J. Meteor. Res.  2019, Vol. 33 Issue (4): 695-704   PDF    
http://dx.doi.org/10.1007/s13351-019-8208-9
The Chinese Meteorological Society
0

Article Information

HOU, Xueyan, Yang HAN, Xiuqing HU, et al., 2019.
Verification of Fengyun-3D MWTS and MWHS Calibration Accuracy Using GPS Radio Occultation Data . 2019.
J. Meteor. Res., 33(4): 695-704
http://dx.doi.org/10.1007/s13351-019-8208-9

Article History

Received January 28, 2019
in final form May 8, 2019
Verification of Fengyun-3D MWTS and MWHS Calibration Accuracy Using GPS Radio Occultation Data
Xueyan HOU1, Yang HAN2, Xiuqing HU3, Fuzhong WENG1     
1. State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, China Meteorological Administration, Beijing 100081;
2. Nanjing University of Information Science & Technology, Nanjing 210044;
3. National Satellite Meteorological Center, China Meteorological Administration, Beijing 100081
ABSTRACT: The newly launched Fengyun-3D (FY-3D) satellite carries microwave temperature sounder (MWTS) and microwave humidity sounder (MWHS), providing the global atmospheric temperature and humidity measurements. It is important to assess the in orbit performance of MWTS and MWHS and understand their calibration accuracy before using them in numerical weather prediction and many other applications such as hurricane monitoring. This study aims at quantifying the biases of MWTS and MWHS observations relative to the simulations from the collocated Global Positioning System (GPS) radio occultation (RO) data. Using the collocated FY-3C Global Navigation Satellite System Occultation Sounder (GNOS) RO data under clear-sky conditions as inputs to Community Radiative Transfer Model (CRTM), brightness temperatures and viewing angles are simulated for the upper level sounding channels of MWTS and MWHS. In order to obtain OB statistics under clear sky conditions, a cloud detection algorithm is developed by using the two MWTS channels with frequencies at 50.3 and 51.76 GHz and the two MWHS channels with frequencies centered at 89 and 150 GHz. The analysis shows that for the upper air sounding channels, the mean biases of the MWTS observations relative to the GPS RO simulations are negative for channels 5–9, with absolute values < 1 K, and positive for channels 4 and 10, with values < 0.5 K. For the MWHS observations, the mean biases in brightness temperature are negative for channels 2–6, with absolute values < 2.6 K and relatively small standard deviations. The mean biases are also negative for channels 11–13, with absolute values < 1.3 K, but with relatively large standard deviations. The biases of both MWTS and MWHS show scan-angle dependence and are asymmetrical across the scan line. The biases for the upper air MWTS and MWHS sounding channels are larger than those previously derived for the Advanced Technology Microwave Sounder.
Key words: satellites     Fengyun satellites     microwave sounding     cross-calibration     radio occultation    
1 Introduction

As a new generation of polar-orbiting meteorological satellite series in China, Fengyun-3 (FY-3) series consists of seven satellites. The fourth satellite, FY-3D, was successfully launched on 15 November 2017. Because its equatorial crossing time in the ascending node is around 1400 local time, FY-3D is referred as an afternoon orbit satellite. Its altitude, inclination angle, and orbital period are 836 km, 98.75°, and 101.49 min, respectively. The FY-3D satellite is a replacement for the FY-3B satellite, which operated in orbit for 8 yr and formed a constellation with the morning-orbit FY-3C satellite. There are 11 instruments onboard the FY-3D satellite, including a microwave temperature sounder (MWTS) and a microwave humidity sounder (MWHS). The MWTS has 13 channels in the 50–60-GHz oxygen band, similar to bands 3–15 of the Advanced Technology Microwave Sounder (ATMS) onboard NASA’s Suomi National Polar-orbiting Partnership satellite. It can provide the vertical structure of atmospheric temperatures from the earth’s surface to a height of 1 hPa. The MWHS has 15 channels, including 5 water vapor sounding channels at 183 GHz, similar to the ATMS, 8 additional sounding channels in the oxygen band near 118 GHz, and 2 window channels at 89 and 150 GHz. Both the MWTS and MWHS can profile the temperature and humidity of the atmosphere under nearly all weather conditions and have a vital role in improving global medium-range weather forecasts, as demonstrated by many numerical weather prediction centers.

Although the pre-launch thermal vacuum data from both the MWTS and MWHS can be used for accurate calibration, post-launch characterization is still required to fully understand their onboard performance. These instruments experienced strong vibrations during the launch of the satellite and are now working in a space environment very different from on-ground conditions (Weng et al., 2013). The atmospheric profiles obtained from Global Positioning System (GPS) radio occultation data are both stable and accurate and are valuable for assessing the performance of microwave sounders in orbit, as demonstrated by previous studies (Weng et al., 2013; Blackwell et al., 2014; Feltz et al., 2014a,b; Zou et al., 2014; Qin and Zou, 2016; Tradowsky et al., 2017). We therefore use atmospheric profiles from the GPS radio occultation data, which are accurate from the mid-troposphere to the stratosphere, to assess the in-orbit performance of the microwave sounders. Global Navigation Satellite System Occultation Sounders (GNOS) onboard the FY-3C satellite now provide atmospheric profiles with a similar quality to the Constellation Observing System for Meteorology, Ionosphere, and Climate (COSMIC) data (Liao et al., 2015, 2016). Specifically, the in-orbit performance of the FY-3D MWTS and MWHS upper air sounding channels can be verified by using the FY-3C GNOS data.

This paper is organized as follows. Section 2 briefly introduces the characteristics of the MWTS and MWHS instruments, the GNOS radio occultation data, and the Community Radiative Transfer Model (CRTM). Section 3 presents the collocation procedure between the FY-3C GNOS and FY-3D MWTS and MWHS instruments, and cloud algorithms are developed for quality control. Sections 4 and 5 discuss the biases in the MWTS and MWHS upper air sounding channels with respect to the simulations. Our summary and conclusions are presented in Section 6.

2 FY-3 instrument description

The FY-3D MWTS is a passive cross-track scanning microwave sounder capable of temperature sounding under nearly all weather conditions. It has 13 channels at frequencies ranging from 50 to 60 GHz (Table 1) with a scan swath of 2250 km. A total of 90 field of view (FOV) samples is taken along each scan line and the resolution is about 32 km near nadir. While FY-3D MWHS is used to detect both temperature and moisture under nearly all weather conditions, it has 15 higher frequencies, including bands at 183 GHz (for humidity) and 118 GHz (Table 2). It is a cross-track scanning sounder with a total of 98 FOVs and a scan swath of 2700 km. The nadir resolution is 32 km at 89 and 118 GHz and 16 km at 150 and 183 GHz. Figure 1 shows the vertical profiles of the weighting function (WF) for the MWTS and MWHS channels. Note that the 13 channels of the MWTS are similar to channels 3–15 of the ATMS and 7 of the MWHS channels are similar to the ATMS channels at 89 GHz (MWHS channel 1 to ATMS channel 16), 150 GHz (MWHS channel 10 to ATMS channel 17), and around 183 GHz (MWHS channels 11–15 to ATMS channels 18–22). The FY-3D MWHS also has 8 channels located in the oxygen band near 118.75 GHz. The two MWHS window channels 1 and 10 are primarily affected by radiation from the earth’s surface and by emission and scattering from ice phase clouds. The two channels can be used to physically retrieve the ice water path (Weng and Grody, 2000).

Table 1 Characteristics of the FY-3D MWTS channels
Channel No. Central frequency (GHz) Bandwidth (MHz) Polarization NEΔT (K)
1 50.30 180 V or H 1.20
2 51.76 400 V or H 0.75
3 52.8 400 V or H 0.75
4 53.596 400 V or H 0.75
5 54.40 400 V or H 0.75
6 54.94 400 V or H 0.75
7 55.50 330 V or H 0.75
8 57.290344 330 V or H 0.75
9 57.290344 ± 0.217 78 V or H 1.20
10 57.290344 ± 0.3222 ± 0.048 36 V or H 1.20
11 57.290344 ± 0.3222 ± 0.022 16 V or H 1.70
12 57.290344 ± 0.3222 ± 0.010 8 V or H 2.40
13 57.290344 ± 0.3222 ± 0.0045 3 V or H 3.60
Table 2 Characteristics of the FY-3D MWHS channels
Channel No. Central frequency (GHz) Bandwidth (MHz) Polarization NEΔT (K)
1 89 1500 V 1.0
2 118.75 ± 0.08 20 H 3.6
3 118.75 ± 0.2 100 H 2.0
4 118.75 ± 0.3 165 H 1.6
5 118.75 ± 0.8 200 H 1.6
6 118.75 ± 1.1 200 H 1.6
7 118.75 ± 2.5 200 H 1.6
8 118.75 ± 3.0 1000 H 1.0
9 118.75 ± 5.0 2000 H 1.0
10 150 1500 V 1.0
11 183.31 ± 1 500 H 1.0
12 183.31 ± 1.8 700 H 1.0
13 183.31 ± 3 1000 H 1.0
14 183.31 ± 4.5 2000 H 1.0
15 183.31 ± 7 2000 H 1.0
Figure 1 Weighting functions of the (a) MWTS and (b) MWHS onboard the FY-3D satellite

To produce an in-orbit truth for validation of the microwave sounders, we used atmospheric temperature and water vapor profiles derived from radio occultation data as inputs to a radiative transfer model and simulated the brightness temperatures at the satellite viewing positions. Radio occultation occurs when the receiver onboard a satellite in a low earth orbit tracks a Global Navigation Satellite System (GNSS) satellite in a higher orbit. It uses the radio signals emitted from the GNSS satellite to detect atmospheric parameters, including pressure, temperature, air density, and humidity. Radio occultation profiles are available from many sensors, e.g., COSMIC, COSMIC2, GNSS Receiver for Atmospheric Soundings (GRAS), KOrea-Multi-Purpose-SATellite-5 (KOMPSAT-5), and GNOS. The first GNOS was carried on the FY-3C satellite and used both the GPS and BeiDou navigation satellites. GNOS increased the number of transmitting sources and greatly improved the measured temperature and moisture profiles in the upper atmosphere (Tang et al., 2014). Atmospheric profiles from GPS radio occultation are both stable and accurate, especially for temperature profiles in the upper troposphere and stratosphere, making radio occultation data sources perfect for the calibration and cross-calibration of microwave sounders.

Like many other GPS radio occultation measurements, GNOS-GPS radio occultation measurements are very accurate and stable from the mid-troposphere to the lower stratosphere, roughly in the height range 5–25 km (Liao et al., 2015, 2016). The GPS radio occultation data are affected by several errors outside this range, such as super-refraction in the lower troposphere, a residual ionospheric effect, and high-altitude initialization above 25 km (Moradi et al., 2015). The upper level sounding channels of the MWTS and MWHS, the weighting functions of which reach a maximum within 5–25 km (20–600 hPa), are of most interest in this study because the GNOS-GPS radio occultation performs best in the height range 5–25 km and the atmospheric radiative transfer models are accurate in the upper troposphere and lower stratosphere (Zou et al., 2014). These are channels 4–10 for the MWTS and channels 2–6 and 11–13 for the MWHS. The following discussion focuses on these channels.

The CRTM developed and distributed by the US Joint Center for Satellite Data Assimilation was used as a forward operator to simulate the MWTS and MWHS brightness temperatures over global environments. The CRTM is a radiative transfer model based on sensor channels and is widely used to assimilate data from infrared and microwave satellites and other remote sensing applications (Weng, 2007; Liu et al., 2012). It includes modules that compute the satellite-measured thermal radiation from gaseous absorption, the absorption and scattering of radiation by aerosols and clouds, and the emission and reflection of radiation by the earth’s surface. The input to the CRTM includes atmospheric and surface state variables and other parameterized information. Forward, tangent-linear, adjoint, and K-matrix models are also available for many other applications. We simulated the brightness temperature for the FY-3D MWTS and MWHS sounding channels using the collocated FY-3C GNOS radio occultation data under clear sky conditions as inputs to the CRTM.

3 Collocation and cloud detection algorithms

GNOS radio occultation soundings collocated with the MWTS/MWHS measurements are selected to assess the accuracy of the MWTS/MWHS measurements. Figure 2 shows the collocation flowchart.

Figure 2 Flow chart for the collocation of MWTS/MWHS and GNOS radio occultation data

Figure 2 shows that the collocation criteria is set by a time difference of ≤ 3 h and a horizontal spatial separation < 50 km. If there are more than one MWTS/MWHS FOV measurement satisfying these collocation criteria, the measurement closest to the related GNOS radio occultation sounding is chosen and the other measurements are discarded. Because the simulations are performed over oceans, the surface wind field is required in the emissivity calculation. We used ECMWF reanalysis data as the input to the CRTM and found that the wind speed affected the bias primarily at the surface-sensitive channels of the MWTS and MWHS. For the upper air sounding channels, the O–B is not sensitive to the surface wind speed. Thus, we set a wind speed of 5 m s -1 at 10-m height. Because the surface emissivity influences vary greatly over land, the collocations of the MWTS/MWHS and GNOS are only constrained over the oceans. Also, the cloud radiative properties at microwave sounding channels vary significantly from clear sky conditions and thus a cloud liquid water index is applied to separate the data under clear sky conditions from the total measurements.

Compared to ATMS, the FY-3D MWTS and MWHS lack two low frequencies at 23.8 and 31.4 GHz and cannot be used to derive the cloud liquid water path. However, the cloud ice water path can be derived from the MWTS and MWHS frequencies at 89.0 and 157.0 GHz (Weng and Grody, 2000; Weng et al., 2003). Han et al. (2015) developed a cloud emission and scattering index using the approximately linear relationship between the paired MWTS oxygen band and the MWHS oxygen band. Inspired by this study, we develop a new cloud scattering index (CSI) using the two MWHS channels with frequencies centered around 89 and 150 GHz. The CSI is used for cloud detection in the MWHS and GNOS collocation and is derived as follows:

(1) Calculate the regression coefficients α and β through a logarithmic fit between the simulated brightness temperatures at 150 and 89 GHz under a clear sky. A logarithmic function is used to make the regression fit more linear:

$ \ln (290 - T_{150}^{{\rm{sim}}}) = \alpha + \beta \ln (290 - T_{89}^{{\rm{sim}}}), $ (1)

where $T_{89}^{{\rm{sim}}}$ and $T_{150}^{{\rm{sim}}}$ are the brightness temperature simulations using the numerical weather prediction model datasets as input to the CRTM. Here, we derived $\alpha = - 10.8231 \;{{\rm{ and}}} \;\; \beta = 3.3143$ .

(2) Calculate the brightness temperature at 150 GHz using the regression model:

$ T_{150}^{{\rm{reg}}} = 290 - \exp [\alpha + \beta \log (290 - T_{89}^{{\rm{obs}}})], $ (2)

where $T_{89}^{{\rm{obs}}}$ is the level 1b MWHS brightness temperature at 89 GHz.

(3) The new CSI is defined by

$ {\rm CSI} = T_{150}^{\rm reg} - T_{150}^{\rm obs}, $ (3)

where $T_{150}^{\rm obs}$ is the level 1b MWHS brightness temperatures at 150 GHz.

By using the new cloud index, the evolution process of Typhoon Maria from 8 to 11 July 2018 could be captured when the threshold for the CSI is set to 5.5 (Fig. 3).

Figure 3 Spatial distributions of CSI calculated from two MWHS window channels for Typhoon Maria (within red circles) from (a–d) 8 to 11 July 2018 when the minimum CSI index is set to 5.5

Similarly, a cloud detection algorithm based on the MWTS channels at 50.3 and 51.76 GHz is applied to separate the data under clear sky conditions over the oceans in the collocation of MWTS and GNOS. The clear sky regions are extracted from the total measurements by setting the threshold of CSI < 5 for MWTS and < 5.5 for MWHS. Figure 4 presents the spatial distribution of the MWTS/MWHS observations collocated with the GNOS radio occultation data under clear sky conditions over the oceans between 60°S and 60°N in July 2018. A total of 1036 (513) MWTS (MWHS) measurements are collocated with the GNOS radio occultation profiles in July 2018. Most of the collocated data are located in mid and high latitudes in both the Northern and Southern Hemispheres, especially for the MWTS. This is mainly due to the overlaps of the microwave instrument swath and the high density of GNOS data in high latitudes. The radio occultation profiles used to simulate the brightness temperature are more reliable because the GNOS radio occultation data are accurate in mid and high altitudes as a result of the limited multipath effects of moist atmospheres (Liao et al., 2016).

Figure 4 Distributions of collocated (a) MWTS and (b) MWHS data over the oceans between 60°S and 60°N in July 2018
4 Estimation of the MWTS biases using GNOS-GPS radio occultation data

The simulations of brightness temperature using the GNOS-GPS profiles as the input to the CRTM are denoted as BGPS. For cloud detection in the MWTS and GNOS collocation, the threshold of the CSI is set to < 5 to separate the data under clear sky conditions from the total measurements. Only the upper level sounding channels 4–10 for MWTS, the weighting functions of which peak within the height range 5–25 km, are of interest here, because the GNOS-GPS radio occultation measurements are more reliable at these heights. Figure 5 presents scatter plots of the brightness temperature from the MWTS observations and the GNOS-GPS radio occultation simulations of channels 4–10 for collocated data points under clear sky conditions over the oceans between 60°S and 60°N in July 2018. In general, the CRTM simulations with the GNOS-GPS radio occultation input profiles correlate fairly well with the MWTS observations. The correlations between the MWTS observations and the CRTM simulations are > 0.98 for channels 4–10. Figure 6 shows the mean biases and standard deviations of the differences between the MWTS observations (O) and the GNOS-GPS radio occultation simulations (BGPS) for channels 4–10. The global biases estimated by the mean differences (OBGPS) are negative for channels 5–9, with absolute values of < 1 K: channel 5 (−1.03), channel 6 (−0.98), channel 7 (−0.76), channel 8 (−0.52), and channel 9 (−0.27). The biases are positive for channels 4 and 10 with values of 0.22 and 0.42, respectively ( Fig. 6). The standard deviations of (OBGPS) for channels 4–10 are very close at 0.67, 0.71, 0.48, 0.67, 1.02, 1.06, and 1.16, respectively. By contrast, the biases and variances for channels 1–3 and 11–13 are much larger than those for channels 4–10, that is, the biases are positive with absolute values > 2 K and standard deviations > 3 K (figure omitted). The differences between the biases of channels 4–10 and the other channels are mainly due to the surface emission and cloud effects for channels 1–3, which are sensitive to the ground emissivity. Besides, the GNOS-GPS radio occultation measurements are not accurate at altitudes > 30 km, where channels 11–13 sounding. Overall, the mean bias between the MWTS observations and the simulations using GPS radio occultation profiles are < 1 K for channels 4–10. Compared with biases generally < 0.5 K for the ATMS temperature sounding channels ( Weng et al., 2013; Zou et al., 2014; Moradi et al., 2015), the accuracies of the MWTS upper air sounding channels (channels 4–10) are slightly larger than those of the ATMS.

Figure 5 Scatter plots of brightness temperature from the MWTS observations and the CRTM simulations with input from collocated GNOS-GPS radio occultation data for channels 4–10 under clear sky conditions over the ocean between 60°S and 60°N in July 2018
Figure 6 (a) Mean bias and (b) standard deviation of the FY-3D MWTS measurements using GPS radio occultation data as the input to the CRTM for channels 4–10

The variation of atmospheric inhomogeneity with scan angle cannot be accurately simulated in the CRTM for cross-track scanning radiometers such as the MWTS and MWHS (Zou et al., 2014). The effects of spacecraft radiation on brightness temperature also vary with the scan angle. Thus, it is important to quantify the biases of the cross-track scanning radiometer measurements that are dependent on the scan angle before using them in the assimilation of the radiance data. Figure 7 presents the biases and standard deviations of MWTS channels 4–10, which are dependent on the scan angle, estimated by using GNOS-GPS radio occultation data. An asymmetrical scan bias pattern is seen for MWTS channels 4–7, where the biases are more negative near the end of the MWTS scan lines. Among MWTS channels 4–10, the standard deviations of channels 8–10 are larger than those of channels 4–7.

Figure 7 Angular dependence of brightness temperature biases (solid curve) and standard deviations (dashed curve) between the MWTS observations and GPS radio occultation simulations (OBGPS) for channels 4–10
5 Estimations of MWHS biases using GNOS-GPS radio occultation data

For cloud detection in the MWHS and GNOS collocation, the threshold of the cloud liquid water index is set to < 5.5 to separate the data under clear sky conditions from the total measurements. The GNOS-GPS radio occultation measurements are not accurate at altitudes lower than 5 km where channels 1, 7–10, 14, and 15 sounding. Thus only the upper level sounding channels 2–6 and 11–13 for the MWHS, the weighting functions of which peak within the height range 5–25 km, are of interest in this study. Figure 8 presents the scatter plots of brightness temperature from the MWHS observations (O) and GNOS-GPS radio occultation simulations (BGPS) of channels 2–6 and 11–13 for collocated data points under clear sky conditions over the oceans between 60°S and 60°N in July 2018. In general, the CRTM simulations with GPS radio occultation input profiles correlate well with the MWHS observations. The correlations between the MWHS observations and the CRTM simulations are > 0.95 for channels 2–6. However, the correlations for channels 11–13 are smaller than those for channels 2–6.

Figure 8 As in Fig. 5, but for MWHS channels 2–6 and 11–13

Figure 9 shows the mean biases and standard deviations of the differences between the MWHS and GPS radio occultation simulations calculated for channels 2–6 and 11–13. The global biases estimated by the mean differences (OBGPS) are negative for channel 2 (−1.17), channel 3 (−2.57), channel 4 (−2.22), channel 5 (−1.63), channel 6 (−1.19), channel 11 (−1.18), channel 12 (−1.30), and channel 13 (−0.07). The standard deviations of (OBGPS) for channels 2–6 are 2.41 1.38, 1.17, 0.89, and 1.12, respectively. By contrast, the variances are much larger for channels 11–13 (4.98, 4.45, and 4.09, respectively) than for channels 2–6. Overall, the mean biases between the MWHS observations and the simulations using GPS radio occultation profiles range between −1.1 and −2.6 K for channels 2–6 and between 0 and −1.3 K for channels 11–13. The statistical uncertainties of the bias vary from 0.8–2.4 K for channels 2–6 to 4.0–5.0 K for channels 11–13. The biases are larger for the MWHS water vapor sounding channels than for the MWTS temperature channels. The similar situation was also reported for the ATMS by Moradi et al. (2015). They found that the difference between the radiosonde and ATMS brightness temperatures was < 0.5 K for the mid- and upper-tropospheric temperature sounding channels, but the difference for the water vapor channels was between 0.5 and 2.0 K. They attribute this to the fact that the radiosonde temperature measurements were more accurate than the sonde humidity measurements ( Moradi et al., 2015).

Figure 9 As in Fig. 6, but for MWHS channels 2–6 and 11–13

Similar to the MWTS measurements, the biases of the MWHS measurements dependent on the scan angle are also quantified. Figure 10 presents the scan-dependent biases and standard deviations of MWHS channels 2–6 and 11–13 estimated using the GNOS-GPS radio occultation data. Because there are only 513 MWHS measurements collocated with GNOS radio occultation profiles during July 2018 when the cloud detection index is set to < 5.5, the curves in Fig. 10 are not smooth. The biases are asymmetrical with respect to the nadir position for channels 2–6 and 11–13. The biases are slightly negative for channels 2 and 6 near the end of the MWHS scan lines, but very negative for channels 3–5 near the end of the scan lines. For channels 2–4 in particular, the biases at the left-most edge (i.e., FOV 1) are larger than those at the right-most edge (i.e., FOV 98), whereas for channels 5 and 6, the biases on the left-hand side of the FOV are smaller than those on the right-hand side of the FOV. Unlike channels 2–6, the biases of channels 11–13 show more symmetrical pattern across the scan line. The sounding channels 11–13 are more negatively biased near the nadir than at the edge of the scan line, and the standard deviations of the bias for channels 11–13 are much larger than those for channels 2–6 at all FOVs.

Figure 10 As in Fig. 7, but for MWHS channels 2–6 and 11–13
6 Summary and Conclusions

The FY-3D MWTS and MWHS provide us with more datasets for monitoring atmospheric temperature, water vapor, and cloud structures at different vertical levels using the dual oxygen absorption band near 60 and 118 GHz. It is important to quantify the post-launch in-orbit performance and to calibrate these instruments before applying them in numerical weather prediction models. Previous studies have suggested that GPS radio occultation has a high vertical resolution and small absolute uncertainty in the upper troposphere and lower stratosphere, making GPS radio occultation data suitable for calibrating and validating microwave sounder observations. GNOS-GPS radio occultation data and MWTS/MWHS sounding observations are collocated under clear sky conditions and over the oceans in July 2018. The CRTM is used to produce the GPS radio occultation simulations. A total of 1036 (513) MWTS (MWHS) measurements are collocated with GNOS radio occultation profiles in July 2018. The analysis shows that, for the upper air sounding channels, the mean biases of the MWTS observations to GPS radio occultation simulations are negative for channels 5–9, with absolute values < 1 K, and positive for channels 4 and 10, with absolute values < 0.5 K. For the MWHS, the mean brightness temperature biases are negative for channels 2–6, with absolute values < 2.6 K and relatively small standard deviations. The mean biases are also negative for channels 11–13, with absolute values < 1.3 K. The statistical uncertainty of the biases vary from 1–2 K for channels 2–6 to 4–5 K for channels 11–13. The distribution of the bias dependent on the scan angle for both the MWTS and MWHS are asymmetrical across the scan line for almost all the channels of interest. The biases for the upper air MWTS and MWHS sounding channels are generally larger than those of < 0.5 K previously derived for the ATMS.

References
Blackwell, W. J., R. Bishop, K. Cahoy, et al., 2014: Radiometer calibration using colocated GPS radio occultation measurements. IEEE Trans. Geosci. Remote Sens., 52, 6423–6433. DOI:10.1109/TGRS.2013.2296558
Feltz, M., R. Knuteson, S. Ackerman, et al., 2014a: Application of GPS radio occultation to the assessment of temperature profile retrievals from microwave and infrared sounders. Atmos. Meas. Tech., 7, 3751–3762. DOI:10.5194/amt-7-3751-2014
Feltz, M. L., R. O. Knuteson, H. E. Revercomb, et al., 2014b: A methodology for the validation of temperature profiles from hyperspectral infrared sounders using GPS radio occultation: Experience with AIRS and COSMIC. J. Geophys. Res. Atmos., 119, 1680–1691. DOI:10.1002/2013JD020853
Han, Y., X. L. Zou, and F. Z. Weng, 2015: Cloud and precipitation features of Super Typhoon Neoguri revealed from dual oxygen absorption band sounding instruments on board FengYun-3C satellite . Geophys. Res. Lett., 42, 916–924. DOI:10.1002/2014gl062753
Liao, M., P. Zhang, Y. M. Bi, et al., 2015: A preliminary estimation of the radio occultation products accuracy from the Fengyun-3C meteorological satellite . Acta Meteor. Sinica, 73, 1131–1140. DOI:10.11676/qxxb2015.072
Liao, M., P. Zhang, G. L. Yang, et al., 2016: Status of radio occultation sounding technology of FY-3C GNOS . Adv. Meteor. Sci. Technol., 6, 83–87. DOI:10.3969/j.issn.2095-1973.2016.01.012
Liu, Q. H., P. Van Delst, Y. Chen, et al., 2012: Community radiative transfer model for radiance assimilation and applications. Proc. IEEE IGARSS, 3700–3703. DOI:10.1109/IGARSS.2012.6350612
Moradi, I., R. R. Ferraro, P. Eriksson, et al., 2015: Intercalibration and validation of observations from ATMS and SAPHIR microwave sounders. IEEE Trans. Geosci. Remote Sens., 53, 5915–5925. DOI:10.1109/TGRS.2015.2427165
Qin, Z. K., and X. L. Zou, 2016: Uncertainty inFengyun-3C microwave humidity sounder measurements at 118 GHz with respect to simulations from GPS RO data . IEEE Trans. Geosci. Remote Sens., 54, 6907–6918. DOI:10.1109/TGRS.2016.2587878
Tang, Y. Q., J. S. Zhang, and J. S. Wang, 2014: FY-3 meteorological satellites and the applications. Chinese J. Space Sci., 34, 703–709. DOI:10.11728/cjss2014.05.703
Tradowsky, J. S., C. P. Burrows, S. B. Healy, et al., 2017: A new method to correct radiosonde temperature biases using radio occultation data. J. Appl. Meteor. Climatol., 56, 1643–1661. DOI:10.1175/JAMC-D-16-0136.1
Weng, F. Z., 2007: Advances in radiative transfer modeling in support of satellite data assimilation. J. Atmos. Sci., 64, 3799–3807. DOI:10.1175/2007JAS2112.1
Weng, F. Z., and N. C. Grody, 2000: Retrieval of ice cloud parameters using a microwave imaging radiometer. J. Atmos. Sci., 57, 1069–1081. DOI:10.1175/1520-0469(2000)057<1069:ROICPU>2.0.CO;2
Weng, F. Z., L. M. Zhao, R. R. Ferraro, et al., 2003: Advanced microwave sounding unit cloud and precipitation algorithms. Radio Sci., 38, 8068–8096. DOI:10.1029/2002rs002679
Weng, F. Z., X. L. Zou, N. H. Sun, et al., 2013: Calibration of Suomi national polar-orbiting partnership advanced technology microwave sounder. J. Geophys. Res. Atmos., 118, 11187–11200. DOI:10.1002/jgrd.50840
Zou, X. L., L. Lin, and F. Z. Weng, 2014: Absolute calibration of ATMS upper level temperature sounding channels using GPS RO observations. IEEE Trans. Geosci. Remote Sens., 52, 1397–1406. DOI:10.1109/TGRS.2013.2250981