• Chaboureau, J-P., , J-P. Cammas, , P. Mascart, , J-P. Pinty, , C. Claud, , R. Roca, , and J-J. Morcrette, 2000: Evaluation of a cloud system life-cycle simulated by Meso-NH during FASTEX using METEOSAT radiances and TOVS-3I cloud retrievals. Quart. J. Roy. Meteor. Soc, 126 , 17351750.

    • Search Google Scholar
    • Export Citation
  • Chevallier, F., , and J-J. Morcrette, 2000: Comparison of model fluxes with surface and top-of-the-atmosphere observations. Mon. Wea. Rev, 128 , 38393852.

    • Search Google Scholar
    • Export Citation
  • Chevallier, F., , P. Bauer, , G. Kelly, , C. Jakob, , and T. McNally, 2001: Model clouds over oceans as seen from space: Comparison with HIRS/2 and MSU radiances. J. Climate, 14 , 42164229.

    • Search Google Scholar
    • Export Citation
  • Courtier, P., , J-N. Thépaut, , and A. Hollingsworth, 1994: A strategy, for operational implementation of 4D-Var, using an incremental approach. Quart. J. Roy. Meteor. Soc, 120 , 13671388.

    • Search Google Scholar
    • Export Citation
  • Ebert, E. E., , and J. A. Curry, 1992: A parameterization of ice optical properties for climate models. J. Geophys. Res, 97D , 38313836.

    • Search Google Scholar
    • Export Citation
  • Eyre, J. R., 1991: A fast radiative transfer model for satellite sounding systems. ECMWF Tech. Memo. 176, 28 pp.

  • Fouquart, Y., , and B. Bonnel, 1980: Computation of solar heating of the earth's atmosphere: A new parameterization. Beitr. Phys. Atmos, 53 , 3562.

    • Search Google Scholar
    • Export Citation
  • Gregory, D., , J-J. Morcrette, , C. Jakob, , A. C. M. Beljaars, , and T. Stockdale, 2000: Revision of convection, radiation and cloud schemes in the ECMWF Integrated Forecasting System. Quart. J. Roy. Meteor. Soc, 126 , 16851710.

    • Search Google Scholar
    • Export Citation
  • Hortal, M., 1999: The development and testing of a new two-time-level semi-Lagrangian scheme (SETTLS) in the ECMWF forecast model. ECMWF Tech. Memo. 292, 17 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Jakob, C., 1999: Cloud cover in the ECMWF reanalysis. J. Climate, 12 , 947959.

  • Jakob, C., 2000: The representation of cloud cover in atmospheric general circulation models. Ph.D. thesis, Ludwig-Maximilians-Universität, Munich, Germany, 193 pp.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., , and S. A. Klein, 2000: A parametrization of the effects of cloud and precipitation overlap for use in general-circulation models. Quart. J. Roy. Meteor. Soc, 126 , 25252544.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., and and Coauthors, 2000: The IFS cycle CY21r4 made operational in October 1999. ECMWF Newsletter, Vol. 87, 2–9.

  • Janisková, M., 2001: Preparatory studies for the use of observations from the earth radiation mission in numerical weather prediction. Report from ESA Contract 13151/98/NL/GD, 79 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Joyce, R., , J. Janowiak, , and G. Huffman, 2001: Latitudinally and seasonally dependent zenith-angle corrections for geostationary satellite IR brightness temperatures. J. Appl. Meteor, 40 , 689703.

    • Search Google Scholar
    • Export Citation
  • Laprise, R., 1992: The resolution of global spectral models. Bull. Amer. Meteor. Soc, 73 , 14531454.

  • Lazzara, M. A., and and Coauthors, 1999: The Man computer Interactive Data Access System (McIDAS): 25 Years of Interactive Processing. Bull. Amer. Meteor. Soc, 80 , 271284.

    • Search Google Scholar
    • Export Citation
  • Mlawer, E. J., , S. J. Taubman, , P. D. Brown, , M. J. Iacono, , and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res, 102 , 1666316682.

    • Search Google Scholar
    • Export Citation
  • Morcrette, J-J., 1991: Evaluation of model-generated cloudiness: Satellite observed and model-generated diurnal variability and brightness temperature. Mon. Wea. Rev, 119 , 12051224.

    • Search Google Scholar
    • Export Citation
  • Morcrette, J-J., 2001: Assessment of the ECMWF model cloudiness and surface radiation fields at the ARM-SGP site. ECMWF Tech. Memo. 327. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Munro, R., , G. Kelly, , and R. Saunders, 2000: Assimilation of Meteosat radiance data within the 4DVAR system at ECMWF. EUMETSAT/ECMWF Fellowship Programme Rep. 8, 41 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., 1991: A recommended specific definition of “resolution.”. Bull. Amer. Meteor. Soc, 72 , 1914.

  • Räisänen, P., 1998: Effective longwave cloud fraction and maximum-random overlap clouds—A problem and a solution. Mon. Wea. Rev, 126 , 33363340.

    • Search Google Scholar
    • Export Citation
  • Rizzi, R., 1994: Raw HIRS/2 radiances and model simulations in the presence of clouds. ECMWF Tech. Memo. 73, 29 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Roca, R., , L. Picon, , M. Desbois, , H. Le Treut, , and J-J. Morcrette, 1997: Direct comparison of Meteosat water vapor channel data and general circulation model results. Geophys. Res. Lett, 24 , 147150.

    • Search Google Scholar
    • Export Citation
  • Rohn, M., , G. Kelly, , and R. W. Saunders, 2001: Impact of new cloud motion wind product from Meteosat on NWP analyses and forecasts. Mon. Wea. Rev, 129 , 23922403.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , and R. A. Schiffer, 1983: The International Satellite Cloud Climatology Project (ISCCP): The first project of the World Climate Research Program. Bull. Amer. Meteor. Soc, 64 , 779784.

    • Search Google Scholar
    • Export Citation
  • Saunders, R., , M. Matricardi, , and P. Brunel, 1999: An improved fast radiative transfer model for assimilation of satellite radiance observations. Quart. J. Roy. Meteor. Soc, 125 , 14071425.

    • Search Google Scholar
    • Export Citation
  • Schmetz, J., , K. Holmlund, , J. Hoffman, , B. Strauss, , B. Mason, , V. Gaertner, , A. Koch, , and L. van de Berg, 1993: Operational cloud-motion winds from Meteosat infrared images. J. Appl. Meteor, 32 , 12061225.

    • Search Google Scholar
    • Export Citation
  • Sherlock, V. J., 1999: ISEM-6: Infrared Surface Emissivity Model for RTTOV-6. Forecasting Research Tech. Rep. FR-299, Met Office, 17 pp. [Available from Met Office, London Road, Bracknell RG12 2SZ, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Smith, E. A., , and L. Shi, 1992: Surface forcing of the infrared cooling profile over the Tibetan plateau. Part I: Influence of relative longwave radiative heating at high altitude. J. Atmos. Sci, 49 , 805822.

    • Search Google Scholar
    • Export Citation
  • Soden, B., , and F. Bretherton, 1994: Upper-tropospheric relative humidity from the GOES 6.7 μm channel—Method and climatology for July 1987. J. Geophys. Res, 99 , 11871210.

    • Search Google Scholar
    • Export Citation
  • Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea. Rev, 121 , 30403061.

  • van de Berg, L. C. J., , J. Schmetz, , and J. Whitlock, 1995: On the calibration of the Meteosat water vapor channel. J. Geophys. Res, 100 , 2106921076.

    • Search Google Scholar
    • Export Citation
  • van den Hurk, B. J. J. M., , P. Viterbo, , A. C. M. Beljaars, , and A. K. Betts, 2000: Offline validation of the ERA40 surface scheme. ECMWF Tech. Memo. 295, 42 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Velden, C. S., , C. M. Hayden, , W. P. Menzel, , J. L. Franklin, , and J. S. Lynch, 1992: The impact of satellite-derived winds on numerical hurricane track forecasting. Wea. Forecasting, 7 , 107118.

    • Search Google Scholar
    • Export Citation
  • Velden, C. S., , S. J. Niemann, , W. P. Menzel, , and S. T. Wanzong, 1997: Upper-tropospheric winds derived from geostationary satellite water vapor observations. Bull. Amer. Meteor. Soc, 78 , 173195.

    • Search Google Scholar
    • Export Citation
  • Washington, W. M., , and D. L. Williamson, 1977: A description of the NCAR GCM. Methods in Computational Physics, J. Chang, Ed., Vol. 17, Academic Press, 111–172.

    • Search Google Scholar
    • Export Citation
  • Wylie, D. P., , W. P. Menzel, , H. M. Woolf, , and K. I. Strabala, 1994: Four years of global cirrus cloud statistics using HIRS. J. Climate, 7 , 19721986.

    • Search Google Scholar
    • Export Citation
  • Yu, W., , L. Garand, , and A. P. Dastoor, 1997: Evaluation of model clouds and radiation at 100 km scale using GOES data. Tellus, 49A , 246262.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    The horizontal wind field at 300 hPa from the ECMWF analyses: (a) Dec 2000 monthly mean and (b) at 1200 UTC 15 Dec 2000

  • View in gallery

    For Dec 2000, Meteosat-7 and -5 mean 11-μm brightness temperature

  • View in gallery

    For Dec 2000, (a) mean and (b) standard deviation of the model surface temperature, in K, in the Meteosat-7 area

  • View in gallery

    For Dec 2000, for each pixel p of a Meteosat picture, the distance (in km) to the first pixel p′ that is correlated in time by less than 0.90 to p is represented

  • View in gallery

    For Dec 2000, Meteosat-7 and -5 standard deviation of the 11-μm brightness temperature

  • View in gallery

    For Dec 2000, temporal correlation between the model-simulated and the observed 11-μm brightness temperatures

  • View in gallery

    For Dec 2000, brightness temperature (a) mean and (b) standard deviation as a function of the hour of the day in the 5°S–equator and 20°–25°E region (Zaire)

  • View in gallery

    For 10 Dec 2000 at 0300 UTC, simulated and observed Meteosat 11-μm images. The simulation was performed on the 3-h forecast. The grayscale is the same in both panels

  • View in gallery

    For 20 Dec 2000 at 0300 UTC, simulated and observed Meteosat 11-μm images. The simulation was performed on the 3-h forecast. In order to highlight the cloud patterns, the grayscale differs between the panels

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 15 15 1
PDF Downloads 8 8 0

Model Clouds as Seen from Space: Comparison with Geostationary Imagery in the 11-μm Window Channel

View More View Less
  • 1 European Centre for Medium-Range Weather Forecasts, Reading, Berkshire, United Kingdom
© Get Permissions
Full access

Abstract

A monitoring of the European Centre for Medium-Range Weather Forecasts (ECMWF) model with the Meteosat infrared window channel during December 2000 is presented. The Meteosat images are simulated every 3 h during the 31 days of the month from the model fields, at a resolution of 35 km. The study of both the spatial and the temporal variabilities of the model cloudiness is based on forecasts, from 3 to 48 h, as well as on analyses. Despite a reduced cloud forcing in the model, the variations of the extratropical cyclones are shown to be well represented in the short-range (up to 1 day) forecasts. The intertropical convergence zone (ITCZ) is well located but the representation of its temporal variations significantly differs from the observations, in particular over land. The variability of the Meteosat brightness temperature time series usually differs by more than 10% from one model grid point to another, whereas the structures described by the model have scales of about three to four grid points at least.

This work prepares the routine monitoring of the ECMWF analysis and forecast system with the raw images from Meteosat, soon to be replaced by the Meteosat Second Generation imager/sounder, and is an important step toward the assimilation of cloud-affected satellite radiances.

Corresponding author address: F. Chevallier, ECMWF, Shinfield Park, Reading, Berkshire RG29AX, United Kingdom. Email: f.chevallier@ecmwf.int

Abstract

A monitoring of the European Centre for Medium-Range Weather Forecasts (ECMWF) model with the Meteosat infrared window channel during December 2000 is presented. The Meteosat images are simulated every 3 h during the 31 days of the month from the model fields, at a resolution of 35 km. The study of both the spatial and the temporal variabilities of the model cloudiness is based on forecasts, from 3 to 48 h, as well as on analyses. Despite a reduced cloud forcing in the model, the variations of the extratropical cyclones are shown to be well represented in the short-range (up to 1 day) forecasts. The intertropical convergence zone (ITCZ) is well located but the representation of its temporal variations significantly differs from the observations, in particular over land. The variability of the Meteosat brightness temperature time series usually differs by more than 10% from one model grid point to another, whereas the structures described by the model have scales of about three to four grid points at least.

This work prepares the routine monitoring of the ECMWF analysis and forecast system with the raw images from Meteosat, soon to be replaced by the Meteosat Second Generation imager/sounder, and is an important step toward the assimilation of cloud-affected satellite radiances.

Corresponding author address: F. Chevallier, ECMWF, Shinfield Park, Reading, Berkshire RG29AX, United Kingdom. Email: f.chevallier@ecmwf.int

1. Introduction

Operational geostationary satellites provide high resolution (a few kilometers) and frequent (1 h or less) observations of the visible and infrared radiation that is emitted or reflected by a large portion of the earth–atmosphere system. In forecast centers, geostationary imagery is essential for nowcasting. Atmospheric motion winds derived from these data are routinely assimilated (Velden et al. 1992; Schmetz et al. 1993; Rohn et al. 2001) and the assimilation of clear-sky raw radiances will be operational in the near future (Munro et al. 2000). The validation of numerical weather prediction (NWP) systems also makes use of derived cloud products (Rossow and Schiffer 1983; Jakob 1999) and of raw radiances (Morcrette 1991; Yu et al. 1997). It has been emphasized in previous studies that, for validation as well as for assimilation, derived products may be ambiguous because they are to be understood as radiatively effective quantities, in the sense that they are defined as seen from the satellite (Velden et al. 1997; Chevallier et al. 2001), and as such they are preprocessed products with questionable errors associated with this processing. Therefore, an effort is made to develop the use of raw radiances.

The high spatial resolution of the geostationary images makes it necessary to perform some averaging when comparing to model data. The resolution of global models has been dramatically increased in recent years so that synoptic scales are taken into account more accurately. As an example, in late autumn 2000 the horizontal resolution of the European Centre for Medium-Range Weather Forecasts (ECMWF) operational forecasting system was increased from a TL319 spectral representation (about 60-km resolution) to a TL511 one (about 40-km resolution) on a global scale. Test versions are being run at TL799 (about 25-km resolution).

In this paper, infrared window (11 μm) images from the two Meteosat satellites operated by the European Organization for the Exploitation of Meteorological Satellites (EUMETSAT) are compared to the simulated images from the ECMWF system. The comparison focuses on the month of December 2000: the Meteosat images are simulated every 3 h. The data and the model are described in section 2. The Meteosat images allow the investigation of both the spatial and the temporal variabilities of the cloudiness in the ECMWF analyses and forecasts. The results are presented in section 3 and are discussed in section 4.

2. The data

a. The Meteosat observations

Two Meteosat satellites were operational during December 2000. Meteosat-7 was positioned at the latitude and longitude origin; Meteosat-5 was located above the Indian Ocean, at longitude 63°E on the equator. Their onboard radiometers measure radiation from the earth's disk every 30 min in three spectral bands: the visible from 0.45 to 1.0 μm, the infrared 5.7–7.1-μm water vapor absorption band, and the infrared window 10.5–12.5-μm band. This study makes use of the data from the infrared window channel, which is mainly sensitive to the vertical distribution of cloud amount and to surface characteristics (temperature and emissivity). The cloud signals that are investigated in the present study are significantly larger than the uncertainty caused by the calibration uncertainty [a few degrees K; e.g., van de Berg et al. 1995). No absolute intercalibration is performed. The resolution is degraded from 5 km to about 35 km in order to compare with the ECMWF model. This is done by remapping the native satellite images to an equal-angle latitude–longitude grid.

b. The model fields

The model version used in this study is the so-called cycle 23r3 of the ECMWF forecast system, which was operational in December 2000.

The assimilation system relies on the four-dimensional variational scheme (4DVAR) described by Courtier et al. (1994). Analyses are produced at nominal times of 0000 and 1200 UTC, with an assimilation window of 12 h. They use information about atmospheric pressure, temperature, water vapor, and winds from conventional and satellite measurements. The clouds are not analyzed, but the cloud fields from the 4DVAR final trajectory, therefore in balance with the analyzed fields, are archived and used here. Most of the results presented here concern the 31 days of December 2000, with a time resolution of 3 h. The times used and the corresponding forecast ranges are summarized in Table 1. Complementary computations, based on the analysis and on the 24- and 48-h forecasts, are also performed but at 1200 UTC only.

The forecast model is a global spectral TL511L60 model. It includes a semi-Lagrangian advection scheme together with a linear Gaussian grid (Hortal 1999). The reduced horizontal grid corresponds to a regular grid size of about 40 km from the equator to the poles. In the vertical, a hybrid coordinate of 60 levels between the surface and the top of the atmosphere is used by the global spectral forecast model. The physics package is based on Gregory et al. (2000). In particular, two prognostic equations describe the time evolution of cloud condensate and cloud cover (Tiedtke 1993), while rain and snow are separate diagnostic quantities. Clouds are formed by convection, diabatic cooling, and boundary layer turbulence. The scheme links their dissipation to adiabatic and diabatic heating, turbulent mixing of cloud air with unsaturated environmental air, and precipitation processes. The modifications to the original formulation from Tiedtke (1993) are given by Jakob and Klein (2000), Jakob (2000), and Jakob et al. (2000). The broadband radiation scheme includes the Rapid Radiative Transfer Model (RRTM: Mlawer et al. 1997) for the infrared and the Fouquart and Bonnel (1980) scheme (with four spectral bands) for the shortwave. At the surface, the Tiled ECMWF Scheme for Surface Exchanges over Land (TESSEL) described by van den Hurk et al. (2000) is used.

The Meteosat infrared window (11 μm) radiances are simulated from the model profiles of temperature, specific humidity, cloud cover, ice water, and liquid water, and from model surface variables (temperature and soil water). The method, which covers both infrared and microwave simulations, is described by Chevallier et al. (2001). It uses well-documented parameterizations that are summarized below. The gas absorption is obtained from the Radiative Transfer for Television and Infrared Observation Satellite (TIROS) Operational Vertical Sounder (RTTOV: Eyre 1991; Saunders et al. 1999). The radiances in the presence of semitransparent cloud layers are expressed as a linear combination of the clear-sky radiance and of the radiances in the presence of single-layered blackbodies (Washington and Williamson 1977). The coefficients of the linear combination are functions of the cloud cover and of the cloud emissivity on model levels. The cloud absorption is computed consistently with the ECMWF version of RRTM, with cloud absorption from Ebert and Curry (1992) for ice water and from Smith and Shi (1992) for liquid water. The maximum-random hypothesis defined by Räisänen (1998) is used to describe the overlap of the cloud layers. Scattering is not considered, which could lead to a slight overestimation of the brightness temperatures by the model, primarily for high thin clouds. Even though the satellite viewing angle is taken into account, the computed brightness temperatures may be overestimated for large zenith angles due to the neighboring pixel clouds that are not taken into account (Joyce et al. 2001). Moreover, the maximum-random hypothesis is also less reliable for large angles. In order to limit these effects, the image pixels for which the satellite zenith angle is more than 70° are discarded (except in Figs. 8 and 9). The estimation of surface emissivity differs from the ECMWF version of RRTM (Gregory et al. 2000). Emissivity is set to 0.99 and 0.93, respectively, on moist and dry land. The Infrared Surface Emissivity Model for RTTOV-6 (ISEM-6; Sherlock 1999) provides the sea emissivity.

c. Method of comparison

In the following, Meteosat images are compared to simulated images from the analyses and forecasts. All image manipulations are performed with the Man computer Interactive Data Access System (McIDAS) software (Lazzara et al. 1999). Statistics are computed in terms of mean, standard deviation, and correlation. The correlation between two series of N numbers {a(i)}i=1,N and {b(i)}i=1,N (here brightness temperatures), with means a and b, and standard deviations σa and σb, are given by
i1520-0493-130-3-712-e1
Two types of correlations are used: the series are either temporal (with i being the time index, with fixed spatial location) or spatial (with i being the space index, with fixed time).

The study focuses on cloud signals as seen from the infrared window channel. However the 11-μm radiances are also sensitive to other quantities, mainly surface temperature, surface emissivity, and water vapor continuum absorption. The uncertainty of the surface temperature in the model is of the order of 1 K over sea and is larger over land. In particular its diurnal cycle is not taken into account over sea, and may be underestimated over land (Morcrette 2001). The surface emissivity depends on the surface characteristics, like skin temperature, soil water content, vegetation coverage, and soil nature. Due to the uncertainty of those quantities, only a rough estimation of the emissivity is used here over land (section 2b). As far as the water vapor is concerned, radiative computations show that it impacts the 11-μm brightness temperature by about 3 K (9 K) when the total column water vapor is about 30 kg m−2 (60 kg m−2). The difficult representation of the humidity fields may affect the reproduction of this signal by the model computation. If the various errors accumulate, errors in the computation of the clear sky brightness temperatures may reach 10 K or more over land. This non-negligible uncertainty is important to keep in mind when analyzing the forthcoming results. As a consequence, an effort is made here to link the patterns of the differences with the atmosphere dynamics, in order to identify them. A parallel study at ECMWF concerns the clear-sky differences but is out of the scope of the present one.

3. Results

a. December 2000 main general circulation patterns

Figure 1 illustrates the structure of the atmospheric flow during December 2000. It represents the mean 300-hPa horizontal wind field (Fig. 1a) and the one on 15 December at 1200 UTC (Fig. 1b), which was analyzed by the ECMWF system. The northern Atlantic jet stream is oriented toward Spain. In Europe it snakes north of the Mediterranean Sea, then southward along the Arabian Peninsula, before strengthening in Southeast Asia. The subtropical jet streams vary around 20°N in the Northern Hemisphere and around 40°S in the Southern Hemisphere. The satellite images (e.g., Figs. 2c and 2d) indicate that the intertropical convergence zone (ITCZ) is located at the equator with a large southward extension over land.

b. Spatial variability

The mean values of the 11-μm brightness temperature for the month of December are presented in Fig. 2. By comparison, the mean model surface temperature in the Meteosat-7 area is shown in Fig. 3a. In the overlap region, the brightness temperatures from the two satellites differ by a few degrees Kelvin or less for instantaneous observations, because of differences in spectral response, calibration, and viewing angle. The general patterns of the Meteosat images are well reproduced by the ECMWF system: the ITCZ over sea as over land, the trade wind regions, the land-based convergence zones, and the midlatitude cyclone tracks. The intensity of the model patterns does not fit the observations so well, in particular over tropical land (Brazil, southern Africa, Sri Lanka, and Indonesia) where the ITCZ is strongly underestimated, with model brightness temperatures warmer than the observations by more than 12 K. Conversely, over tropical oceans, the ITCZ intensity is overestimated by about 8 K or more. West of Senegal, the differences are about 10 K, the model being warmer than the observations. In the midlatitude regions, the model is about 5 K warmer than the observations with larger differences over Europe: about 10 K. The agreement of the bias patterns with the cloud patterns, as well as its amplitude, suggests that it is caused by deficiencies of the cloud radiative forcing in the model. The cloud radiative forcing is defined as the difference between the clear-sky brightness temperature and the full-sky brightness temperature.

Focusing on Meteosat-7, Table 2 presents some statistics (mean and standard deviations) about the temporal evolution of the spatial correlation between the observed and the simulated images at 1200 UTC over an extended period of time: from 12 October to 31 December 2000. The correlations for the analyses vary between 0.7 and 0.8 over ocean, with a mean of 0.765 and a standard deviation of 0.025. Over land, the variations are stronger, between 0.6 and 0.9 with a standard deviation of 0.054, depending on the spatial structure of the cloudiness. All latitudes are taken into account in these statistics. They are mostly representative of midlatitude regions (poleward of 20°N and 20°S). When restricting the area of analysis to the 20°N–20°S band, correlations are poor: about 0.55 over sea as over land (not shown).

For the forecasts, Table 2 shows that the longer the range, the lower the correlation. Over the sea, the 12-h forecast mean correlation is 0.02 lower than the mean correlation using the analyses. The 24-h forecast mean correlation is 0.02 farther below the mean, which is still a limited degradation in the forecast quality on an average. When reaching the 48-h range, the mean correlation goes down to 0.68. The good quality of the forecasts up to the 24-h range is confirmed by direct examination of the images from both satellites (not shown). Over land, the degradation compared to the analysis correlation is stronger than over the sea for the 12- and 24-h ranges. Also the standard deviations are about twice as high, which makes the forecasts less reliable over land. Focusing on December 2000, the correlations using the 3-, 6-, and 9-h forecasts (see Table 1) consistently vary between the analysis and the 12-h correlations (not shown).

The Meteosat images allow validation of the scales of the spatial structures that the model describes. For each pixel of the Meteosat image (either simulated or observed), the nearest pixel that is correlated in time by less than 0.9 to that pixel is found. Figure 4 shows the distance to that nearest pixel for December 2000. In the following, the word pixel relates to the reduced-resolution (35 km) images (see section 2b). Over most regions, the observed images have distances of less than 40 km, which means that from one pixel to the next, the variability is usually more than 10%. In the midlatitudes, in the Northern Hemisphere subtropical jet stream, and in the land-based convergence zones, the distance is about 50 km. Higher values are seen in the desert areas. In the model-simulated images, the distances are about 90 km (i.e., about three pixels), with lower values (about 60 km) in the trade winds and southern Africa, and higher values (more than 140 km) in the desert areas. This result confirms that the effective resolution of a model is lower than its numerical resolution (Pielke 1991; Laprise 1992).

c. Temporal variability

The brightness temperature standard deviations are presented in Fig. 5. By comparison, the standard deviation of the model surface temperature in the Meteosat-7 area is shown in Fig. 3b: they are usually less than 4 K. The brightness temperature standard deviations are much higher in most regions. They are lower in the model than in the observations by about 4 K over most areas, confirming the lack of representation of atmospheric variability illustrated in the previous section. Larger negative differences are found over the land parts of the ITCZ, where the standard deviations of the model are about 10 K lower than those of the observations, west of Senegal, with negative differences up to minus 15 K, and in the Northern Hemisphere midlatitude disturbance regions, with negative differences of about 7 K. This is consistent with the underestimation of the cloud forcing that is shown in Fig. 2. Also, a model is expected to simulate an atmosphere smoother than the real one, because of the difficult representation of subgrid processes.

Figure 6 shows the spatial variation of the temporal correlation between the model and the observations. Largest correlations (above 0.8) are found over the regions with the least cloudiness, like the Sahara and Yemen, which seems to indicate a reasonable representation of the diurnal surface temperature cycle in these areas. Correlations above 0.7 are also located over midlatitude lands and seas, on the border of the ITCZ, west of Senegal, east of Brazil, and over Namibia. The ITCZ, the trade wind regions, and the high elevations in Asia, like the Himalayas and Mongolia, have low correlations. As an example of the model behavior in the ITCZ over land, Fig. 7 shows the brightness temperature mean and standard deviation as a function of the hour of the day in a 5° latitude by 5° longitude region over Zaire. Consistent with Fig. 2, the model mean brightness temperature appears to be overestimated (between 10 and 20 K). Its diurnal cycle is less pronounced than in the observations, with a minimum during the afternoon and the night due to both surface temperature and cloudiness variations. The model standard deviation has a low diurnal variability, whereas the observations have a significant diurnal cycle that exhibits the afternoon maximum of convection. To summarize, the model locates the ITCZ well, but inadequately represents the time variations of convection, in particular over land.

4. Discussion

As in previous works (e.g., Morcrette 1991; Soden and Bretherton 1994; Roca et al. 1997; Chaboureau et al. 2000), the present study emphasizes the use of geostationary satellites, like those of the Meteosat series, for the validation of numerical models of the atmosphere. The high spatial and temporal resolution of these instruments allows extensive statistics to be gathered, so that models can be quantitatively monitored. For instance, it was shown that, in the Meteosat data degraded to 35-km resolution, the variability of the brightness temperature time series usually differs by more than 10% from one grid point to another, whereas the structures described by the model have scales of about three to four grid points at least (Fig. 4). Also, forecast quality when going from a 3-h range to a 24-h range is very similar from the correlation point of view (Table 2). Unlike over the sea, the spatial correlation over land is significantly degraded between the analysis and the forecasts. Moreover, the temporal variation of convection in the Tropics is not well represented, in particular over land (Fig. 6). Improvements are needed in the modeling of the interaction between the soil, the planetary boundary layer, and the atmosphere.

Some of the results presented confirm previous validation studies of the top-of-the-atmosphere radiation in earlier versions of the ECMWF model (Rizzi 1994; Chevallier and Morcrette 2000; Chevallier et al. 2001). These works rely on different instruments, different versions of the model, and on different periods of time. In the extratropical cyclones, and to a lesser extent in the ITCZ, Chevallier et al. (2001) show that the model underestimates the radiative impact of high clouds, possibly because of insufficient cloud ice. As a result, the top-of-the-atmosphere longwave radiation is overestimated in the frontal regions and in the ITCZ (Fig. 2). In the oceanic ITCZ, the excessive frequency of cloud occurrence over time counterbalances this effect and leads to an underestimation of the top-of-the-atmosphere longwave radiation, when averaged over long periods of time, as is done here. The overestimation of the radiation west of Senegal, despite good temporal correlations, is a feature that was not noted before. The Atlantic and African ITCZs provide moisture for the generation of cirrus clouds. This cirrus outflow is insufficiently represented in the model. The misrepresentation of cirrus clouds in this region is confirmed by independent validation using the CO2-slicing technique (Wylie et al. 1994) to retrieve cloud-top pressure from the High-Resolution Infrared Radiation Sounder (HIRS) observations and from the model (not shown). The underestimation of the cloud forcing by the model partly explains the underestimation of the temporal variability of the top-of-the-atmosphere radiation (Fig. 5). Poleward of 30°S and 30°N, the relatively high correlations in Fig. 6 show that despite a weak cloud forcing in the model, the variations of the extratropical cyclones are well represented. This is illustrated in Figs. 8 and 9 for 10 December: the model reproduces the location of the midlatitude fronts very well (Fig. 9), even over land. In comparison, the representation of the ITCZ is poor, in particular over Kenya (Fig. 8).

The extension of the validation of the top-of-the-atmosphere radiation in NWP models is a consequence of the higher level of skill required from them, when they have to satisfy an extended range of applications from analysis to forecasting and to climate studies. It is planned to routinely use the diagnostics shown here for the monitoring of the ECMWF system, in addition to the current diagnostics based on the clear radiances (Munro et al. 2000). They are expected to help improve the model parameterizations. They will be extended to the other meteorological geostationary satellites. Additional information about the model will be diagnosed as new instruments become available such as the Spinning Enhanced Visible InfraRed Imager (SEVIRI) instrument, on board the Meteosat Second Generation (MSG), that combines the advantages of an imager and those of an infrared sounder. Finally, ongoing work aims at an objective feedback from the observed differences on the model through variational data assimilation of the cloud-affected radiances (Janisková 2001).

Acknowledgments

This work was done at the Satellite Application Facility on Numerical Weather Prediction, which is, cosponsored by EUMETSAT. The authors would like to thank J.-J. Morcrette and C. Köpken for fruitful discussions about radiation in general and the Meteosat observations in particular. J.-J. Morcrette, A. Simmons, and J.-N. Thépaut helped improve the initial version of the text.

REFERENCES

  • Chaboureau, J-P., , J-P. Cammas, , P. Mascart, , J-P. Pinty, , C. Claud, , R. Roca, , and J-J. Morcrette, 2000: Evaluation of a cloud system life-cycle simulated by Meso-NH during FASTEX using METEOSAT radiances and TOVS-3I cloud retrievals. Quart. J. Roy. Meteor. Soc, 126 , 17351750.

    • Search Google Scholar
    • Export Citation
  • Chevallier, F., , and J-J. Morcrette, 2000: Comparison of model fluxes with surface and top-of-the-atmosphere observations. Mon. Wea. Rev, 128 , 38393852.

    • Search Google Scholar
    • Export Citation
  • Chevallier, F., , P. Bauer, , G. Kelly, , C. Jakob, , and T. McNally, 2001: Model clouds over oceans as seen from space: Comparison with HIRS/2 and MSU radiances. J. Climate, 14 , 42164229.

    • Search Google Scholar
    • Export Citation
  • Courtier, P., , J-N. Thépaut, , and A. Hollingsworth, 1994: A strategy, for operational implementation of 4D-Var, using an incremental approach. Quart. J. Roy. Meteor. Soc, 120 , 13671388.

    • Search Google Scholar
    • Export Citation
  • Ebert, E. E., , and J. A. Curry, 1992: A parameterization of ice optical properties for climate models. J. Geophys. Res, 97D , 38313836.

    • Search Google Scholar
    • Export Citation
  • Eyre, J. R., 1991: A fast radiative transfer model for satellite sounding systems. ECMWF Tech. Memo. 176, 28 pp.

  • Fouquart, Y., , and B. Bonnel, 1980: Computation of solar heating of the earth's atmosphere: A new parameterization. Beitr. Phys. Atmos, 53 , 3562.

    • Search Google Scholar
    • Export Citation
  • Gregory, D., , J-J. Morcrette, , C. Jakob, , A. C. M. Beljaars, , and T. Stockdale, 2000: Revision of convection, radiation and cloud schemes in the ECMWF Integrated Forecasting System. Quart. J. Roy. Meteor. Soc, 126 , 16851710.

    • Search Google Scholar
    • Export Citation
  • Hortal, M., 1999: The development and testing of a new two-time-level semi-Lagrangian scheme (SETTLS) in the ECMWF forecast model. ECMWF Tech. Memo. 292, 17 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Jakob, C., 1999: Cloud cover in the ECMWF reanalysis. J. Climate, 12 , 947959.

  • Jakob, C., 2000: The representation of cloud cover in atmospheric general circulation models. Ph.D. thesis, Ludwig-Maximilians-Universität, Munich, Germany, 193 pp.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., , and S. A. Klein, 2000: A parametrization of the effects of cloud and precipitation overlap for use in general-circulation models. Quart. J. Roy. Meteor. Soc, 126 , 25252544.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., and and Coauthors, 2000: The IFS cycle CY21r4 made operational in October 1999. ECMWF Newsletter, Vol. 87, 2–9.

  • Janisková, M., 2001: Preparatory studies for the use of observations from the earth radiation mission in numerical weather prediction. Report from ESA Contract 13151/98/NL/GD, 79 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Joyce, R., , J. Janowiak, , and G. Huffman, 2001: Latitudinally and seasonally dependent zenith-angle corrections for geostationary satellite IR brightness temperatures. J. Appl. Meteor, 40 , 689703.

    • Search Google Scholar
    • Export Citation
  • Laprise, R., 1992: The resolution of global spectral models. Bull. Amer. Meteor. Soc, 73 , 14531454.

  • Lazzara, M. A., and and Coauthors, 1999: The Man computer Interactive Data Access System (McIDAS): 25 Years of Interactive Processing. Bull. Amer. Meteor. Soc, 80 , 271284.

    • Search Google Scholar
    • Export Citation
  • Mlawer, E. J., , S. J. Taubman, , P. D. Brown, , M. J. Iacono, , and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res, 102 , 1666316682.

    • Search Google Scholar
    • Export Citation
  • Morcrette, J-J., 1991: Evaluation of model-generated cloudiness: Satellite observed and model-generated diurnal variability and brightness temperature. Mon. Wea. Rev, 119 , 12051224.

    • Search Google Scholar
    • Export Citation
  • Morcrette, J-J., 2001: Assessment of the ECMWF model cloudiness and surface radiation fields at the ARM-SGP site. ECMWF Tech. Memo. 327. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Munro, R., , G. Kelly, , and R. Saunders, 2000: Assimilation of Meteosat radiance data within the 4DVAR system at ECMWF. EUMETSAT/ECMWF Fellowship Programme Rep. 8, 41 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., 1991: A recommended specific definition of “resolution.”. Bull. Amer. Meteor. Soc, 72 , 1914.

  • Räisänen, P., 1998: Effective longwave cloud fraction and maximum-random overlap clouds—A problem and a solution. Mon. Wea. Rev, 126 , 33363340.

    • Search Google Scholar
    • Export Citation
  • Rizzi, R., 1994: Raw HIRS/2 radiances and model simulations in the presence of clouds. ECMWF Tech. Memo. 73, 29 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Roca, R., , L. Picon, , M. Desbois, , H. Le Treut, , and J-J. Morcrette, 1997: Direct comparison of Meteosat water vapor channel data and general circulation model results. Geophys. Res. Lett, 24 , 147150.

    • Search Google Scholar
    • Export Citation
  • Rohn, M., , G. Kelly, , and R. W. Saunders, 2001: Impact of new cloud motion wind product from Meteosat on NWP analyses and forecasts. Mon. Wea. Rev, 129 , 23922403.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , and R. A. Schiffer, 1983: The International Satellite Cloud Climatology Project (ISCCP): The first project of the World Climate Research Program. Bull. Amer. Meteor. Soc, 64 , 779784.

    • Search Google Scholar
    • Export Citation
  • Saunders, R., , M. Matricardi, , and P. Brunel, 1999: An improved fast radiative transfer model for assimilation of satellite radiance observations. Quart. J. Roy. Meteor. Soc, 125 , 14071425.

    • Search Google Scholar
    • Export Citation
  • Schmetz, J., , K. Holmlund, , J. Hoffman, , B. Strauss, , B. Mason, , V. Gaertner, , A. Koch, , and L. van de Berg, 1993: Operational cloud-motion winds from Meteosat infrared images. J. Appl. Meteor, 32 , 12061225.

    • Search Google Scholar
    • Export Citation
  • Sherlock, V. J., 1999: ISEM-6: Infrared Surface Emissivity Model for RTTOV-6. Forecasting Research Tech. Rep. FR-299, Met Office, 17 pp. [Available from Met Office, London Road, Bracknell RG12 2SZ, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Smith, E. A., , and L. Shi, 1992: Surface forcing of the infrared cooling profile over the Tibetan plateau. Part I: Influence of relative longwave radiative heating at high altitude. J. Atmos. Sci, 49 , 805822.

    • Search Google Scholar
    • Export Citation
  • Soden, B., , and F. Bretherton, 1994: Upper-tropospheric relative humidity from the GOES 6.7 μm channel—Method and climatology for July 1987. J. Geophys. Res, 99 , 11871210.

    • Search Google Scholar
    • Export Citation
  • Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea. Rev, 121 , 30403061.

  • van de Berg, L. C. J., , J. Schmetz, , and J. Whitlock, 1995: On the calibration of the Meteosat water vapor channel. J. Geophys. Res, 100 , 2106921076.

    • Search Google Scholar
    • Export Citation
  • van den Hurk, B. J. J. M., , P. Viterbo, , A. C. M. Beljaars, , and A. K. Betts, 2000: Offline validation of the ERA40 surface scheme. ECMWF Tech. Memo. 295, 42 pp. [Available from ECMWF, Shinfield Park, Reading, Berkshire RG2 9AX, United Kingdom.].

    • Search Google Scholar
    • Export Citation
  • Velden, C. S., , C. M. Hayden, , W. P. Menzel, , J. L. Franklin, , and J. S. Lynch, 1992: The impact of satellite-derived winds on numerical hurricane track forecasting. Wea. Forecasting, 7 , 107118.

    • Search Google Scholar
    • Export Citation
  • Velden, C. S., , S. J. Niemann, , W. P. Menzel, , and S. T. Wanzong, 1997: Upper-tropospheric winds derived from geostationary satellite water vapor observations. Bull. Amer. Meteor. Soc, 78 , 173195.

    • Search Google Scholar
    • Export Citation
  • Washington, W. M., , and D. L. Williamson, 1977: A description of the NCAR GCM. Methods in Computational Physics, J. Chang, Ed., Vol. 17, Academic Press, 111–172.

    • Search Google Scholar
    • Export Citation
  • Wylie, D. P., , W. P. Menzel, , H. M. Woolf, , and K. I. Strabala, 1994: Four years of global cirrus cloud statistics using HIRS. J. Climate, 7 , 19721986.

    • Search Google Scholar
    • Export Citation
  • Yu, W., , L. Garand, , and A. P. Dastoor, 1997: Evaluation of model clouds and radiation at 100 km scale using GOES data. Tellus, 49A , 246262.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

The horizontal wind field at 300 hPa from the ECMWF analyses: (a) Dec 2000 monthly mean and (b) at 1200 UTC 15 Dec 2000

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Fig. 2.
Fig. 2.

For Dec 2000, Meteosat-7 and -5 mean 11-μm brightness temperature

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Fig. 3.
Fig. 3.

For Dec 2000, (a) mean and (b) standard deviation of the model surface temperature, in K, in the Meteosat-7 area

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Fig. 4.
Fig. 4.

For Dec 2000, for each pixel p of a Meteosat picture, the distance (in km) to the first pixel p′ that is correlated in time by less than 0.90 to p is represented

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Fig. 5.
Fig. 5.

For Dec 2000, Meteosat-7 and -5 standard deviation of the 11-μm brightness temperature

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Fig. 6.
Fig. 6.

For Dec 2000, temporal correlation between the model-simulated and the observed 11-μm brightness temperatures

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Fig. 7.
Fig. 7.

For Dec 2000, brightness temperature (a) mean and (b) standard deviation as a function of the hour of the day in the 5°S–equator and 20°–25°E region (Zaire)

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Fig. 8.
Fig. 8.

For 10 Dec 2000 at 0300 UTC, simulated and observed Meteosat 11-μm images. The simulation was performed on the 3-h forecast. The grayscale is the same in both panels

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Fig. 9.
Fig. 9.

For 20 Dec 2000 at 0300 UTC, simulated and observed Meteosat 11-μm images. The simulation was performed on the 3-h forecast. In order to highlight the cloud patterns, the grayscale differs between the panels

Citation: Monthly Weather Review 130, 3; 10.1175/1520-0493(2002)130<0712:MCASFS>2.0.CO;2

Table 1.

Forecast ranges used for the simulation of the Meteosat images during Dec 2000

Table 1.
Table 2.

For Meteosat-7, at 1200 UTC. Mean and standard deviations of the time series of the spatial correlations between the simulated images and the observations from 12 Oct to 31 Dec 2000. Distinction is made between land and sea geotypes. The simulations use the analysis, the 12-h, 24-h, or 48-h forecasts (respectively, AN, FC12, FC24, and FC48)

Table 2.
Save