Meteorological satellites have become an irreplaceable weather and ocean observing tool in China.Recently,much progress has been made in direct assimilation of satellite radiance measurements in numerical weather prediction(NWP).The Microwave Atmospheric Humidity Sounder(MWHS),carried onboard the Chinese new generation of polar-orbiting weather satellites,the Feng-Yun(FY-3) series,provides passive measurements of the radiation emitted from Earth's surface and throughout the atmosphere.MWHS is similar to AMSU-B and the Microwave Humidity Sounder of NOAA,with a primary purpose of moisture sounding in cloudy regions,designed to obtain information on atmospheric humidity profiles,water vapor,rainfall,cloud liquid water,and so on.
Several operational NWP centers currently rely on a variational analysis system to define the initial state for their NWP models.From the point of view of variational assimilation,errors from observations and models should have normal and unbiased distributions.However,statistical errors of radiance between observation and prediction from the NWP model are usually not random but systemic.MWHS is a cross-track scanning microwave radiometer.A feature of a cross-track sounder is that the measurements vary with scan angle because of the change in the optical path length between Earth and the satellite.This feature is called the limb effect.Cross-track scan bias correction is a key part of an NWP assimilation system.In order to use radiances from MWHS,biases between the observed radiances and those simulated from the model first guess must be corrected.The bias of global FY-3A MWHS observed brightness temperature(T_O) and simulated brightness temperature(T_B) by the Community Radiative Transfer Model(CRTM) based on 6 h forecast fields of the NCEP GFS from 28 August to 6 September 2010 are calculated in this study.
The statistical features of observation increment(T_O minus T_B) in channels 3,4 and 5 of MWHS are analyzed.The scan angle bias of channels 3-5 varying along scan position reveal that the scan angle bias generally increases with the scan position far away from the nadir,but does not increase in a strictly progressive manner.Large positive scan biases are present at large scan angles at all latitudes for channels 3 and 4 on both sides of the scan lines.Relatively large negative scan biases for channel 5 are seen at most scan angles and are highly asymmetric on both sides of nadir.The closer to the surface the peak height of the channel weighting functions,the more serious the asymmetry.All of this further verifies the importance of scan angle bias correction.
Analysis of radiance data shows a significant residual scan bias that depends strongly on latitude for these channels.The scheme applies a latitudinally dependent scan correction to take this into account.In this study,the correction coefficients are calculated within every 5° latitudinal band for each scan position.Moreover,nadir biases are subtracted.These biases will be subtracted from T_O minus-values in the formulation of data assimilation.Therefore,a constant scan bias for each channel needs be removed before MWHS radiance assimilation.The statistical distribution of brightness temperature bias(the departures of T_O minus T_B) after correction locates mostly in the vicinity of zero,with a more Gaussian distribution.The bias and standard deviations of T_O minus T_B differences between observations and model simulations are reduced after bias correction.
The scan-angle and latitudinal-dependence of the observation increment for the MWHS channels 3 to 5 within 5° latitudinal bands are given quantitatively.This scan-bias correction coefficient can be used directly for MWHS data assimilation.This work can be extended to the FY-3 MWHS series.The methodology will be incorporated into the Global/Regional Assimilation and Prediction System,operational NWP data assimilation systems in China,or other NWP modeling systems.