• Andersson, E., , Pailleux J. , , Thepaut J.-N. , , Eyre J. R. , , McNally A. P. , , Kelly G. A. , , and Courtier P. , 1994: Use of cloud-cleared radiances in three-/four-dimensional variational data assimilation. Quart. J. Roy. Meteor. Soc., 120, 627653.

    • Search Google Scholar
    • Export Citation
  • Bauer, P., , Lopez P. , , Benedetti A. , , Salmond D. , , and Moreau E. , 2006: Implementation of 1D+4D-Var assimilation of precipitation-affected microwave radiances at ECMWF. I: 1D-Var. Quart. J. Roy. Meteor. Soc., 132, 22772306.

    • Search Google Scholar
    • Export Citation
  • Bauer, P., , Geer A. J. , , Lopez P. , , and Salmond D. , 2010: Direct 4D-Var assimilation of all-sky radiances. Part I: Implementation. Quart. J. Roy. Meteor. Soc., 136, 18681885.

    • Search Google Scholar
    • Export Citation
  • Courtier, P., and Coauthors, 1998: The ECMWF implementation of three-dimensional variational assimilation (3D-Var). I: Formulation. Quart. J. Roy. Meteor. Soc., 124, 17831807.

    • Search Google Scholar
    • Export Citation
  • Derber, J. C., , and Wu W.-S. , 1998: The use of TOVS cloud-cleared radiances in the NCEP SSI analysis system. Mon. Wea. Rev., 126, 22872299.

    • Search Google Scholar
    • Export Citation
  • Di Vittorio, A. V., , and Emery W. J. , 2002: An automated dynamic threshold cloud-masking algorithm for daytime AVHRR images over land. IEEE Trans. Geosci. Remote Sens., 40, 16821694.

    • Search Google Scholar
    • Export Citation
  • Dong, C., and Coauthors, 2009: An overview of a new Chinese weather satellite FY-3A. Bull. Amer. Meteor. Soc., 90, 15311544.

  • Ferraro, R., and Coauthors, 2005: NOAA operational hydrological products derived from the Advanced Microwave Sounding Unit. IEEE Trans. Geosci. Remote Sensing,43, 1036–1049.

  • Klaes, D., , and Schraidt R. , 1999: The European ATOVS and AVHRR processing package (AAPP). Proc. 10th Int. ATOVS Study Conf., Boulder, CO, WMO, 288–294.

  • Kozo, O., , Masahiro K. , , and Hiromi O. , 2005: The assimilation of ATOVS radiances in the JMA global analysis system. J. Meteor. Soc. Japan, 83, 201217.

    • Search Google Scholar
    • Export Citation
  • Lansante, J. R., 1996: Resistant, robust and non-parametric techniques for the analysis of climate data: Theory and examples, including applications to historical radiosonde. Int. J. Climatol., 16, 11971226.

    • Search Google Scholar
    • Export Citation
  • Li, Y., , Rong Z. , , Zheng Z. , , Liu J. , , Zhang L.-J. , , and Zhang L.-Y. , 2009: Post launch site calibration of visible and near-infrared channels of FY-3A visible and infrared radiometers. Opt. Precis. Eng.,17, 2966–2974.

  • Lorenc, A. C., 1986: Analysis methods for numerical weather prediction. Quart. J. Roy. Meteor. Soc., 112, 11771194.

  • Lu, Q., , Bell W. , , Bauer P. , , Bormann N. , , and Peubey C. , 2010: An initial evaluation of FY-3A satellite data. ECMWF Tech. Memo. 631, 58 pp.

  • McNally, A. P., , Derber J. C. , , Wu W. , , and Katz B. B. , 2000: The use of TOVS level-1b radiances in the NCEP SSI analysis system. Quart. J. Roy. Meteor. Soc., 126, 689724.

    • Search Google Scholar
    • Export Citation
  • Mo, T., 1999: Calibration of the Advanced Microwave Sounding Unit-A radiometers for NOAA-L and NOAA-M. NESDIS/NOAA Tech. Rep. 92, 53 pp.

  • Prigent, C., , Rossow W. L. , , and Matthews E. , 1997: Microwave land surface emissivities estimated from SSM/I observations. J. Geophys. Res., 102 (D18), 21 86721 890.

    • Search Google Scholar
    • Export Citation
  • Rohn, M., , Kelly G. , , and Saunders R. , 1998: Experiments with atmospheric motion vectors at ECMWF. Proc. Fourth Int. Winds Workshop, EUM P24, Saanenmoser, Switzerland, EUMETSAT, 139–146.

  • Van Delst, P., and Coauthors, 2011: CRTM: V2.0 User Guide, Joint Center for Satellite Data Assimilation, Camp Springs, Maryland.

  • Wang, X., , and Zou X. , 2012: An assessment of the FY-3A Microwave Temperature Sounder using the NCEP numerical weather prediction model. IEEE Trans. Geosci. Remote Sens., 50, 48604874.

    • Search Google Scholar
    • Export Citation
  • Weng, F., , and Grody N. C. , 1994: Retrieval of cloud liquid water over oceans using Special Sensor Microwave Imager (SSM/I). J. Geophys. Res., 99 (D12), 25 53525 551.

    • Search Google Scholar
    • Export Citation
  • Weng, F., , Zou X. , , Wang X. , , Yang S. , , and Goldberg M. D. , 2012: Introduction to Suomi National Polar-orbiting Partnership Advanced Technology Microwave Sounder for numerical weather prediction and tropical cyclone applications. J. Geophys. Res.,117, D19112, doi:10.1029/2012JD018144.

  • Weng, F., , Yang H. , , and Zou X. , 2013: On convertibility from antenna to sensor brightness temperature for Advanced Technology Microwave Sounder (ATMS). IEEE Geosci. Remote Sens. Lett., 10, 771775.

    • Search Google Scholar
    • Export Citation
  • Yang, J., , Dong C. , , Lu N. , , Yang Z. , , Shi J. , , Zhang P. , , Liu Y. , , and Cai B. , 2009: FY-3A: The new generation polar-orbiting meteorological satellite of China. Acta Meteor. Sin., 67, 501509.

    • Search Google Scholar
    • Export Citation
  • You, R., , Gu S. , , Guo Y. , , Chen W. , , and Yang H. , 2012: Long-term calibration and accuracy assessment of the FengYun-3 Microwave Temperature Sounder radiance measurements. IEEE Trans. Geosci. Remote Sens., 50, 48544859.

    • Search Google Scholar
    • Export Citation
  • Zhang, P., , Yang J. , , Dong C. , , Lu N. , , Yang Z. , , and Shi J. , 2009: General introduction on payloads, ground segment and data application of FengYun 3A. Front. Earth Sci. China, 3, 367373.

    • Search Google Scholar
    • Export Citation
  • Zou, X., , and Zeng Z. , 2006: A quality control procedure for GPS radio occultation data. J. Geophys. Res., 111, D02112, doi:10.1029/2005JD005846.

    • Search Google Scholar
    • Export Citation
  • Zou, X., , Wang X. , , Weng F. , , and Li G. , 2011: Assessments of Chinese FengYun microwave temperature sounder (MWTS) measurements for weather and climate applications. J. Atmos. Oceanic Technol., 28, 12061227.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (a) Global percentage of MWTS FOVs with the cloud fraction being greater than fVIRR for data during July 2011. The three threshold values, fVIRR = 37%, 97%, and 100%, are indicated in (a) by the vertical dotted lines. (b) Daily variation of cloudy MWTS FOV percentage with fVIRR = 37% (green), 97% (blue), and 100% (red) over the entire globe (solid) and ocean only (dashed).

  • View in gallery

    Distributions of cloudy FOVs when values of two variables (a) fVIRR and (b) LWP are within four different ranges for data during 0300–1500 UTC 1 Jul 2011.

  • View in gallery

    Scatterplots of brightness temperature differences between observations and model simulations for data within 30°N–30°S on 1 July 2011. Outliers identified by the biweighting check are in red.

  • View in gallery

    Percentage of outliers identified by land, coastal FOVs, sea ice, scan edge, terrain altitude >500 m, cloud detection, and O-B biweighting check as well as the remaining data (the last three bars) for MWTS channels 2, 3, and 4 in July 2011.

  • View in gallery

    Scatterplots of brightness temperature differences between observations and model simulations for MWTS channels (a),(b) 2, (c),(d) 3, and (e),(f) 4 for (left) outliers and (right) data that passed quality control during 1–5 July 2011.

  • View in gallery

    Global distribution of brightness temperature differences (K) between observations and model simulations for channels (a),(b) 2, (c),(d) 3, and (e),(f) 4 for (left) outliers and (right) the remaining data on 0300–1500 UTC 1 July 2011.

  • View in gallery

    Latitudinal dependences of scan biases (shaded) for channels (a),(b) 2, (c),(d) 3, and (e),(f) 4 (left) before and (right) after quality control. Globally averaged scan biases are shown by the blue solid line, y-axis values on the right of each panel. Nadir biases are subtracted.

  • View in gallery

    Frequency distributions of O-B differences for channels (top) 2, (middle) 3, and (bottom) 4 before (hatched bars) and after (solid bars) quality control for MWTS channels 2–4. Biases are removed for the frequency distributions of data after quality control.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 45 45 8
PDF Downloads 31 31 5

A Quality Control Procedure for FY-3A MWTS Measurements with Emphasis on Cloud Detection Using VIRR Cloud Fraction

View More View Less
  • 1 Numerical Prediction Center, China Meteorological Administration, Beijing, China
  • | 2 Department of Earth, Ocean and Atmospheric Sciences, The Florida State University, Tallahassee, Florida
© Get Permissions
Full access

Abstract

A quality control (QC) procedure for satellite radiance assimilation is proposed and applied to radiance observations from the Microwave Temperature Sounder (MWTS) on board the first satellite of the Chinese polar-orbiting Fengyun-3 series (FY-3A). A cloud detection algorithm is incorporated based on the cloud fraction product provided by the Visible and Infrared Radiometer (VIRR) on board FY-3A. Analysis of the test results conducted in July 2011 indicates that most clouds are identifiable by applying an FY-3A VIRR cloud fraction threshold of 37%. This result is verified with the cloud liquid water path data from the Meteorological Operational Satellite A (MetOp-A). On average, 56.1% of the global MWTS data are identified as cloudy by the VIRR-based cloud detection method. Other QC steps include the following: (i) two outmost field of views (FOVs), (ii) use of channel 3 if the terrain altitude is greater than 500 m, (iii) channel 2 over sea ice and land, (iv) coastal FOVs, and (v) outliers with large differences between model simulations and observations. About 82%, 74%, and 29% of the MWTS observations are removed by the proposed QC for channels 2–4, respectively. An approximate 0.5-K scan bias improvement is achieved with QC, with a large impact at edges of the field of regard for channels 2–4. After QC, FY-3A MWTS global data more closely resemble the National Centers for Environmental Prediction (NCEP) forecast data, the global biases and standard deviations are reduced significantly, and the frequency distribution of the differences between observations and model simulations become more Gaussian.

Corresponding author address: Dr. X. Zou, Department of Earth, Ocean and Atmospheric Science, Florida State University, 110 S. Woodward Ave., Tallahassee, FL 32306-4520. E-mail: xzou@fsu.edu

Abstract

A quality control (QC) procedure for satellite radiance assimilation is proposed and applied to radiance observations from the Microwave Temperature Sounder (MWTS) on board the first satellite of the Chinese polar-orbiting Fengyun-3 series (FY-3A). A cloud detection algorithm is incorporated based on the cloud fraction product provided by the Visible and Infrared Radiometer (VIRR) on board FY-3A. Analysis of the test results conducted in July 2011 indicates that most clouds are identifiable by applying an FY-3A VIRR cloud fraction threshold of 37%. This result is verified with the cloud liquid water path data from the Meteorological Operational Satellite A (MetOp-A). On average, 56.1% of the global MWTS data are identified as cloudy by the VIRR-based cloud detection method. Other QC steps include the following: (i) two outmost field of views (FOVs), (ii) use of channel 3 if the terrain altitude is greater than 500 m, (iii) channel 2 over sea ice and land, (iv) coastal FOVs, and (v) outliers with large differences between model simulations and observations. About 82%, 74%, and 29% of the MWTS observations are removed by the proposed QC for channels 2–4, respectively. An approximate 0.5-K scan bias improvement is achieved with QC, with a large impact at edges of the field of regard for channels 2–4. After QC, FY-3A MWTS global data more closely resemble the National Centers for Environmental Prediction (NCEP) forecast data, the global biases and standard deviations are reduced significantly, and the frequency distribution of the differences between observations and model simulations become more Gaussian.

Corresponding author address: Dr. X. Zou, Department of Earth, Ocean and Atmospheric Science, Florida State University, 110 S. Woodward Ave., Tallahassee, FL 32306-4520. E-mail: xzou@fsu.edu

1. Introduction

On 27 May 2008, the satellite Fengyun-3A (FY-3A) was launched with 11 instruments on board. It is a sun-synchronous polar-orbiting environmental research satellite. The local equator crossing time (LECT) for the FY-3A ascending node and descending node is around 1000 and 2200 LT, respectively. The second satellite in the FY-3 series, FY-3B, was successfully launched on 5 November 2010, carrying the same instruments as those on board FY-3A. FY-3A carries three sounding instruments: the Microwave Temperature Sounder (MWTS), the Microwave Humidity Sounder (MWHS), and the Infrared Atmospheric Sounder (IRAS), providing China's first ever global atmospheric sounding data (Zhang et al. 2009). The performance of the atmospheric sounding instruments meets or exceeds the prelaunch specifications (Dong et al. 2009; Yang et al. 2009). Besides MWTS, MWHS, and IRAS, a visible and infrared instrument called the Visible and Infrared Radiometer (VIRR) is also on board FY-3A for the purpose of providing cloud cover, clear-sky sea surface temperatures, and cloud-top brightness temperatures. It is of great value to incorporate the MWTS, MWHS, and IRAS data from the FY-3 series into numerical weather prediction (NWP) models. This study describes a quality control procedure that could be implemented prior to the assimilation of the MWTS data into NWP models.

MWTS is a four-channel cross-track microwave scanning radiometer that is similar to the Microwave Sounding Unit (MSU) on board the early National Oceanic and Atmospheric Administration (NOAA) satellites, from the Television and Infrared Observation Satellite-N series (TIROS-N) through NOAA-14 from 1979 to 2007. It is also similar to the Advanced Microwave Sounding Unit-A (AMSU-A) channels 3, 5, 7, and 9 on board NOAA-15, NOAA-16, NOAA-17, NOAA-18, NOAA-19, and Aqua (You et al. 2012). AMSU-A data have long been incorporated in almost all operational NWP systems in the world, including the National Centers for Environmental Prediction (NCEP) and the European Center for Medium-Range Weather Forecasts (ECMWF). It is widely accepted that direct assimilation of radiance observations from microwave temperature sounding channels can significantly improve the accuracy of global and regional weather analysis and forecasts (Andersson et al. 1994; Courtier et al. 1998; Derber and Wu 1998; McNally et al. 2000; Kozo et al. 2005). An initial evaluation of FY-3A MWTS against NOAA-18 AMSU-A by Zou et al. (2011) showed that the MWTS data compare favorably with the AMSU-A data in terms of their global bias, with MWTS standard deviations slightly larger than those of AMSU-A data. A similar comparison of the data from the four observing instruments on board FY-3A and similar instruments on board the Meteorological Operational Satellite A (MetOp-A) can be found in Lu et al. (2010). Moreover, Wang and Zou (2012) showed that the temperature dependence of the FY-3A MWTS measurement biases is mainly introduced by a postlaunch frequency shift found by Lu et al. (2010). By incorporating the shifted frequencies into radiative transfer models, MWTS biases are nearly constant with the scene temperature. The addition of FY-3 vertical atmospheric sounding radiance data increases the current global coverage of satellite observations. It is anticipated that the MWTS data might be useful for NWP modeling systems if properly assimilated.

Quality control (QC) is important for the assimilation of any satellite data. Observations have errors, forward models have limitations, and forecast models have uncertainty. Since data assimilation systems, such as three-dimensional/four-dimensional variational (3D-Var/4D-Var) systems are very sensitive to erroneous observations and are inclined to converge toward outliers (Lorenc 1986), observations that are either unreliable or cannot be simulated accurately by the fast radiative transfer model (RTM) and NWP models must be identified and eliminated before being assimilated. Development of an effective QC procedure is thus important for all observations to optimally improve numerical weather forecasts (Rohn et al. 1998). As mentioned above, MWTS is the first atmospheric sounding instrument developed by China. It is important to have an effective and efficient QC procedure for MWTS measurements from the FY-3 series before being incorporated into the Chinese NWP or other data assimilations systems. This study develops a series of quality control procedures for MWTS radiance assimilation and compares the statistical features of differences between MWTS observations and radiance simulations using the NCEP Global Forecast System (GFS) 6-h forecast fields.

This paper is organized as followed: section 2 briefly describes the MWTS instrument on board FY-3A, satellite observations, and model simulations. Section 3 presents the quality control procedure that consists of cloud detection, surface identification, scan position check, and large differences between observations and model simulations based on the NCEP GFS 6-h forecast fields. The numerical results on some data characteristics (e.g., outliers, scan-dependent biases, and frequency distribution) before and after QC are provided in section 4. The conclusion is given in section 5.

2. Data description

a. MWTS instrument characteristics and data description

The absorption and emission of microwave radiation by atmospheric oxygen enables the MWTS to passively sense temperature through the atmosphere as a function of altitude. The MWTS has four channels in the oxygen band at frequencies between 50.3 and 57.3 GHz for atmospheric temperature profiling from the earth's surface to about 16 km (or 90 hPa). Level-1b radiance data from the FY-3A MWTS are employed for this study for the month of July 2011. The two-point calibration technique for AMSU-A (Mo 1999) is used for the calibration of the MWTS data (You et al. 2012). Table 1 lists a few selected channel characteristics of the FY-3A MWTS, including channel frequency, peak weighting function height, and radiometric temperature sensitivity [Noise Equivalent Differential Temperature (NEΔT); Dong et al. 2009]. There are only 15 field-of-view (FOV) scenes along each MWTS scan line. The horizontal FOV resolution at nadir is 62 km. The MWTS channel 1 frequency is centered at 50.3 GHz and is sensitive to surface and cloud liquid water. MWTS channels 2–4 are for the sounding of the atmospheric temperature in the troposphere and low stratosphere.

Table 1.

Channel characteristics of FY-3A MWTS.

Table 1.

b. Global simulations of brightness temperatures

In this study, the Community Radiative Transfer Model (CRTM 2.0) is used to simulate MWTS brightness temperatures (Van Delst et al. 2011). The 6-h forecasts of the vertical profiles of temperature, specific humidity, and the surface pressure from the NCEP GFS are used as input to CRTM. The NCEP GFS 6-h forecast fields have a horizontal resolution of 1° × 1° and 26 vertical levels. The highest vertical level is around 10 hPa.

Global simulations of brightness temperature are used as a reference for examining the performance of the MWTS instrument. Statistical features of the differences between the MWTS observed and model-simulated brightness temperatures are examined. For brevity, the difference between MWTS observations and model simulations generated by CRTM with the 6-h forecasts from the NCEP GFS as its input will be denoted by O-B (i.e., differences between observations and model simulations of the AMSU-A channel 1).

3. Cloud detection and other quality control procedures

a. The importance of cloud detection

Challenges to the assimilation of satellite radiance data are associated with radiance biases, cloud contamination, and surface emissivity. Microwave instruments at MWTS frequencies may penetrate through clouds except when there is heavy precipitation. MWTS radiance observations in nonheavy precipitation contain useful information about clouds and precipitation. However, rain-contaminated radiances are difficult to simulate accurately using a fast radiative transfer model (e.g., CRTM). Assimilating cloud- and precipitation-affected microwave imager radiances, though challenging, is being done operationally by ECMWF (Bauer et al. 2006, 2010). Our current work of assimilating clear-sky microwave temperature sounding radiances is a first step toward the assimilation of all-weather data. Therefore, data in cloudy conditions are eliminated from tropospheric microwave temperature sounding channels such as MWTS channels 2 and 3.

There were several cloud detection methods developed for microwave satellite measurements. The scatter index (SI) method is applied to the Advanced TIROS Operational Vertical Sounder (ATOVS) and Advanced Very High Resolution Radiometer (AVHRR) data by a preprocessing package (AAPP) to detect scattering hydrometeors over the ocean (Klaes and Schraidt 1999), where SI is defined by the difference between AMSU-A channel 15 observations and a physical retrieval-derived AMSU-A channel 15 simulation using a linear regression model between AMSU-A channel 15 and AMSU-A channels 1, 2, and 3 observations over the ocean. Another cloud detection algorithm used in the Microwave Surface and Precipitation Products System (MSPPS) is based on the cloud liquid water path (LWP) estimated by AMSU-A channels 1 and 2 (Weng and Grody 1994). However, these methods are mainly developed for the AMSU-A channels. Since the two surface-sensitive channels that could be used for LWP retrievals are not available from the MWTS data, the cloud detection algorithms develop for the AMSU-A data cannot be directly applied to the MWTS data. Fortunately, the cloud fraction within the MWTS FOVs can be derived from the VIRR cloud mask, providing a direct method for the MWTS cloud detection needed in the MWTS data assimilation. This is one of the advantages of having the VIRR on board the same satellite platform as the MWTS.

b. VIRR cloud detection algorithm

The VIRR is a visible and infrared instrument that has 10 channels spanning the spectrum from 0.455 to 12.5 μm (Table 2; Dong et al. 2009). The horizontal resolution of the pixel at nadir is 1.1 km. The wavelengths and resolution of channels 1–5 are similar to the AVHRR on board the NOAA series of satellites. Details on data characteristics, calibration, and the accuracy of the VIRR can be found in Li et al. (2009).

Table 2.

Channels and characteristics of VIRR.

Table 2.

The FY-3A VIRR cloud detection algorithm is described as follows. For each pixel, a cloud-masking algorithm based on dynamic thresholds is used for identifying whether the pixel is cloudy or not using 10 VIRR channels (Di Vittorio and Emery 2002). A total of 10 cloud masks are created for each scene. The dynamic thresholds for each of the 10 cloud masks are determined by the inflection point on a frequency histogram of reflectivity or brightness temperature for each of the 10 VIRR channel. More details on the determination of these thresholds can be found in Di Vittorio and Emery (2002). The cloud masks can be obtained by comparing the observations and thresholds. The cloud masks derived from all channels are used in the daytime. In the nighttime, cloud masks from channels 3–5 are applied. For each pixel, another four cloud masks are produced by checking the values of the quantities
eq1
against a given set of thresholds, where refi is the reflectance of the ith channel (i = 1 and 2), and is the brightness temperature of the ith channel (i = 3, 4, and 5). The first two are designed for the daytime cloud detection and are used for the nighttime. A VIRR pixel is classified as cloudy when more than six (in the daytime) or three (in the nighttime) cloud masks identified the pixel as cloudy. Otherwise, it is classified as clear sky.

Since the spatial resolution of the MWTS is much coarser than that of the VIRR, a cloud fraction is calculated for an individual MWTS FOV, which is defined as the ratio of the number of cloudy pixels to the total number of VIRR pixels collocated within the MWTS FOV. If all VIRR pixels situated in a MWTS FOV are cloudy, the cloud fraction of this single MWTS FOV will be 100%. An MWTS FOV with a cloud fraction greater than a threshold fVIRR, which will be determined in the following subsection, will be identified as a cloudy radiance in this study. For brevity, the FY-3A MWTS cloud detection algorithm will be denoted the FY3A-VIRR method.

The pixel-level VIRR data are first collocated with each MWTS FOV. The time differences between any VIRR pixels matched with a target MWTS FOV are less than 32 s. It is required that the distance between the center of the matched VIRR pixel and a target MWTS FOV at either nadir or off nadir be less than 0.6 times the length of the long and short axis in across-track and along-track directions, respectively. It is also required that the sum of the distances from the center of the matched VIRR pixel to the two focus points of an elliptical MWTS FOV is less than or equal to the length of the long axis of the MWTS FOV. Also considered in the collocation procedure is the increasing footprint size as the instrument scans away from the satellite subpoint.

c. Another cloud detection method

In this study, the MetOp-A AMSU-A cloud LWP products, obtained from the MSPPS, will be used for checking the consistency between the FY-3A VIRR cloud fraction and the MetOp-A AMSU-A cloud LWP (Weng and Grody 1994; Ferraro et al. 2005). MetOp-A is chosen because its LECT is around 0930, which is only a half hour earlier than the LECT of FY-3A. The MetOp-A cloud LWP is retrieved over the ocean only (without sea ice) and varies from 0.01 to 2 kg m−2. In this study, an FOV with cloud LWP more than 0.01 kg m−2 is marked as cloudy. The global percentage of the cloudy radiances are compared with those obtained based on the VIRR cloud fraction.

d. Comparisons between two cloud detection methods

Figure 1a presents the percentage of MWTS FOVs with their cloud fraction being greater than fVIRR, whose value is varied from 0% to 100% using 1-month data in July 2011. As expected, the global percentage of MWTS FOVs with cloud fraction being greater than fVIRR decreases as fVIRR increases. The three thresholds of 37%, 97%, and 100% of fVIRR are indicated in Fig. 1a by the vertical dotted lines. Table 3 compares cloud percentages over the ocean at two selected thresholds for each of two cloud detection methods as described in sections 3b and 3c. For MetOp-A AMSU-A cloud LWP retrieval, data are grouped into three categories: LWP ≥ 0.01, 0.1, and 0.14 kg m−2. At these thresholds, percentages of cloudy AMSU-A FOVs are 58.1%, 24.2%, and 18.6%, respectively, which are comparable to global oceanic percentages of cloudy FY-3A MWTS FOVs as identified by fVIRR when it is set to be greater than 37%, 97%, and 100%. It is seen that when fVIRR equals 37%, the cloud coverage is around 58.3% over the ocean, which is almost equivalent to the cloud percentage of 58.1% over the ocean obtained by setting LWP > 0.01 g m−2. The value of 37% is thus chosen as the threshold for fVIRR. Figure 1b displays the daily variation of cloud percentage estimated by setting fVIRR to 37%, 97%, or 100% during the entire month of July 2011. As can be seen, similar daily changes of global cloud percentage can be found for three thresholds, and it is generally stable in a month.

Fig. 1.
Fig. 1.

(a) Global percentage of MWTS FOVs with the cloud fraction being greater than fVIRR for data during July 2011. The three threshold values, fVIRR = 37%, 97%, and 100%, are indicated in (a) by the vertical dotted lines. (b) Daily variation of cloudy MWTS FOV percentage with fVIRR = 37% (green), 97% (blue), and 100% (red) over the entire globe (solid) and ocean only (dashed).

Citation: Journal of Atmospheric and Oceanic Technology 30, 8; 10.1175/JTECH-D-12-00164.1

Table 3.

Percentages of cloudy radiances over the global ocean estimated from FY-3A VIRR cloud fraction data and MetOp-A AMSU-A cloud LWP retrievals using 1-month data in July 2011.

Table 3.

Figure 2 shows the distributions of cloudy FOVs when the values of two variables fVIRR and LWP are within four different ranges for data during 0330–0930 UTC 1 July 2011. Note that the MWTS swaths do not overlap the AMSU-A swaths, and the size of data points plotted in this figure does not reflect the increase of FOV size with a scan angle. Note that MWTS observation resolutions are coarser than AMSU-A observations. For example, MWTS nadir FOV diameter is 62 km and that of AMSU-A is 48 km. Nevertheless, the general patterns of cloudiness are comparable between Figs. 2a and 2b. Results using fVIRR for cloud detection compare well with those of LWP over their overlapping portions of swaths. Note that the comparison between Figs. 2a and 2b is only of qualitative value since the fVIRR method identifies all types of clouds, whereas the LWP products are only for liquid clouds.

Fig. 2.
Fig. 2.

Distributions of cloudy FOVs when values of two variables (a) fVIRR and (b) LWP are within four different ranges for data during 0300–1500 UTC 1 Jul 2011.

Citation: Journal of Atmospheric and Oceanic Technology 30, 8; 10.1175/JTECH-D-12-00164.1

For all data points in July 2011, 58.3% and 58.1% of global oceanic FOVs are found to be cloudy based on the thresholds of fVIRR > 37% and LWP > 0.01 kg m−2, respectively. The remaining clear-sky pixels over the ocean are thus 41.7% and 41.9% based on the thresholds of fVIRR ≤ 37% and LWP ≤ 0.01 kg m−2, respectively.

e. Surface identification and scan position check

Besides clouds, surface emissivity causes another type of challenge for satellite data assimilation. Surface emissivity varies with frequency, surface type, soil moisture content, vegetation cover, and viewing angle. The ocean emissivity is obtained using CRTM's internal emissivity model. The land surface emissivity is from the observationally derived emissivity dataset that is used in the Radiative Transfer for TIROS Operational Vertical Sounder (RTTOV-10; Prigent et al. 1997). Simulations of surface-sensitive channels show less agreement with observations, especially over land, snow, ice, and coastal areas. The large uncertainty in surface emissivity and the significant impact of surface emissivity on the radiance simulations also makes the detection of cloud/precipitation much more difficult over land, snow, ice, and coastal areas. Additional QC steps dealing with problems associated with surface emissivity, high terrain, and outliers are implemented for FY-3A MWTS data.

Table 4 provides a channel selection scheme for the MWTS. When fVIRR > 37%, only channel 4 data are used because the weighting function of this channel is in the low stratosphere, and clouds have negligible effects on this channel. Channel 2 is used only over the ocean when there is no sea ice. The MWTS channel 2 over land and channel 3 over high terrain (e.g., terrain height is greater than 500 m) is not used because of uncertainties in surface emissivity (Prigent et al. 1997). The identification of an underlying land/sea/coast is based on a land mask database with a 0.05° longitudinal and latitudinal resolution. Sea ice is identified when sea surface temperature (SST) is less than 273.15 K using the China Meteorological Administration (CMA) daily SST data from the Global Telecommunication System (GTS). The two outmost FOVs (i.e., FOV 1, 2, 14, and 15) for all channels are excluded for data assimilation because of larger inhomogeneous limb effects that could not be accurately accounted for in forward radiative transfer models and a larger geolocation error near the swath edge. Channel 1 is not included in data assimilation because it is strongly sensitive to surface conditions.

Table 4.

Channel selection based on cloud fraction, terrain height z, and surface type.

Table 4.

f. Biweight check

To remove the remaining outliers, a biweighting quality control procedure is implemented, aiming at remove those radiance data that deviate greatly from the background fields. First, the biweight mean μbm and biweight standard deviation σbsd of the following variable are calculated (Lansante 1996; Zou and Zeng 2006):
eq2
where and represent the observed and model-simulated brightness temperature, respectively. Then, a z score for all data that pass all the checks on the cloud, terrain height, and surface types listed in Table 4 is calculated:
eq3
where the subscript i indicates the ith datum. Data with a z score of more than two are removed. Considering the variations of the mean states of the atmosphere at different latitudes, the biweighting quality control is implemented in three separate latitudinal bands separately: the tropics (30°N–30°S), midlatitudes (30°–60°N and 30°–60°S), and high latitudes (60°–90°N and 60°–90°S).

Figure 3 presents scatterplots of differences of brightness temperatures between the observations and model simulations for data within the tropics on 1 July 2011. Outliers identified by the biweighting check are indicated. It is seen that outliers identified by the biweighting method are located on both sides of the mean, which could result from the NWP model not being able to describe the exact cloud patterns, water vapor distributions, or surface emissivity.

Fig. 3.
Fig. 3.

Scatterplots of brightness temperature differences between observations and model simulations for data within 30°N–30°S on 1 July 2011. Outliers identified by the biweighting check are in red.

Citation: Journal of Atmospheric and Oceanic Technology 30, 8; 10.1175/JTECH-D-12-00164.1

Elimination of data by the MWTS quality control is carried out sequentially in the following order: 1) coastal FOVs are removed; 2) FOVs of channel 2 over sea ice are removed; 3) FOVs at the scan edges (two from both sides of a scan line) are removed; 4) FOVs of channel 3 over terrain higher than 500 m are removed; 5) FOVs of channel 2 over land are removed; 6) cloudy FOVs of channel 2 and 3 with fVIRR greater than 37% are identified and removed; and 7) the remaining outliers identified by the biweighting check are eliminated. The final results of the percentages of data eliminated by the above seven steps are shown in Fig. 4 for data in July 2011. Also shown in Fig. 4 are the data counts that pass the MWTS quality control, which are about 18%, 26%, and 71% of data for channels 2, 3, and 4, respectively.

Fig. 4.
Fig. 4.

Percentage of outliers identified by land, coastal FOVs, sea ice, scan edge, terrain altitude >500 m, cloud detection, and O-B biweighting check as well as the remaining data (the last three bars) for MWTS channels 2, 3, and 4 in July 2011.

Citation: Journal of Atmospheric and Oceanic Technology 30, 8; 10.1175/JTECH-D-12-00164.1

4. Comparison of data characteristics before and after QC

Figure 5 presents scatterplots in 1–5 July 2011, and Fig. 6 shows the global distributions during 0300–1500 UTC of the O-B differences at channels 2–4 for outliers (left panels) and data that pass quality control (right panels), respectively. Outliers removed by specific quality control tests are indicated in colors. The O-B differences of outliers and their standard deviations are much larger than those of the remaining data after quality control. For channel 2, the observed cloudy radiances could be more than 15 K colder than model simulations (Fig. 5a). After quality control, the data spread for all three channels is significantly reduced and the O-B differences do not show obvious temperature dependency (Fig. 5b).

Fig. 5.
Fig. 5.

Scatterplots of brightness temperature differences between observations and model simulations for MWTS channels (a),(b) 2, (c),(d) 3, and (e),(f) 4 for (left) outliers and (right) data that passed quality control during 1–5 July 2011.

Citation: Journal of Atmospheric and Oceanic Technology 30, 8; 10.1175/JTECH-D-12-00164.1

Fig. 6.
Fig. 6.

Global distribution of brightness temperature differences (K) between observations and model simulations for channels (a),(b) 2, (c),(d) 3, and (e),(f) 4 for (left) outliers and (right) the remaining data on 0300–1500 UTC 1 July 2011.

Citation: Journal of Atmospheric and Oceanic Technology 30, 8; 10.1175/JTECH-D-12-00164.1

Angular-dependent biases between the observed brightness temperatures and those simulated from the radiative transfer models are anticipated for the cross-track MWTS instrument. The size of the FOV increases with an increasing scan angle. The larger the FOV is, the higher the atmospheric inhomogeneity is within the FOV. The atmospheric inhomogeneity within FOVs at larger scan angles cannot be accurately simulated in radiative transfer models. Furthermore, satellite observations at large scan angles could be obstructed by the spacecraft radiation (Weng et al. 2013). This is also difficult to be taken into account in the forward model and calibration process. Figure 7 presents the latitudinal variation of scan biases of the MWTS channels 2–4. They are calculated within every 5° latitudinal band for each scan position for the entire month of July 2011. Moreover, nadir biases are subtracted. These biases will be subtracted from O-B values in the formulation of data assimilation (Weng et al. 2012).

Fig. 7.
Fig. 7.

Latitudinal dependences of scan biases (shaded) for channels (a),(b) 2, (c),(d) 3, and (e),(f) 4 (left) before and (right) after quality control. Globally averaged scan biases are shown by the blue solid line, y-axis values on the right of each panel. Nadir biases are subtracted.

Citation: Journal of Atmospheric and Oceanic Technology 30, 8; 10.1175/JTECH-D-12-00164.1

Before quality control, large negative scan biases are present at large scan angles at all latitudes for channel 2 near the end of scan lines and channel 3 at both sides of the scan lines. Relatively large negative scan biases for channel 4 are seen in high latitudes at large scan angles. After quality control, the scan biases become small and homogeneous without a strong latitudinal dependence. Therefore, a constant scan bias for each channel needs be removed before MWTS radiance assimilation.

Frequency distributions of O-B differences before and after quality control for MWTS channels 2–4 are provided in Fig. 8. Angular biases are removed for the frequency distributions of data after quality control. The residual averaged O-B biases are also removed. The frequency distributions of all data without quality control are asymmetric, with more data located on the left-hand side of the distributions. The bias and standard deviations of the O-B differences between observations and model simulations are reduced after quality control and bias correction.

Fig. 8.
Fig. 8.

Frequency distributions of O-B differences for channels (top) 2, (middle) 3, and (bottom) 4 before (hatched bars) and after (solid bars) quality control for MWTS channels 2–4. Biases are removed for the frequency distributions of data after quality control.

Citation: Journal of Atmospheric and Oceanic Technology 30, 8; 10.1175/JTECH-D-12-00164.1

5. Summary and conclusions

Since 2008, China has launched the second generation of polar-orbiting satellites called the Fengyun-3 (FY-3) series. The first two satellites in the FY-3 series, FY-3A and FY-3B, were successfully launched into a morning orbit and an afternoon orbit on 27 May 2008 and 5 November 2010, respectively. Both FY-3A and FY-3B carry 11 instruments. After the intensive calibration/validation period of its first 6 months in orbit, it is concluded that the performance of the FY-3 series atmospheric sounding instruments met or exceeded system specifications (Dong et al. 2009; Yang et al. 2009).

In parallel with advances in satellite instruments, China also had begun its development of a numerical weather prediction data assimilation system and model called the Global/Regional Assimilation and Prediction System (GRAPES). It is deemed important to incorporate data from the FY-3 series into GRAPES. As the first step, a quality control procedure is developed for FY-3A MWTS data and is presented in this study.

A cloud detection scheme for MWTS data is tested using a VIRR cloud fraction retrieval product. The global percentages of clouds from applying this variable to MWTS data are compared with those obtained by applying cloud LWP products derived from two MetOp-A AMSU-A surface-sensitive channels. Analysis indicates that most clouds are identifiable by setting a VIRR cloud fraction threshold at 37%. On average, the FY-3A VIRR-based cloud detection method found 56.1% of MWTS FOVs to be cloudy globally.

In addition to cloud detection, observations over land and sea ice are also removed for channels 2 or 3. Two outermost FOVs as well as coastal FOVs are also eliminated for all channels. Finally, outliers that deviate greatly from the model-simulated MWTS radiances are also eliminated. About 18%, 26%, and 71% of observations are retained after quality control.

After quality control, FY-3A MWTS data show a more coherent pattern when compared to model simulations. The frequency distributions of the differences between observations and model simulations for all three MWTS sounding channels are more Gaussian after quality control and bias corrections.

This work can be extended to the FY-3 Microwave Temperature Sounder measurements. The methodology will be incorporated into GRAPES or other NWP modeling systems. Incorporating FY-3 vertical atmospheric sounding radiance data can increase the current global coverage of satellite observations in operational numerical weather prediction. An impact experiment of FY-3 MWTS on NWP models is being carried out and results will be presented in a separate paper.

Acknowledgments

This work was jointly supported by Chinese Ministry of Science and Technology Project 2010CB951600 and Chinese Ministry of Finance Projects GYHY200906006 and GYHY201106008.

REFERENCES

  • Andersson, E., , Pailleux J. , , Thepaut J.-N. , , Eyre J. R. , , McNally A. P. , , Kelly G. A. , , and Courtier P. , 1994: Use of cloud-cleared radiances in three-/four-dimensional variational data assimilation. Quart. J. Roy. Meteor. Soc., 120, 627653.

    • Search Google Scholar
    • Export Citation
  • Bauer, P., , Lopez P. , , Benedetti A. , , Salmond D. , , and Moreau E. , 2006: Implementation of 1D+4D-Var assimilation of precipitation-affected microwave radiances at ECMWF. I: 1D-Var. Quart. J. Roy. Meteor. Soc., 132, 22772306.

    • Search Google Scholar
    • Export Citation
  • Bauer, P., , Geer A. J. , , Lopez P. , , and Salmond D. , 2010: Direct 4D-Var assimilation of all-sky radiances. Part I: Implementation. Quart. J. Roy. Meteor. Soc., 136, 18681885.

    • Search Google Scholar
    • Export Citation
  • Courtier, P., and Coauthors, 1998: The ECMWF implementation of three-dimensional variational assimilation (3D-Var). I: Formulation. Quart. J. Roy. Meteor. Soc., 124, 17831807.

    • Search Google Scholar
    • Export Citation
  • Derber, J. C., , and Wu W.-S. , 1998: The use of TOVS cloud-cleared radiances in the NCEP SSI analysis system. Mon. Wea. Rev., 126, 22872299.

    • Search Google Scholar
    • Export Citation
  • Di Vittorio, A. V., , and Emery W. J. , 2002: An automated dynamic threshold cloud-masking algorithm for daytime AVHRR images over land. IEEE Trans. Geosci. Remote Sens., 40, 16821694.

    • Search Google Scholar
    • Export Citation
  • Dong, C., and Coauthors, 2009: An overview of a new Chinese weather satellite FY-3A. Bull. Amer. Meteor. Soc., 90, 15311544.

  • Ferraro, R., and Coauthors, 2005: NOAA operational hydrological products derived from the Advanced Microwave Sounding Unit. IEEE Trans. Geosci. Remote Sensing,43, 1036–1049.

  • Klaes, D., , and Schraidt R. , 1999: The European ATOVS and AVHRR processing package (AAPP). Proc. 10th Int. ATOVS Study Conf., Boulder, CO, WMO, 288–294.

  • Kozo, O., , Masahiro K. , , and Hiromi O. , 2005: The assimilation of ATOVS radiances in the JMA global analysis system. J. Meteor. Soc. Japan, 83, 201217.

    • Search Google Scholar
    • Export Citation
  • Lansante, J. R., 1996: Resistant, robust and non-parametric techniques for the analysis of climate data: Theory and examples, including applications to historical radiosonde. Int. J. Climatol., 16, 11971226.

    • Search Google Scholar
    • Export Citation
  • Li, Y., , Rong Z. , , Zheng Z. , , Liu J. , , Zhang L.-J. , , and Zhang L.-Y. , 2009: Post launch site calibration of visible and near-infrared channels of FY-3A visible and infrared radiometers. Opt. Precis. Eng.,17, 2966–2974.

  • Lorenc, A. C., 1986: Analysis methods for numerical weather prediction. Quart. J. Roy. Meteor. Soc., 112, 11771194.

  • Lu, Q., , Bell W. , , Bauer P. , , Bormann N. , , and Peubey C. , 2010: An initial evaluation of FY-3A satellite data. ECMWF Tech. Memo. 631, 58 pp.

  • McNally, A. P., , Derber J. C. , , Wu W. , , and Katz B. B. , 2000: The use of TOVS level-1b radiances in the NCEP SSI analysis system. Quart. J. Roy. Meteor. Soc., 126, 689724.

    • Search Google Scholar
    • Export Citation
  • Mo, T., 1999: Calibration of the Advanced Microwave Sounding Unit-A radiometers for NOAA-L and NOAA-M. NESDIS/NOAA Tech. Rep. 92, 53 pp.

  • Prigent, C., , Rossow W. L. , , and Matthews E. , 1997: Microwave land surface emissivities estimated from SSM/I observations. J. Geophys. Res., 102 (D18), 21 86721 890.

    • Search Google Scholar
    • Export Citation
  • Rohn, M., , Kelly G. , , and Saunders R. , 1998: Experiments with atmospheric motion vectors at ECMWF. Proc. Fourth Int. Winds Workshop, EUM P24, Saanenmoser, Switzerland, EUMETSAT, 139–146.

  • Van Delst, P., and Coauthors, 2011: CRTM: V2.0 User Guide, Joint Center for Satellite Data Assimilation, Camp Springs, Maryland.

  • Wang, X., , and Zou X. , 2012: An assessment of the FY-3A Microwave Temperature Sounder using the NCEP numerical weather prediction model. IEEE Trans. Geosci. Remote Sens., 50, 48604874.

    • Search Google Scholar
    • Export Citation
  • Weng, F., , and Grody N. C. , 1994: Retrieval of cloud liquid water over oceans using Special Sensor Microwave Imager (SSM/I). J. Geophys. Res., 99 (D12), 25 53525 551.

    • Search Google Scholar
    • Export Citation
  • Weng, F., , Zou X. , , Wang X. , , Yang S. , , and Goldberg M. D. , 2012: Introduction to Suomi National Polar-orbiting Partnership Advanced Technology Microwave Sounder for numerical weather prediction and tropical cyclone applications. J. Geophys. Res.,117, D19112, doi:10.1029/2012JD018144.

  • Weng, F., , Yang H. , , and Zou X. , 2013: On convertibility from antenna to sensor brightness temperature for Advanced Technology Microwave Sounder (ATMS). IEEE Geosci. Remote Sens. Lett., 10, 771775.

    • Search Google Scholar
    • Export Citation
  • Yang, J., , Dong C. , , Lu N. , , Yang Z. , , Shi J. , , Zhang P. , , Liu Y. , , and Cai B. , 2009: FY-3A: The new generation polar-orbiting meteorological satellite of China. Acta Meteor. Sin., 67, 501509.

    • Search Google Scholar
    • Export Citation
  • You, R., , Gu S. , , Guo Y. , , Chen W. , , and Yang H. , 2012: Long-term calibration and accuracy assessment of the FengYun-3 Microwave Temperature Sounder radiance measurements. IEEE Trans. Geosci. Remote Sens., 50, 48544859.

    • Search Google Scholar
    • Export Citation
  • Zhang, P., , Yang J. , , Dong C. , , Lu N. , , Yang Z. , , and Shi J. , 2009: General introduction on payloads, ground segment and data application of FengYun 3A. Front. Earth Sci. China, 3, 367373.

    • Search Google Scholar
    • Export Citation
  • Zou, X., , and Zeng Z. , 2006: A quality control procedure for GPS radio occultation data. J. Geophys. Res., 111, D02112, doi:10.1029/2005JD005846.

    • Search Google Scholar
    • Export Citation
  • Zou, X., , Wang X. , , Weng F. , , and Li G. , 2011: Assessments of Chinese FengYun microwave temperature sounder (MWTS) measurements for weather and climate applications. J. Atmos. Oceanic Technol., 28, 12061227.

    • Search Google Scholar
    • Export Citation
Save