The accuracy of the sea surface temperature measured by the MOS-1 MSR has been degraded by both the rotated polarization due to antenna scanning and antenna sidelobes. This paper describes data correction methods for them. First, a simple function has been introduced to correct the polarization error due to antenna scanning. Second, an inverse filtering has been used to suppress the error due to antenna sidelobes in both simulation and real data obtained by the MOS-1 MSR.
The MOS-1 MSR is a microwave sensor that rotates the main dish antenna with mechanically conical scanning. At the same time the observing polarization is also rotated by the mechanical scanning. Because sea surface brightness temperature depends on polarization, the rotation of the polarization causes measurement errors. The amount of the errors are estimated about 3 K at 23 GHz and 4 K at 31 GHz, respectively.
The sea surface brightness temperature varies as nearly cosine or sine curves according to the rotated polarization. Then we made simple equations for error correction. Using the equations, the error correction for both 23 GHz and 31 GHz has been done and the errors after correction became within ±0.1 K.
In order to evaluate the influence of antenna sidelobes on observed brightness temperature, we have calculated the brightness temperature on a simulation image. The results are as follows.
If the main beam of the antenna is directed to sea and the antenna sidelobes look at ground, the sea surface brightness temperature are observed higher than true data because of the influence of the higher temperature of the ground. On the other hand if the main beam and the sidelobes are directed to land and sea area respectively, the land brightness temperature is observed lower than true data. The amount of the errors due to the antenna sidelobes are about 10 K higher at sea and 20 K lower at land.
An inverse filtering method for the error correction has been applied to both simulation images and real data of the MOS-1 MSR. By this method we can correct more than 5 K and 10 K of the brightness temperature at sea surface and land, respectively.
Finally, we discussed the effect of pixel fractionization for smoothing.