• Adler, R. F., , and D. D. Fenn, 1979: Thunderstorm intensity as determined from satellite data. J. Appl. Meteor., 18, 502517.

  • Breon, F. M., , D. L. Jackson, , and J. J. Bates, 2000: Calibration of the Meteosat water vapor channel using collocated NOAA/HIRS 12 measurements. J. Geophys. Res., 105 (D9), 11 92511 933.

    • Search Google Scholar
    • Export Citation
  • Brogniez, H., , R. Roca, , and L. Picon, 2009: Study of the free tropospheric humidity interannual variability using Meteosat data and an advection-condensation transport model. J. Climate, 22, 67736787.

    • Search Google Scholar
    • Export Citation
  • Burpee, R. W., 1972: The origin and structure of easterly waves in the lower troposphere of North Africa. J. Atmos. Sci., 29, 7790.

  • Clough, S. A., , M. W. Shephard, , E. J. Mlawer, , J. S. Delamere, , M. J. Iacono, , K. Cady-Pereira, , S. Boukabara, , and P. D. Brown, 2005: Atmospheric radiative transfer modeling: A summary of the AER codes. J. Quant. Spectrosc. Radiat. Transfer, 91, 233244.

    • Search Google Scholar
    • Export Citation
  • Desbois, M., , T. Kayiranga, , B. Gnamien, , S. Guessous, , and L. Picon, 1988: Characterization of some elements of the Sahelian climate and their annual variations for July 1983, 1984, and 1985 from the analysis of METEOSAT ISCCP data. J. Climate, 1, 867904.

    • Search Google Scholar
    • Export Citation
  • Grist, J. P., , and S. E. Nicholson, 2001: A study of the dynamic factors influencing the rainfall variability in the West African Sahel. J. Climate, 14, 13371359.

    • Search Google Scholar
    • Export Citation
  • Hastenrath, S., 2000: Interannual and longer term variability of upper-air circulation over the tropical Atlantic and West Africa in boreal summer. Int. J. Climatol., 20, 14151430.

    • Search Google Scholar
    • Export Citation
  • Heidinger, A. K., , C. O’Dell, , R. Bennartz, , and T. Greenwald, 2006: The successive-order-of-interaction radiative transfer model. Part I: Model development. J. Appl. Meteor. Climatol., 45, 13881402.

    • Search Google Scholar
    • Export Citation
  • Hewison, T., , and M. König, 2008: Inter-calibration of Meteosat imagers and IASI. Proc. 2008 EUMETSAT Meteorological Satellite Conf., Darmstadt, Germany, EUMETSAT, 52–59.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., and Coauthors, 1997: The Global Precipitation Climatology Project (GPCP) combined precipitation dataset. Bull. Amer. Meteor. Soc., 78, 520.

    • Search Google Scholar
    • Export Citation
  • Kossin, J. P., , S. J. Camargo, , and M. Sitkowski, 2010: Climate modulation of North Atlantic hurricane tracks. J. Climate, 23, 30573076.

    • Search Google Scholar
    • Export Citation
  • Laing, A. G., , J. M. Fritsch, , and A. J. Negri, 1999: Contribution of mesoscale convective complexes to rainfall in Sahelian Africa: Estimates from geostationary infrared and passive microwave data. J. Appl. Meteor., 38, 957964.

    • Search Google Scholar
    • Export Citation
  • Laurent, H., , N. D’Amato, , and T. Lebel, 1998: How important is the contribution of the mesoscale convective complexes to the Sahelian rainfall? Phys. Chem. Earth, 23, 629633.

    • Search Google Scholar
    • Export Citation
  • Machado, L. A. T., , W. B. Rossow, , R. L. Guedes, , and A. W. Walker, 1998: Life cycle variations of mesoscale convective systems over the Americas. Mon. Wea. Rev., 126, 16301654.

    • Search Google Scholar
    • Export Citation
  • Mathon, V., , and H. Laurent, 2001: Life cycle of Sahelian mesoscale convective cloud systems. Quart. J. Roy. Meteor. Soc., 127, 377406.

    • Search Google Scholar
    • Export Citation
  • Matthews, A. J., 2004: Intraseasonal variability over tropical Africa during northern summer. J. Climate, 17, 24272440.

  • Nicholson, S., 2005: On the question of the “recovery” of the rains in the West African Sahel. J. Arid Environ., 63, 615641.

  • Nicholson, S., 2009: A revised picture of the structure of the “monsoon” and land ITCZ over West Africa. Climate Dyn., 32, 11551171.

    • Search Google Scholar
    • Export Citation
  • Nicholson, S., , and J. P. Grist, 2001: A conceptual model for understanding rainfall variability in the West African Sahel on interannual and interdecadal timescales. Int. J. Climatol., 21, 17331757.

    • Search Google Scholar
    • Export Citation
  • Picon, L., , R. Roca, , S. Serrar, , J. L. Monge, , and M. Desbois, 2003: A new METEOSAT “water vapor” archive for climate studies. J. Geophys. Res., 108, 4301, doi:10.1029/2002JD002640.

    • Search Google Scholar
    • Export Citation
  • Reed, R. J., , D. C. Norquist, , and E. E. Recker, 1977: Structure and properties of African wave disturbances as observed during phase III of GATE. Mon. Wea. Rev., 105, 317333.

    • Search Google Scholar
    • Export Citation
  • Roca, R., , J. P. Lafore, , C. Piriou, , and J. L. Redelsperger, 2005: Extratropical dry-air intrusions into the West African monsoon midtroposphere: An important factor for the convective activity over the Sahel. J. Atmos. Sci., 62, 390407.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , and R. A. Schiffer, 1991: ISCCP cloud data products. Bull. Amer. Meteor. Soc., 72, 220.

  • Rossow, W. B., , and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80, 22612287.

  • Rudolf, B., , and U. Schneider, 2005: Calculation of gridded precipitation data for the global land-surface using in-situ gauge observations. Proc. Second Workshop of the International Precipitation Working Group IPWG, Darmstadt, Germany, EUMETSAT, 231–247. [Available from Am Kavalleriesand 31, D-64295 Darmstadt, Germany.]

    • Search Google Scholar
    • Export Citation
  • Schroeder, M., , M. König, , and J. Schmetz, 2009: Deep convection observed by the Spinning Enhanced Visible and Infrared Imager on board Meteosat 8: Spatial distribution and temporal evolution over Africa in summer and winter 2006. J. Geophys. Res., 114, D05109, doi:10.1029/2008JD010653.

    • Search Google Scholar
    • Export Citation
  • Schulz, J., and Coauthors, 2009: Operational climate monitoring from space: The Satellite Application Facility on Climate Monitoring. Atmos. Chem. Phys., 9, 16871709.

    • Search Google Scholar
    • Export Citation
  • Sohn, B. J., , J. Schmetz, , S. Tjemkes, , M. Koenig, , H. Lutz, , A. Arriaga, , and E. S. Chung, 2000: Intercalibration of the Meteosat-7 water vapor channel with SSM/T-2. J. Geophys. Res., 105, 15 67315 680.

    • Search Google Scholar
    • Export Citation
  • Thorncroft, C., , and K. Hodges, 2001: African easterly wave variability and its relationship to Atlantic tropical cyclone activity. J. Climate, 14, 11661179.

    • Search Google Scholar
    • Export Citation
  • Tian, B. J., , B. J. Soden, , and X. Q. Wu, 2004: Diurnal cycle of convection, clouds, and water vapor in the tropical upper troposphere: Satellites versus a general circulation model. J. Geophys. Res., 109, D10101, doi:10.1029/2003JD004117.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., 2002: Changes in tropical clouds and radiation. Science, 296, 2095.

  • Trenberth, K. E., and Coauthors, 2007: Observations: Surface and atmospheric climate change. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 235–336.

    • Search Google Scholar
    • Export Citation
  • Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 29613012.

  • Waliser, D. E., , and M. Moncrieff, 2008: The Year of Tropical Convection (YOTC) Science Plan: A joint WCRP–WWRP/THORPEX international initiative. WMO Tech. Doc. 1452, WCRP 130, WWRP/THORPEX 9, Geneva, Switzerland, 34 pp.

    • Search Google Scholar
    • Export Citation
  • Williams, M., , and R. A. Houze, 1987: Satellite-observed characteristics of winter monsoon cloud clusters. Mon. Wea. Rev., 115, 505519.

    • Search Google Scholar
    • Export Citation
  • Zipser, E. J., , D. J. Cecil, , C. T. Liu, , S. W. Nesbitt, , and D. P. Yorty, 2006: Where are the most intense thunderstorms on earth? Bull. Amer. Meteor. Soc., 87, 10571071.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    (top) Atmospheric transmission and (bottom) corresponding simulated brightness temperatures for a midlatitude summer atmosphere (water vapor column amount: 29.3 kg m−2, surface temperature: about 294 K) at 60° zenith angle. The different colored lines in the top panel show the different MVIRI infrared channel filter functions used for Meteosat-2–7. For plotting purposes filter functions are normalized to a peak value of one.

  • View in gallery

    The three areas selected for the homogenization study (AREA 1–3). The Sahel area (SAH) used later on in the study is highlighted as well.

  • View in gallery

    Comparison of observed and simulated brightness temperature time series for Meteosat-2–7. Results are provided separately for the three comparison areas. The comparison is restricted to the months of July–September. The colored dots are observed area-mean daily 1200 UTC brightness temperatures. The corresponding colored lines provide the time-averaged observed brightness temperature for each satellite. The corresponding black dots and lines give the simulation results.

  • View in gallery

    Mean features of convective systems derived from 20 yr of MVIRI observations. Data are binned in 1° × 1° boxes according to their origin. Values are provided for the (a) total number of convective events identified, (b) coldest temperature reached during system lifetime (mean over all systems in box), (c) mean system velocity, and (d) mean lifetime of the systems. (right) Corresponding zonal mean plots for two longitude bands (60°–15°W and 15°W–10°E). The line between 10° and 20°N is the average position of the 600-hPa zonal wind maximum for the 20-yr period (from NCEP reanalysis).

  • View in gallery

    Tracks of individual convective events. (top) The actual tracks of all convective events that lasted longer than 2 days. Starting points are the blue dots, end points the red dots. The size of the blue dots gives an indication of the lifetime of the individual events. (bottom) A conceptualization of four different origins (open blue ellipses) of convection and preferred direction of the tracks (solid arrows). Also shown are the levels of the 925-, 600-, and 200-hPa easterly wind maxima and the position of 0 meridional wind velocity at 925 and 200 hPa from NCEP reanalysis. For details see text.

  • View in gallery

    Long-term statistics for convection over the Sahel region at 1900 local solar time (LST). (top) The mean area covered by convection. (middle) The number of convective cells. (bottom) The mean temperature of the cells. The red curves show results with homogenization, the black curves without homogenization. The dashed lines give results of linear regression and the straight solid lines show the long-term average.

  • View in gallery

    (top) Sahel rainfall derived from GPCC gridded reanalysis. Precipitation anomalies (middle) vs total convective area and (bottom) vs number of convective events. For plotting purposes the time series in the middle and bottom panels are normalized by subtracting the mean value and dividing by the standard deviation. The correlation between precipitation anomalies and convective area is 0.66, and between precipitation anomalies and number of convective events is 0.76.

  • View in gallery

    Anomaly patterns of convection for the five low–Sahel rainfall years (see Table 3) compared to the entire 20-yr period (see Fig. 4). The black line between 10° and 20°N is the position of the 600-hPa zonal wind maximum for the five low-rain years (from NCEP reanalysis).

  • View in gallery

    As in Fig. 8, but for five high–Sahel rainfall years.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 10 10 5
PDF Downloads 6 6 2

Convective Activity over Africa and the Tropical Atlantic Inferred from 20 Years of Geostationary Meteosat Infrared Observations

View More View Less
  • 1 Atmospheric and Oceanic Sciences, University of Wisconsin—Madison, Madison, Wisconsin
  • 2 Satellite Application Facility on Climate Monitoring, Deutscher Wetterdienst, Offenbach, Germany
© Get Permissions
Full access

Abstract

A 20-yr (1986–2005) time series of Meteosat Visible and Infrared Imager (MVIRI) geostationary infrared observations was used to study deep convection over Africa and the tropical Atlantic. The 20-yr time period is covered by six consecutive satellites (Meteosat-2–7). To correct for possible systematic differences between instruments on the different satellite platforms, a time series of Meteosat infrared observations over cloud-free ocean surfaces was compared to reanalysis-based radiative transfer results. Based on the comparison of simulations with observations, a homogenization was performed for the MVIRI infrared channel. The homogenized 20-yr dataset was then subjected to a tracking analysis for deep convection over Africa and the tropical Atlantic for the boreal summer months of July–September.

The mean state of convection as well as anomalies for high– and low–Sahel rainfall years were studied. Comparisons with the Global Precipitation Climatology Center’s (GPCC) rainfall estimates were performed for the Sahel region and interannual variability was evaluated comparing convection for the five driest and five wettest Sahel years. Results support earlier findings that precipitation in the Sahel region is strongly linked to the latitudinal position of the African Easterly Jet with deep convection being triggered more strongly if the jet is displaced northward. A relationship between the jet position and long-lived convective systems over the tropical Atlantic was found as well.

Corresponding author address: Ralf Bennartz, Atmospheric and Oceanic Sciences, University of Wisconsin—Madison, Madison, WI 53706. E-mail: bennartz@aos.wisc.edu

Abstract

A 20-yr (1986–2005) time series of Meteosat Visible and Infrared Imager (MVIRI) geostationary infrared observations was used to study deep convection over Africa and the tropical Atlantic. The 20-yr time period is covered by six consecutive satellites (Meteosat-2–7). To correct for possible systematic differences between instruments on the different satellite platforms, a time series of Meteosat infrared observations over cloud-free ocean surfaces was compared to reanalysis-based radiative transfer results. Based on the comparison of simulations with observations, a homogenization was performed for the MVIRI infrared channel. The homogenized 20-yr dataset was then subjected to a tracking analysis for deep convection over Africa and the tropical Atlantic for the boreal summer months of July–September.

The mean state of convection as well as anomalies for high– and low–Sahel rainfall years were studied. Comparisons with the Global Precipitation Climatology Center’s (GPCC) rainfall estimates were performed for the Sahel region and interannual variability was evaluated comparing convection for the five driest and five wettest Sahel years. Results support earlier findings that precipitation in the Sahel region is strongly linked to the latitudinal position of the African Easterly Jet with deep convection being triggered more strongly if the jet is displaced northward. A relationship between the jet position and long-lived convective systems over the tropical Atlantic was found as well.

Corresponding author address: Ralf Bennartz, Atmospheric and Oceanic Sciences, University of Wisconsin—Madison, Madison, WI 53706. E-mail: bennartz@aos.wisc.edu

1. Introduction

Tropical convection plays an important role in the hydrological and energetic balance of the tropics and subtropics. Trends in time or shifts–expansions in space would affect the water and energy balance of large regions of the globe. Tropical convection over the African continent has experienced increased interest with ongoing discussions linking continental convective activity and hurricane activity (Thorncroft and Hodges 2001, and references therein), the effect of ENSO and Indian Ocean sea surface temperature on precipitation in East Africa (Trenberth et al. 2007), the effect of waves triggered by the Madden–Julian oscillation on precipitation in East and West Africa (Matthews 2004), and the strong precipitation anomalies over the Sahel. Tropical convection is also one focus of research, modeling, and forecasting activities in the framework of the year of tropical convection (Waliser and Moncrieff 2008). Africa is also one of the key regions of interest of the Coordinated Regional Climate Downscaling Experiment (CORDEX; http://wcrp.ipsl.jussieu.fr/SF_RCD_CORDEX.html). Initiated by the World Climate Research Programme, its objective is to “develop a framework to evaluate and possibly improve RCD (regional climate downscaling) techniques for use in downscaling global climate projections.”

Trenberth et al. (2007) state that “the decreasing rainfall and devastating droughts in the Sahel region during the last three decades of the 20th century … are among the largest climate changes anywhere.” The boreal summer months are particularly important since precipitation in the Sahel peaks in those months. In the Sahel, mesoscale convective systems (i.e., systems with a size larger than 5000 km2) predominantly move westward (Mathon and Laurent 2001) and are the dominant source of precipitation [Laurent et al. 1998; Laing et al. 1999, using Tropical Rainfall Measuring Mission (TRMM) data]. Based on comparisons between wet and dry months of July 1983–85, Desbois et al. (1988) observed a more northward position of squall lines in the Sahel during the wet years. An increased likelihood of fast-moving and long-lasting convective systems was observed by Roca et al. (2005) in the presence of extratropical dry intrusions. Nicholson (2009) introduced a new picture of the land ITCZ structure over western Africa decoupling the tropical rain belt from the ITCZ [see Fig. 18 of Nicholson (2009), also showing the position of the Tropical and African Easterly Jets (AEJ)]. Looking at individual convective systems using TRMM data, Zipser et al. (2006) defined storm intensities by height, minimum microwave brightness temperatures, and lightning rate. Their analysis over the global tropics revealed that the most intense storms occur frequently in semiarid areas, like the Sahel, while heavy rain events such as over Indonesia are often not associated with strong storms.

Most long-term analyses of tropical convection and jet–wave activities are generally based on reanalyses or long-term, ground-based observations (e.g., rain gauge networks). Geostationary satellite datasets with sufficient spatial and temporal resolution to resolve convective activity are only available since the mid-1980s. Within this study we use the 20-yr time series of Meteosat Visible and Infrared Imager (MVIRI) observations to study deep convection over Africa and the tropical Atlantic. We evaluate the entire MVIRI data record (boreal summer only) using a tracking system for deep convection (Schroeder et al. 2009), which allows convective events to be followed in time and space. In this publication the following parameters of individual convective events are assessed quantitatively: origin, lifetime, track, and maximum intensity (as measured by minimum temperatures obtained during the lifetime of individual convective events). Using this dataset, we evaluate the mean state of convection as well as anomalies over Africa and the tropical Atlantic. The evaluation is carried out also for high– and low–Sahel rainfall years.

Any long-term analyses of tropical convection using satellite observations require carefully homogenized time series from satellite (e.g., Trenberth 2002). Creating such homogenized datasets is a major effort and requires significant resources. Recently, several programs have been initiated to generate such datasets [e.g., the European Organisation for the Exploitation of Meteorological Satellites’ (EUMETSAT) Satellite Application Facility on Climate Monitoring (Schulz et al. 2009) or the World Meteorological Organization’s Global Space-Based Intercalibration System (GSICS)]. Regarding MVIRI, the first attempts were made to homogenize/calibrate MVIRI water vapor observations at 6.3 μm by Sohn et al. (2000) for the period 1983–94, by Picon et al. (2003) using International Satellite Cloud Climatology Project (ISCCP-B3) data, and more recently by Brogniez et al. (2009). Their approach corrects not only for effects related to platform and instrument changes, but also for calibration events. A first homogenization of the 6.3-μm channel using HIRS water vapor observations was carried out by Breon et al. (2000).

In contrast, the study presented here focuses on the 11-μm infrared (IR) window channel rather than the 6.3-μm water vapor channel. While the Meteosat time series has already received significant attention for climate studies [e.g., as an integral part of ISCCP (Rossow and Schiffer 1991, 1999)], we perform an independent assessment of this homogenization. Despite the lack of an absolute calibration reference, especially for the early part of the time series, and although earlier studies have emphasized homogenization as well (e.g., ISCCP), the growing demand for well-understood, long-term datasets justifies revisiting this crucial topic. In addition, the MVIRI data are used at its native spatial and temporal resolution in this publication.

The remainder of this paper is organized as follows: section 2 describes the satellite dataset and the homogenization approach. Section 3 describes the tracking algorithm. In section 4 the mean state of convection over the entire period and the tropical part of the Meteosat disc is evaluated. Section 5 provides an analysis related to rainfall in the Sahel region.

2. MVIRI observations and homogenization

a. MVIRI data

The Meteosat Visible and Infrared Imager is a three-channel imaging radiometer flown consecutively on Meteosat-2–7. It covers Earth from a geostationary orbit every 30 min. Coverage at 0° latitude is between 1982 and 2006. Here only the MVIRI infrared channel observing in the infrared window roughly centered around 11 μm is used. The spatial resolution of individual observations is around 5 km at nadir. Data and coincident calibration coefficients are available for 20 yr from 1986 to 2005. MVIRI counts were converted to radiances using the operational calibration coefficients available from EUMETSAT’s Web site. Radiance to brightness temperature conversions were performed using lookup tables provided by EUMETSAT (M. König 2010, personal communication). MVIRI infrared filter functions were available partly from the EUMETSAT Web site (Meteosat-5–7) and partly from the International Satellite Cloud Climatology’s Web site (Rossow and Schiffer 1991, 1999; Meteosat-2–4). EUMETSAT also recommends using the Meteosat-7 filter function for Meteosat-5 and -6. The filter functions are in general not believed to be well-known, thus complicating the absolute calibration of the different MVIRIs. This effect of uncertainties in filter functions on simulated brightness temperatures is discussed in the modeling section.

b. Radiative transfer module

The radiative transfer module presented here consists of two components. First, a lookup table of gas absorption coefficients tabulated for different pressures, temperatures, and absorber masses. The lookup table is based on optical depths calculated using the line-by-line radiative transfer model (LBLRTM; Clough et al. 2005). The second component is a simple nonscattering radiative transfer module solving Schwarzschild’s integral radiative transfer equation and simulating MVIRI brightness temperatures.

1) LBLRTM-based lookup table

The lookup table used within the forward radiative transfer is based on LBLRTM version 11.6. Monochromatic gas absorption coefficients were calculated separately for water vapor, carbon dioxide, and other gases including continuum absorption. Table 1 provides an overview of the various parameters.

Table 1.

Variables and their corresponding ranges, step-widths, and resulting dimensions in the lookup tables. For each absorber the lookup tables provide the volume absorption coefficient and are 4D depending on temperature, pressure, wavenumber, and absorber mass. Note that in this table wavenumber is used as the spectral dimension (consistent with LBLRTM).

Table 1.

If temperature, pressure, and relative humidity (carbon dioxide mixing ratios) are provided, the lookup table can be interpolated in four dimensions to obtain individual absorption coefficients for water vapor (carbon dioxide). For carbon dioxide, only values at 385 ppmv were provided and a linear scaling of the resulting absorption coefficient with the actual carbon dioxide concentration is performed, effectively reducing the dimension of the lookup table for carbon dioxide by one. In contrast, coefficients are interpolated in relative humidity space for water vapor absorption. Other gases are considered using climatological mean values. To obtain the total monochromatic gas absorption coefficient of a given layer, the three absorption coefficients for water vapor, carbon dioxide, and other gases are added.

2) Nonscattering forward radiative transfer

The radiative transfer module interfaces with the lookup table so that for given temperatures, pressures, and absorber amounts monochromatic gas absorption coefficients can be derived. Based on those coefficients and for a given atmospheric profile the monochromatic Schwarzschild equation is solved for a cloud-free atmosphere using a prescribed, but potentially spectrally dependent, surface emissivity.

It is assumed that the reflected downwelling radiation at the surface obeys specular reflection. This simplification can be justified by the relatively large surface emissivity of water (about 0.985) and the relatively low atmospheric transmission in the atmospheric window (of around 0.8 at nadir for a midlatitude standard atmosphere). The vertical integration of the radiative transfer equation is performed with brute force, assuming the optical properties as well as the source function to be constant throughout each layer. This assumption could be relaxed with little additional cost using a vertically variable thermal source but is uncritical for the infrared window (e.g., Heidinger et al. 2006). It will, however, introduce systematic errors if applied to other more opaque spectral regions depending on the narrowness of the weighting functions and the slope of the temperature profile. Figure 1 shows an example of the so-derived upwelling monochromatic brightness temperatures and one-way transmittances for a midlatitude summer atmosphere and an observation zenith angle of 60°.

Fig. 1.
Fig. 1.

(top) Atmospheric transmission and (bottom) corresponding simulated brightness temperatures for a midlatitude summer atmosphere (water vapor column amount: 29.3 kg m−2, surface temperature: about 294 K) at 60° zenith angle. The different colored lines in the top panel show the different MVIRI infrared channel filter functions used for Meteosat-2–7. For plotting purposes filter functions are normalized to a peak value of one.

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

MVIRI observes broadband radiances convolved over its filter function (see colored lines in Fig. 1). Thus, after monochromatic radiative transfer has been performed, the simulated monochromatic radiances have to be convolved to MVIRI resolution and converted back to equivalent brightness temperatures.

For the example presented in Fig. 1, the equivalent brightness temperatures are 288.98 K (Meteosat-2–3), 289.15 K (Meteosat-4), and 288.75 K (Meteosat-5–7). The spread between brightness temperatures for the different filter functions thus is on the order of 0.5 K. This number also gives a rough estimate of the order of magnitude of errors introduced by the incomplete knowledge of the actual shape of the filter functions for the MVIRIs from the different satellites.

c. Homogenization of Meteosat-2–7 MVIRI

Because of known differences in the different sensors’ filter function, the homogenization of the different MVIRIs is clearly critical. In addition, the filter functions for most MVIRIs are not necessarily characterized well and systematic differences and/or trends in MVIRI calibration have been reported by other studies (Hewison and König 2008). The approach followed relies on comparisons of forward radiative transfer results based on 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40; Uppala et al. 2005) meteorological fields with MVIRI observations. For this purpose, three comparison areas over the Atlantic Ocean were defined as shown in Fig. 2. For the months of July–September and for each of those areas individually, the warmest 10 observed brightness temperatures at 1200 UTC were selected. Each of the areas selected holds approximately 50 000 individual observations for each observation time, so that the warmest 10 pixels correspond to roughly the warmest 0.02% of the observations within each area at any given time. This stringent a priori selection serves the purpose of excluding possible cloud-contaminated observations. This approach will inevitably produce a positive bias in the observations entering the comparison, simply because preferably warm pixels are sampled. Thus, only the relative calibration between different MVIRIs can be evaluated, while the absolute calibration might not be reasonably assessed using this method.

Fig. 2.
Fig. 2.

The three areas selected for the homogenization study (AREA 1–3). The Sahel area (SAH) used later on in the study is highlighted as well.

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

For each of the 10 individual observations, a collocation with ERA-40 profiles of temperature, relative humidity, and pressure is performed, radiative transfer simulations are carried out, and comparisons between observations and simulations are performed. Note that the ERA-40 time series ends in 2002, so the years 2003–06 could not be taken into account for the comparison.

Results of these comparisons are shown separately for the three comparison areas in Fig. 3 and Table 2. With few exceptions, the observed brightness temperatures are warmer than the simulations. This might reflect the aforementioned potential positive bias introduced by the data screening. Indeed, more elaborate comparisons between Meteosat-7 and Infrared Atmospheric Sounding Interferometer (IASI) as well as the High Resolution Infrared Radiation Sounder (HIRS) performed in the framework of the Global Space-Based Intercalibration System suggest that Meteosat-7 is actually biased negatively (Hewison and König 2008) rather than positively. Despite this apparent inconsistency in absolute calibration, two important questions can be answered using the approach pursued here. First, the homogenization of the different MVIRIs can be assessed. Second, potentially artificial trends can be identified for each of the individual instrument’s time series.

Fig. 3.
Fig. 3.

Comparison of observed and simulated brightness temperature time series for Meteosat-2–7. Results are provided separately for the three comparison areas. The comparison is restricted to the months of July–September. The colored dots are observed area-mean daily 1200 UTC brightness temperatures. The corresponding colored lines provide the time-averaged observed brightness temperature for each satellite. The corresponding black dots and lines give the simulation results.

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

Table 2.

Mean observation–simulation bias and slope of regression of the observation–simulation difference vs time (averaged over all three areas). Values are given with respect to monthly means. Slope values for MVIRI on Meteosat-3 and Meteosat-6 are not provided because of the short service time of these two instruments.

Table 2.

The mean biases (observations, simulations) over all three areas are given in Table 2. Biases increase almost linearly and range from +2.98 K for Meteosat-2 to +1.29 K for Meteosat-7, introducing a negative trend in the (up to this point) uncorrected time series of observed brightness temperatures. This apparent trend is also visible for each of the selected areas individually by comparing the differences between respective colored lines with the black line in Fig. 3. A similar analysis performed using the monthly median values led to virtually identical results (not shown).

In addition to these differences between different MVIRIs, potential calibration drifts within each MVIRI time series were also investigated. Meteosat-3 and Meteosat-6 were excluded from this particular analysis since they only provide 1 and 2 yr of data, respectively (see Fig. 3). For each of the other MVIRIs, a linear trend analysis was performed on the difference between observations and simulations. The resulting regression slopes are listed in Table 2. Different pictures emerge for the different satellites with trends between −0.3 and +0.2 K yr−1, resulting in maximum linear trends of about 1 K over the maximum 5-yr service time of each satellite. A t test reveals that none of the trends are statistically significant at a 95% confidence level.

The results presented suggest that differences between different instruments are significant. While the analysis performed here cannot fully rule out potential impacts of calibration drifts within the lifetime of individual instruments, they appear to not have a major impact on the entire time series. Based on the results presented here, the following simple approach is used to homogenize the different MVIRIs. The bias values given in Table 2 are renormalized to a mean value of zero by subtracting the mean over all six bias values. The so-derived values are then subtracted from the brightness temperatures of the individual sensors. This leads to corrections on the order of −0.9 K for Meteosat-2 and about +0.9 K for Meteosat-7. Note that this simple correction does not attempt to improve or correct the absolute calibration, but rather only the homogenization between the different MVIRIs. An assessment of the absolute calibration of the different MVIRIs is not within the scope of the method mainly because of the a priori selection of observations.

In addition to the analysis performed here, the impact of satellite homogenization on convective tracking results and detection of deep convection is also further discussed below. It will be shown that the satellite homogenization performed based on warm, cloud-free observations will lead to a more stable time series for deep convection as well.

3. Tracking of deep convection

The general approach used for Lagrangian tracking of deep convection is based on the work of Schroeder et al. (2009) using a maximum overlap analysis between consecutive images as proposed by Williams and Houze (1987). Consistent with the work of Schroeder et al. (2009), we use a temperature threshold of 230 K in the MVIRI infrared channel to identify convective activity. The chosen threshold is identical to the studies by Tian et al. (2004) and is similar to other commonly used values [e.g., 226 and 235 K in Adler and Fenn (1979) and Machado et al. (1998), respectively]. The area size of a convective system has to exceed 100 pixels. In addition, extremely short-lived events with a lifetime smaller than 3 h are excluded from the dataset. Splitting and merging of convective events is monitored as well. In total, 28 parameters are tracked over the lifetime of a convective event. Out of these parameters we focus here on the minimum temperature, the area size, the system velocity, and the lifetime. Data are analyzed on a monthly basis. To avoid the misclassification of systems generated at the end of the previous month or systems entering the next month, the first and last 3 days of each month are ignored.

Data were available for 20 yr (1986–2005). This study focuses on the months of July–September. Subsequently, we evaluate the mean state of convection over Africa and the tropical Atlantic before potential trends on minimum temperature and size in the Sahel region.

4. Mean features derived from 20 yr of data

Figure 4 shows the 20-yr climatology of summertime convection over tropical Africa and the tropical North Atlantic. The data are binned so that every system identified is binned into the 1° × 1° grid box in which it was first identified (i.e., where the above selection criteria on size and minimum temperature were first met). Given the 1° × 1° gridbox size, the first detection can be regarded as the origin of the convective system with good accuracy.

Fig. 4.
Fig. 4.

Mean features of convective systems derived from 20 yr of MVIRI observations. Data are binned in 1° × 1° boxes according to their origin. Values are provided for the (a) total number of convective events identified, (b) coldest temperature reached during system lifetime (mean over all systems in box), (c) mean system velocity, and (d) mean lifetime of the systems. (right) Corresponding zonal mean plots for two longitude bands (60°–15°W and 15°W–10°E). The line between 10° and 20°N is the average position of the 600-hPa zonal wind maximum for the 20-yr period (from NCEP reanalysis).

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

Accordingly, Fig. 4a provides the total number of convective events found. Over tropical Africa a clear preference for the development of convective systems can be found for the Guinea highlands, the coast of Nigeria and Cameroon, the Darfur region, as well as for various mountain ranges, including the Cameroon Mountains and the Jos Plateau. Those areas have already been identified in earlier work as focal points for initiation of convection (e.g., Schroeder et al. 2009). Over the tropical Atlantic, convection is mostly generated in a latitude band between 5° and 15°N, roughly coinciding with the southern part of the Atlantic hurricane development region, where typically more intensive and long-lived hurricanes form (see, e.g., cluster 3 in Kossin et al. 2010). The systems formed over continental Africa reach on average much lower temperatures than those formed over the Atlantic (Fig. 4b), with minimum temperatures on average below 200 K for most of the continental areas and around 205 K for the Atlantic convective systems. The coldest convective systems are mainly generated around and north of a line defined by the 600-hPa zonal wind maximum (indicating the position of the AEJ), where the lifetime for ocean systems also maximizes. Significant differences between convection over land and over ocean are found in the velocity of the systems. The highest velocities are found for the northernmost continental systems around 15°N. Over the African continent, more southerly systems are typically associated with lower velocities. A north–south gradient is not apparent for the systems initiated over the Atlantic Ocean.

The high velocities found over the continent are potentially associated with the AEJ and resulting disturbances (African Easterly Waves) (Grist and Nicholson 2001; Hastenrath 2000; Thorncroft and Hodges 2001). Two distinct wave tracks, south and north of the AEJ, have already been identified in the early work by Burpee (1972) and Reed et al. (1977). The vast majority of systems identified form south of the climatological jet position, consistent with earlier findings (summarized, e.g., in Nicholson 2009) that precipitation is associated mostly with wave disturbances south of the position of the AEJ. Waves forming to the north of the jet near 20°N are typically confined to the lower troposphere and not associated with precipitation. The technique used here relies on cold, deep convective clouds, and cannot identify disturbances that are not associated with such clouds.

Figure 5 shows trajectories of the longest-living system that could be tracked for more than 48 h. Within the entire 20-yr period, 106 events were found in the dataset, corresponding to about two systems per month on average. This analysis shows four main regions where those long-lasting convective events are generated. Each of those regions is associated with a preferred directional motion of the storms.

Fig. 5.
Fig. 5.

Tracks of individual convective events. (top) The actual tracks of all convective events that lasted longer than 2 days. Starting points are the blue dots, end points the red dots. The size of the blue dots gives an indication of the lifetime of the individual events. (bottom) A conceptualization of four different origins (open blue ellipses) of convection and preferred direction of the tracks (solid arrows). Also shown are the levels of the 925-, 600-, and 200-hPa easterly wind maxima and the position of 0 meridional wind velocity at 925 and 200 hPa from NCEP reanalysis. For details see text.

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

Over the African continent storms are initiated in a region north of the climatological position of the African Easterly Jet, which is identified in the plots by the 600-hPa wind maximum. We stress the word climatological here, because the occurrence of these intense systems is typically associated with a significant northward shift of the jet position as shown in section 5d. The systems typically move toward the west with a slight southerly component crossing the climatological jet position and ending at latitudes south of 10°N. They typically do not move out onto the Atlantic Ocean beyond about 20°W. The strong alignment of these systems with the position of the African Easterly Jet suggests that those systems might be initiated by jet instabilities.

Over the Atlantic Ocean, long-lasting convective systems are generated west of 20°W and north of about 8°N, corresponding well with the definition of the hurricane development region. Two different processes seem to be relevant in generating those systems. First, along 15°N, system generation appears to still be aligned with the position of the 600-hPa wind maximum, indicating a relation with the African Easterly Jet. The systems generated farther south align well with the position of zero meridional velocity at 925 and 200 hPa. These lines roughly separate convergent boundary layer flow and divergent upper-tropospheric flow, respectively. A distinction between systems generated along the 600-hPa jet and systems generated in association with lower-tropospheric convergence was also identified earlier by Thorncroft and Hodges (2001), purely based on model data. The longest-lasting systems appear to be generated near 15°N.

The systems generated over the Atlantic can be separated into roughly three classes depending on their preferred direction of movement. Systems initiated south of about 10°N typically tend to move westward with a slight southerly component, whereas systems generated north of 10°N have a tendency to move northward. Whether or not a system curves northward appears to be related to the upper-tropospheric meridional flow υ. Systems south of the 200-hPa υ = 0 line (black, dashed) have a tendency to move southward, whereas systems generated north of this line appear to curve northward. The systems generated north of this line can be separated in two clusters, one in the western and one in the eastern Atlantic. For systems generated west of about 38°W, the northward component is stronger than for systems generated east of about 38°W. Both of these areas fall within cluster 3 described by Kossin et al. (2010) and show a large number of extremely long-lived convective systems possibly associated with hurricane activity.

Another interesting feature is the lack of observed convection over the African continent around 20°N, where the near-surface meridional flow converges (gray, dashed curve). As pointed out by Nicholson (2009), this is an area of surface convergence and the center of the African heat low. In this area deep convection is efficiently suppressed by upper-level subsidence.

5. Trends and interannual variability for the Sahel region (10°–20°N, 15°W–10°E)

In this investigation we study potential trends and interannual variability in a sensitive and important region, namely the sub-Saharan Sahel region. The boundaries of the region can be identified in Fig. 2.

a. Variability and correlation with rainfall estimates

Figure 6 shows the time series of convective activity for the Sahel region derived from the entire 20-yr dataset. To account for diurnal cycle effects, only data for 1900 local solar time are shown in Fig. 6 (as well as Fig. 7 and Table 3). The following three parameters were extracted in Fig. 6: first, the average area covered by a convective event (top panel), second, the total number of cells (middle panel), and third, the mean temperature of the convective systems (bottom panel).

Fig. 6.
Fig. 6.

Long-term statistics for convection over the Sahel region at 1900 local solar time (LST). (top) The mean area covered by convection. (middle) The number of convective cells. (bottom) The mean temperature of the cells. The red curves show results with homogenization, the black curves without homogenization. The dashed lines give results of linear regression and the straight solid lines show the long-term average.

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

Fig. 7.
Fig. 7.

(top) Sahel rainfall derived from GPCC gridded reanalysis. Precipitation anomalies (middle) vs total convective area and (bottom) vs number of convective events. For plotting purposes the time series in the middle and bottom panels are normalized by subtracting the mean value and dividing by the standard deviation. The correlation between precipitation anomalies and convective area is 0.66, and between precipitation anomalies and number of convective events is 0.76.

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

Table 3.

Statistics for mean values and regression slopes for the Sahel region (corresponding to Fig. 6). The mean values correspond to the solid lines in the plots. The slopes correspond to the dashed lines in the plots. Slopes are presented in percent relative to the corresponding mean values. The numbers given in parentheses after the slope values are the percent t-test likelihood values of the slopes being statistically significant. The last two rows give the 5 yr with the highest and lowest rainfall anomalies within the 20-yr period based on the GPCC’s (Rudolf and Schneider 2005) gridded rain gauge analysis.

Table 3.

The mean values corresponding to Fig. 6, regression slopes, and corresponding statistical significances are given in Table 3. While interannual variability is large, none of the regressions is statistically significant at a 95% level. The highest significance level is found for the number of convective cells (78%), which appears to increase slightly over the course of the 20 yr. The lack of statistical significance together with the comparably short time series precludes any further trend analysis of the dataset. Also, interannual variability is about a factor of 10 larger compared to any of the trends over the entire time period.

b. Impact of satellite homogenization on tracking

The homogenization of the different satellites was performed using cloud-free, warm observations from the reference areas in Fig. 2. Here we investigate the impact of the homogenization on the time series of deep convection shown in Fig. 6. For this purpose, we have rerun the entire tracking for all 3 months and all years without homogenization.

The results for the Sahel area are shown in Fig. 6. Without homogenization all three quantities exhibit a trend, while only the maximum number of cells shows an appreciable trend when the homogenization is performed. Most importantly, if homogenization is neglected, trends become apparent in convective area (+4.0% decade−1) and in minimum temperature (−0.4% decade−1) and are statistically significant at the 67% and 92% levels, in contrast to the values given in Table 3 for the corrected dataset.

Especially since long-term absolute calibration references for cold targets are not available, these results highlight once more that trend analyses are difficult and results have to be interpreted with caution. In fact, none of the trends in this study are found to be statistically significant at a 95% level. While this purely statistical argument does not preclude the existence of any trends, the homogenization does reduce trends apparent in the uncorrected dataset, suggesting that actual trends are small compared to the interannual variability.

c. Comparison with the GPCC

Figure 7 relates the number of cells as well as the total convective area (mean size of individual cells × number of cells) to the total rainfall amount in the Sahel region as obtained from the Global Precipitation Climatology Center’s (GPCC) gridded rain gauge reanalysis (Rudolf and Schneider 2005). This dataset purely relies on independent but sometimes sparse (especially in sub-Saharan Africa; see, e.g., Nicholson 2005) rain gauges. A moderate correlation between the observed convective activity and rainfall estimates from GPCC is found. This is reassuring but not surprising. In fact, merged satellite–gauge analyses such as the Global Precipitation Climatology Project (GPCP; Huffman et al. 1997) use convective activity derived from infrared observations to estimate rain amount. Subsequently, we study convective anomalies for the five driest and five wettest years (according to the top panel of Fig. 7), which are listed also in Table 3.

d. Convective anomalies for dry and wet years

Based on the GPCC rainfall estimates we compare anomalies in convection for the five driest and the five wettest years in the dataset (see Table 3 for the particular years). Figure 8 shows results for the five low-rainfall years and Fig. 9 anomalies for the five high-rainfall years. The five lowest (highest) rainfall years exhibit on average 13% less (19% more) rain compared to the average over the entire 20-yr period. In dry Sahel years, convection generation is suppressed over Africa in a latitude band around 15°N and between roughly 15°W and 10°E. Over Cameroon and in coastal regions near the Gulf of Guinea, convection increases (see Fig. 8, top panel). For the high-rainfall years, convection over Africa increases throughout, except for a narrow zonal band at around 8°N.

Fig. 8.
Fig. 8.

Anomaly patterns of convection for the five low–Sahel rainfall years (see Table 3) compared to the entire 20-yr period (see Fig. 4). The black line between 10° and 20°N is the position of the 600-hPa zonal wind maximum for the five low-rain years (from NCEP reanalysis).

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

Fig. 9.
Fig. 9.

As in Fig. 8, but for five high–Sahel rainfall years.

Citation: Journal of Climate 25, 1; 10.1175/2011JCLI3984.1

Nicholson and Grist (2001) provide a conceptual model of rainfall over western Africa and identify four modes of precipitation depending on anomalies in the Sahel and the coastal regions near the western and eastern sides of the Gulf of Guinea. The patterns observed here for the wet Sahel years clearly correspond to the “Sahel Mode,” whereas the patterns for the dry years appear to fit with the “Guinea Coast Mode (Wet Phase)” (see Fig. 2 in Nicholson and Grist 2001). According to the analysis of Grist and Nicholson (2001), the latitudinal position of the AEJ and associated instabilities are possibly the most important factors governing Sahel rainfall. They find a northward shift of the AEJ in wet years during summer (jet location at around 17°N for July/August) and a corresponding southward shift for the dry years (at around 11°N for July/August).

Our own analysis of National Centers for Environmental Prediction (NCEP) reanalysis data confirms this shift for the dry versus wet years, although to a lesser extent. The peak latitudes of the 600-hPa zonal wind fields over the Sahel were 12°N (15°N) for dry (wet) years (see lines overlaid in Figs. 8 and 9). The generation region for the intense systems previously shown in Fig. 5 was located north of the climatological mean jet position. The northward shift of the AEJ during wet years (when most of the intense systems occur) suggests that these systems still form south of the actual (rather than climatological) position of the AEJ. A visual analysis of wind fields for individual days with strong development supports this finding. Stronger systems therefore appear to form if the AEJ is displaced northward, but they tend to occur south of the actual position of the jet.

Our analysis further suggests that over Africa in years with positive rain anomalies, convection generation is higher in a region of about ±5° around the AEJ center position. In low-rain anomaly years a different pattern emerges, with convection suppressed in the core region of the AEJ between 15°E and 10°W. At the same time convection of the southern part of the Gulf of Guinea appears enhanced. The minimum convective temperature anomalies over Africa appear to reverse between low- and high-rain anomaly years (Figs. 8 and 9, second panel from top). For high-rain anomaly years, convective temperature anomalies north (south) of the AEJ position are colder (warmer), suggesting more (less) intense convection in the Sahel region. This pattern is reversed for the low-rain anomaly years.

The velocity fields (Figs. 8 and 9, row 3) show bands of increased easterly velocities over Africa at around 10°N (low-rain years) and at 18°–20°N (high-rain years), possibly directly coinciding with the jet position. Hastenrath (2000) also finds a weakening of the Tropical Easterly Jet (around 200 hPA and 5°–8°N) for dry Sahel years, which we could identify in the NCEP reanalysis as well (not shown).

Figures 8 and 9 reveal one more highly interesting feature: a strong change in convective activity between low– and high–Sahel rainfall years does not only occur over Africa but over the entire tropical Atlantic where latitudinal dipole and tripole patterns in convective activity are observed. In dry Sahel years, convection over the Atlantic around 8°N (15°N) appears to be suppressed (enhanced). In wet years convection around 10°N is suppressed but increases are found around 15°N as well as around 5°N.

6. Conclusions

We have reexamined 20 yr of MVIRI geostationary satellite observations to study convection over Africa and the tropical Atlantic. We find the homogenized time series of convective activity and minimum convective temperatures to be remarkably stable over time. Potential trends are not statistically significant and about an order of magnitude smaller than interannual variability. Comparing reanalysis-derived positions of zonal wind maxima and meridional convergence positions with tracks of convective systems, four different regions of convective initiation could be identified. Over tropical Africa, as well as the tropical Atlantic, the longest-living convective systems are initiated near the African Easterly Jet position. Over the Atlantic, the position of the 200-hPa meridional divergence maximum appears to separate systems with a tendency to curve northward from those that move south.

For the Sahel region we compare satellite-based results with GPCC-derived rainfall anomalies. Anomalies are highly correlated with general convective activity observed from our dataset, also giving further confidence in the quality of the derived dataset. Consistent with earlier studies, the position of the African Easterly Jet appears to be crucially important to understanding Sahel rainfall anomalies. The results found here suggest that a shift of the jet position and resulting shifts in precipitation are not the only factors affecting Sahel precipitation. In addition to the shift, our analysis suggests that when the jet is shifted northward, not only the position of the convection changes accordingly but the intensity and frequency of convection increase as well. Similarly, a southward shift of the jet appears to be associated with generally decreased convective activity, which could possibly be caused by decreased wave activity in the jet zone during dry years.

Various issues could not be addressed and are left for future studies. In particular, the descriptive nature of the study does not allow for drawing new conclusions on the interrelation and causalities between the different phenomena observed. Further investigations are needed to better understand relations between tropical convection initiation, dust, precipitation, hurricane frequency, and other parameters. The contribution of this study is to highlight a dataset that might act as a direct link between many of these parameters and has so far not been explored to the fullest extent possible.

Acknowledgments

This work was carried out during a visiting scientist stay of the first author at EUMETSAT Headquarters and at the EUMETSAT’s Satellite Application Facility on Climate Monitoring. Both are gratefully acknowledged for their hospitality and support of this work. The authors are particularly indebted to Johannes Schmetz, Byung-Ju Sohn, Marianne König, Tim Hewison, Mark Kulie, Joerg Schulz, and Stephen Tjemkes for many helpful scientific discussions. We would also like to acknowledge the excellent contributions of two anonymous reviewers.

REFERENCES

  • Adler, R. F., , and D. D. Fenn, 1979: Thunderstorm intensity as determined from satellite data. J. Appl. Meteor., 18, 502517.

  • Breon, F. M., , D. L. Jackson, , and J. J. Bates, 2000: Calibration of the Meteosat water vapor channel using collocated NOAA/HIRS 12 measurements. J. Geophys. Res., 105 (D9), 11 92511 933.

    • Search Google Scholar
    • Export Citation
  • Brogniez, H., , R. Roca, , and L. Picon, 2009: Study of the free tropospheric humidity interannual variability using Meteosat data and an advection-condensation transport model. J. Climate, 22, 67736787.

    • Search Google Scholar
    • Export Citation
  • Burpee, R. W., 1972: The origin and structure of easterly waves in the lower troposphere of North Africa. J. Atmos. Sci., 29, 7790.

  • Clough, S. A., , M. W. Shephard, , E. J. Mlawer, , J. S. Delamere, , M. J. Iacono, , K. Cady-Pereira, , S. Boukabara, , and P. D. Brown, 2005: Atmospheric radiative transfer modeling: A summary of the AER codes. J. Quant. Spectrosc. Radiat. Transfer, 91, 233244.

    • Search Google Scholar
    • Export Citation
  • Desbois, M., , T. Kayiranga, , B. Gnamien, , S. Guessous, , and L. Picon, 1988: Characterization of some elements of the Sahelian climate and their annual variations for July 1983, 1984, and 1985 from the analysis of METEOSAT ISCCP data. J. Climate, 1, 867904.

    • Search Google Scholar
    • Export Citation
  • Grist, J. P., , and S. E. Nicholson, 2001: A study of the dynamic factors influencing the rainfall variability in the West African Sahel. J. Climate, 14, 13371359.

    • Search Google Scholar
    • Export Citation
  • Hastenrath, S., 2000: Interannual and longer term variability of upper-air circulation over the tropical Atlantic and West Africa in boreal summer. Int. J. Climatol., 20, 14151430.

    • Search Google Scholar
    • Export Citation
  • Heidinger, A. K., , C. O’Dell, , R. Bennartz, , and T. Greenwald, 2006: The successive-order-of-interaction radiative transfer model. Part I: Model development. J. Appl. Meteor. Climatol., 45, 13881402.

    • Search Google Scholar
    • Export Citation
  • Hewison, T., , and M. König, 2008: Inter-calibration of Meteosat imagers and IASI. Proc. 2008 EUMETSAT Meteorological Satellite Conf., Darmstadt, Germany, EUMETSAT, 52–59.

    • Search Google Scholar
    • Export Citation
  • Huffman, G. J., and Coauthors, 1997: The Global Precipitation Climatology Project (GPCP) combined precipitation dataset. Bull. Amer. Meteor. Soc., 78, 520.

    • Search Google Scholar
    • Export Citation
  • Kossin, J. P., , S. J. Camargo, , and M. Sitkowski, 2010: Climate modulation of North Atlantic hurricane tracks. J. Climate, 23, 30573076.

    • Search Google Scholar
    • Export Citation
  • Laing, A. G., , J. M. Fritsch, , and A. J. Negri, 1999: Contribution of mesoscale convective complexes to rainfall in Sahelian Africa: Estimates from geostationary infrared and passive microwave data. J. Appl. Meteor., 38, 957964.

    • Search Google Scholar
    • Export Citation
  • Laurent, H., , N. D’Amato, , and T. Lebel, 1998: How important is the contribution of the mesoscale convective complexes to the Sahelian rainfall? Phys. Chem. Earth, 23, 629633.

    • Search Google Scholar
    • Export Citation
  • Machado, L. A. T., , W. B. Rossow, , R. L. Guedes, , and A. W. Walker, 1998: Life cycle variations of mesoscale convective systems over the Americas. Mon. Wea. Rev., 126, 16301654.

    • Search Google Scholar
    • Export Citation
  • Mathon, V., , and H. Laurent, 2001: Life cycle of Sahelian mesoscale convective cloud systems. Quart. J. Roy. Meteor. Soc., 127, 377406.

    • Search Google Scholar
    • Export Citation
  • Matthews, A. J., 2004: Intraseasonal variability over tropical Africa during northern summer. J. Climate, 17, 24272440.

  • Nicholson, S., 2005: On the question of the “recovery” of the rains in the West African Sahel. J. Arid Environ., 63, 615641.

  • Nicholson, S., 2009: A revised picture of the structure of the “monsoon” and land ITCZ over West Africa. Climate Dyn., 32, 11551171.

    • Search Google Scholar
    • Export Citation
  • Nicholson, S., , and J. P. Grist, 2001: A conceptual model for understanding rainfall variability in the West African Sahel on interannual and interdecadal timescales. Int. J. Climatol., 21, 17331757.

    • Search Google Scholar
    • Export Citation
  • Picon, L., , R. Roca, , S. Serrar, , J. L. Monge, , and M. Desbois, 2003: A new METEOSAT “water vapor” archive for climate studies. J. Geophys. Res., 108, 4301, doi:10.1029/2002JD002640.

    • Search Google Scholar
    • Export Citation
  • Reed, R. J., , D. C. Norquist, , and E. E. Recker, 1977: Structure and properties of African wave disturbances as observed during phase III of GATE. Mon. Wea. Rev., 105, 317333.

    • Search Google Scholar
    • Export Citation
  • Roca, R., , J. P. Lafore, , C. Piriou, , and J. L. Redelsperger, 2005: Extratropical dry-air intrusions into the West African monsoon midtroposphere: An important factor for the convective activity over the Sahel. J. Atmos. Sci., 62, 390407.

    • Search Google Scholar
    • Export Citation
  • Rossow, W. B., , and R. A. Schiffer, 1991: ISCCP cloud data products. Bull. Amer. Meteor. Soc., 72, 220.

  • Rossow, W. B., , and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80, 22612287.

  • Rudolf, B., , and U. Schneider, 2005: Calculation of gridded precipitation data for the global land-surface using in-situ gauge observations. Proc. Second Workshop of the International Precipitation Working Group IPWG, Darmstadt, Germany, EUMETSAT, 231–247. [Available from Am Kavalleriesand 31, D-64295 Darmstadt, Germany.]

    • Search Google Scholar
    • Export Citation
  • Schroeder, M., , M. König, , and J. Schmetz, 2009: Deep convection observed by the Spinning Enhanced Visible and Infrared Imager on board Meteosat 8: Spatial distribution and temporal evolution over Africa in summer and winter 2006. J. Geophys. Res., 114, D05109, doi:10.1029/2008JD010653.

    • Search Google Scholar
    • Export Citation
  • Schulz, J., and Coauthors, 2009: Operational climate monitoring from space: The Satellite Application Facility on Climate Monitoring. Atmos. Chem. Phys., 9, 16871709.

    • Search Google Scholar
    • Export Citation
  • Sohn, B. J., , J. Schmetz, , S. Tjemkes, , M. Koenig, , H. Lutz, , A. Arriaga, , and E. S. Chung, 2000: Intercalibration of the Meteosat-7 water vapor channel with SSM/T-2. J. Geophys. Res., 105, 15 67315 680.

    • Search Google Scholar
    • Export Citation
  • Thorncroft, C., , and K. Hodges, 2001: African easterly wave variability and its relationship to Atlantic tropical cyclone activity. J. Climate, 14, 11661179.

    • Search Google Scholar
    • Export Citation
  • Tian, B. J., , B. J. Soden, , and X. Q. Wu, 2004: Diurnal cycle of convection, clouds, and water vapor in the tropical upper troposphere: Satellites versus a general circulation model. J. Geophys. Res., 109, D10101, doi:10.1029/2003JD004117.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., 2002: Changes in tropical clouds and radiation. Science, 296, 2095.

  • Trenberth, K. E., and Coauthors, 2007: Observations: Surface and atmospheric climate change. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 235–336.

    • Search Google Scholar
    • Export Citation
  • Uppala, S. M., and Coauthors, 2005: The ERA-40 Re-Analysis. Quart. J. Roy. Meteor. Soc., 131, 29613012.

  • Waliser, D. E., , and M. Moncrieff, 2008: The Year of Tropical Convection (YOTC) Science Plan: A joint WCRP–WWRP/THORPEX international initiative. WMO Tech. Doc. 1452, WCRP 130, WWRP/THORPEX 9, Geneva, Switzerland, 34 pp.

    • Search Google Scholar
    • Export Citation
  • Williams, M., , and R. A. Houze, 1987: Satellite-observed characteristics of winter monsoon cloud clusters. Mon. Wea. Rev., 115, 505519.

    • Search Google Scholar
    • Export Citation
  • Zipser, E. J., , D. J. Cecil, , C. T. Liu, , S. W. Nesbitt, , and D. P. Yorty, 2006: Where are the most intense thunderstorms on earth? Bull. Amer. Meteor. Soc., 87, 10571071.

    • Search Google Scholar
    • Export Citation
Save