space-borne microwave radiometer in China. Due to lack of in-orbit calibration

the measurements have to be quantatively evaluated before applications. In the paper

we put forward a vicarious external calibration method for solving the problem. The method is to re-calibrate all RAD channels using the two type of targets with known microwave radiation characteristics on the surface of the Earth. The first target was global open ocean. We collected synchro 252 sea surface truths with RAD from NDBC buoys combined with radiosonde profiles from nearby islands

as well as geophysical products of TMI （TRMM Microwave imager）

including sea surface temperature

wind speed

water vapor contents

and liquid water contents. After calculation using our radiative transfer equation developed for RAD

we got a synchro oceanic parameter database for RAD brightness temperature calibration. The warm end we chose was Amazon Rain Forest. We obtained 8 profiles for two stations within 50 km from the RAD nearest overpasses. The brightness temperatures were derived from applying a simpled emissivity model and atmospheric absorption model. We compared the re-calibrated brightness temperatures of RAD with those from SSM/I within 0.5 degree of latitude and longtitude and within 2 hour of RAD overpasses above global open ocean. We found there exited a good consistency between those equivalent channales of RAD and SSM/I

i.e.

19.35GHz and 37.0 GHz. The correlation coefficient of the channales are 74.4% and 60.1% repectively for vertical and horizontal polarization of 19.35GHz. The poor correlation in 37GHz channels may contribute to the path differences of them since the incidence angles of RAD is 42.6 degree

while SSM/I 53.3 degree. We also compared brightness temperatures of RAD with measurements from a ground based microwave radiometer employed in the Nanhai experiment carried in February

2003. The largest discrepancies also showed in those of 37GHz. In general

due to the insuitable location of ground test and poor sync time

and only few provable measurements for comparisons with the temperatures of RAD were made.