• Asnani, G. C., 2005: Tropical Meteorology. 3 Vols. Indian Institute of Tropical Meteorology, 2241 pp.

  • Atlas, R., and Coauthors, 2005: Hurricane forecasting with the high-resolution NASA finite-volume general circulation model. Geophys. Res. Lett., 32, L03807, doi:10.1029/2004GL021513.

    • Search Google Scholar
    • Export Citation
  • Bosilovich, M. G., and Coauthors, cited 2010: NASA’s modern era retrospective-analysis for research and applications: Integrating earth observations. Earthzine. [Available online at http://www.earthzine.org/2008/09/26/nasas-modern-era-retrospective-analysis.]

    • Search Google Scholar
    • Export Citation
  • Carlson, T. N., , and Prospero J. M. , 1972: The large-scale movement of Saharan air outbreaks over the northern equatorial Atlantic. J. Appl. Meteor., 11, 283297.

    • Search Google Scholar
    • Export Citation
  • Carlson, T. N., , and Wendling P. , 1977: Reflected radiance measured by NOAA 3 VHRR as a function of optical depth for Saharan dust. J. Appl. Meteor., 16, 13681371.

    • Search Google Scholar
    • Export Citation
  • Carlson, T. N., , and Benjamin S. G. , 1980: Radiative heating rates for Saharan dust. J. Atmos. Sci., 37, 193213.

  • Colarco, P., , da Silva A. , , Chin M. , , and Diehl T. , 2010: Online simulations of global aerosol distributions in the NASA GEOS-4 model and comparison to satellite and ground-based aerosol optical depth. J. Geophys. Res., 115, D14207, doi:10.1029/2009JD012820.

    • Search Google Scholar
    • Export Citation
  • Dunion, J., , and Velden C. S. , 2004: The impact of the Saharan air layer on Atlantic tropical cyclone activity. Bull. Amer. Meteor. Soc., 85, 353365.

    • Search Google Scholar
    • Export Citation
  • Hsieh, J.-S., , and Cook K. , 2005: Generation of African easterly wave disturbances: Relationship to the African easterly jet. Mon. Wea. Rev., 133, 13111327.

    • Search Google Scholar
    • Export Citation
  • Hsu, N.-C., , Tsay S.-C. , , King M. D. , , and Herman J. R. , 2006: Deep-blue retrievals of Asian aerosol properties during ACE-Asia. IEEE Trans. Geosci. Remote Sens., 44, 31803195.

    • Search Google Scholar
    • Export Citation
  • Jeong, M. J., , Tsay S.-C. , , Ji Q. , , Hsu N. C. , , Hansell R. A. , , and Lee J. , 2008: Ground-based measurements of airborne Saharan dust in marine environment during the NAMMA field experiment. Geophys. Res. Lett., 35, L20805, doi:10.1029/2008GL035587.

    • Search Google Scholar
    • Export Citation
  • Kanamitsu, M., , Ebisuzaki W. , , Woollen J. , , Yang S.-K. , , Hnilo J. J. , , Fiorino M. , , and Potter G. L. , 2002: NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 16311643.

    • Search Google Scholar
    • Export Citation
  • Kim, K.-M., , Lau W. K. M. , , Sud Y. C. , , and Walker G. K. , 2010: Influence of aerosol-radiative forcings on the diurnal and seasonal cycles of rainfall over West Africa and eastern Atlantic Ocean using GCM simulations. Climate Dyn., 35, 115126, doi:10.1007/s00382-010-0750-1.

    • Search Google Scholar
    • Export Citation
  • Kim, M.-K., , Lau W. K. M. , , Chin M. , , Kim K.-M. , , Sud Y. C. , , and Walker G. K. , 2006: Atmospheric teleconnection over Eurasia induced by aerosol radiative forcing during boreal spring. J. Climate, 19, 47004718.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , and Kim K.-M. , 2007: How nature foiled the 2006 hurricane forecasts. Eos, Trans. Amer. Geophys. Union, 88, 105117.

  • Lau, K. M., , Kim K.-M. , , Sud Y. C. , , and Walker G. K. , 2009: A GCM study of the response of the atmospheric water cycle of West Africa and the Atlantic to Saharan dust radiative forcing. Ann. Geophys., 27, 40234037.

    • Search Google Scholar
    • Export Citation
  • Lin, S.-J., 2004: A “vertically Lagrangian” finite-volume dynamical core for global models. Mon. Wea. Rev., 132, 22932307.

  • Masaru, Y., , Mahowald N. M. , , Conley A. J. , , Collins W. D. , , Fillmore D. W. , , Zender C. S. , , and Coleman D. B. , 2007: Impact of desert dust radiative forcing on Sahel precipitation: Relative importance of dust compared to sea surface temperature variations, vegetation changes, and greenhouse gas warming. J. Climate, 20, 14451467.

    • Search Google Scholar
    • Export Citation
  • Omar, A. H., and Coauthors, 2009: The CALIPSO automated aerosol classification and lidar ratio selection algorithm. J. Atmos. Oceanic Technol., 26, 19942014.

    • Search Google Scholar
    • Export Citation
  • Onogi, K., and Coauthors, 2007: The JRA-25 Reanalysis. J. Meteor. Soc. Japan, 85, 369432.

  • Pratt, A. S., , and Evans J. L. , 2009: Potential impacts of the Saharan air layer on numerical model forecasts of North Atlantic tropical cyclogenesis. Wea. Forecasting, 24, 420435.

    • Search Google Scholar
    • Export Citation
  • Reale, O., , and Lau W. K. , 2010: Reply. J. Atmos. Sci., 67, 24112415.

  • Reale, O., , Susskind J. , , Rosenberg R. , , Brin E. , , Liu E. , , Riishojgaard L. P. , , Terry J. , , and Jusem J. C. , 2008: Improving forecast skill by assimilation of quality-controlled AIRS temperature retrievals under partially cloudy conditions. Geophys. Res. Lett., 35, L08809, doi:10.1029/2007GL033002.

    • Search Google Scholar
    • Export Citation
  • Reale, O., , Lau W. K. , , Kim K.-M. , , and Brin E. , 2009a: Atlantic tropical cyclogenetic processes during SOP-3 NAMMA in the GEOS-5 global data assimilation and forecast system. J. Atmos. Sci., 66, 35633578.

    • Search Google Scholar
    • Export Citation
  • Reale, O., , Lau W. K. , , Susskind J. , , Brin E. , , Liu E. , , Riishojgaard L. P. , , Fuentes M. , , and Rosenberg R. , 2009b: AIRS impact on the analysis and forecast track of Tropical Cyclone Nargis in a global data assimilation and forecasting system. Geophys. Res. Lett., 36, L06812, doi:10.1029/2008GL037122.

    • Search Google Scholar
    • Export Citation
  • Remer, L. A., and Coauthors, 2008: Global aerosol climatology from the MODIS satellite sensors. J. Geophys. Res., 113, D14S07, doi:10.1029/2007JD009661.

    • Search Google Scholar
    • Export Citation
  • Rienecker, M. M., and Coauthors, 2008: The GEOS-5 Data Assimilation System: Documentation versions 5.0.1, 5.1.0 and 5.20. NASA Tech. Rep. Series on Global Modeling and Data Assimilation, Vol. 27, NASA/TM-2008-104606, 118 pp. [Available online at http://gmao.gsfc.nasa.gov/pubs/tm/.]

    • Search Google Scholar
    • Export Citation
  • Sun, D. L., , Lau K. M. , , and Kafatos M. , 2008: Contrasting the 2007 and 2005 hurricane seasons: Evidence of possible impacts of Saharan dry air and dust on tropical cyclone activity in the Atlantic basin. Geophys. Res. Lett., 35, L15405, doi:10.1029/2008GL034529.

    • Search Google Scholar
    • Export Citation
  • Susskind, J., , Barnet C. , , Blaisdell J. , , Iredell L. , , Keita F. , , Kouvaris L. , , Molnar G. , , and Chahine M. , 2006: Accuracy of geophysical parameters derived from Atmospheric Infrared Sounder/Advanced Microwave Sounding Unit as a function of fractional cloud cover. J. Geophys. Res., 111, D09S17, doi:10.1029/2005JD006272.

    • Search Google Scholar
    • Export Citation
  • Susskind, J., , Barnet C. , , Blaisdell J. , , Iredell L. , , Keita F. , , Kouvaris L. , , Molnar G. , , and Chahine M. , 2007: Improved atmospheric soundings and error estimates from analysis of AIRS/AMSU data. Atmospheric and Environmental Remote Sensing Data Processing and Utilization III: Readiness for GEOSS, M. D. Goldberg et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 6684), doi:10.1117/12.734336.

    • Search Google Scholar
    • Export Citation
  • Thorncroft, C. D., , and Hoskins B. J. , 1994: An idealized study of African easterly waves. Part I: A linear view. Quart. J. Roy. Meteor. Soc., 120, 953982.

    • Search Google Scholar
    • Export Citation
  • Thorncroft, C. D., and Coauthors, 2003: The JET2000 Project: Aircraft observations of the African easterly jet and African easterly waves. Bull. Amer. Meteor. Soc., 84, 337351.

    • Search Google Scholar
    • Export Citation
  • Tompkins, A. M., , Cardinali C. , , Morcrette J.-J. , , and Rodwell M. , 2005: Influence of aerosol climatology on forecasts of the African easterly jet. Geophys. Res. Lett., 32, L10801, doi:10.1029/2004GL022189.

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

  • Wilcox, E. M., , Lau K. M. , , and Kim K.-M. , 2010: A northward shift of the North Atlantic Ocean intertropical convergence zone in response to summertime Saharan dust outbreaks. Geophys. Res. Lett., 37, L04804, doi:10.1029/2009GL041774.

    • Search Google Scholar
    • Export Citation
  • Wong, S., , and Dessler A. E. , 2005: Suppression of deep convection over the tropical North Atlantic by the Saharan air layer. Geophys. Res. Lett., 32, L09808, doi:10.1029/2004GL022295.

    • Search Google Scholar
    • Export Citation
  • Wong, S., , Dessler A. E. , , Mahowald N. M. , , Yang P. , , and Feng Q. , 2009: Maintenance of lower-tropospheric temperature inversion in the Saharan air layer by dust and dry anomaly. J. Climate, 22, 51495162.

    • Search Google Scholar
    • Export Citation
  • Wu, M.-L., , Reale O. , , Schubert S. , , Suarez M. J. , , Koster R. , , and Pegion P. , 2009: African easterly jet: Structure and maintenance. J. Climate, 22, 44594480.

    • Search Google Scholar
    • Export Citation
  • Wu, W.-S., , Purser R. J. , , and Parrish D. F. , 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 29052916.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Anomaly correlation plots for 500-hPa height, averaged from 90°S to 90°N: black line, interactive aerosol; red line, no aerosol. Virtually no global impact can be detected from the insertion of interactive aerosol treatment into the model.

  • View in gallery

    (top) The 36-h forecast (initialized at 0000 UTC 24 Aug 2006) of extinction optical thickness computed from the GEOS-5 at 550 nm and (bottom) observed MODIS optical depth at 1200 UTC 25 Aug 2006.

  • View in gallery

    As in Fig. 2, but for next day.

  • View in gallery

    Vertical meridional cross section of temperature (°C) at 10°W in the GEOS-5 NOA25 simulation at 1200 UTC 25 Aug 2006 (36-h forecast, initialized at 0000 UTC 24 Aug) (solid lines with whole numbers); IAA25 − NOA25 difference (color shading); and corresponding dust mixing ratio (10−6 kg kg−1) (solid lines with decimal numbers).

  • View in gallery

    As in Fig. 4, but for the 36-h forecast initialized at 0000 UTC 25 Aug and verified at 1200 UTC 26 Aug 2006.

  • View in gallery

    Vertical profiles of (left) temperature (°C) and (right) zonal wind u (m s−1) at Cape Verde (14.92°N, 23.49°W) in the GEOS-5 NOA25 (green) and CLA50 (blue) simulations at 1200 UTC 29 Aug 2006 (108-h forecast, initialized at 0000 UTC 25 Aug), verifying analyses (black, open circles), compared with observed soundings (purple, open circles) at Cape Verde from the SOP-3 campaign. (left) The orange and blue lines represent the IAA25 − NOA25 and CLA50 − NOA50 differences, respectively (values are provided along the lower axis); (right) the orange and blue lines represent the actual IAA25 and CLA50 u components of the wind, respectively.

  • View in gallery

    Vertical meridional cross sections of (left) 5-day-averaged zonal wind (m s−1) and (right) temperature (°C) at 10°W in the GEOS-5 NOA25 simulation initialized at 0000 UTC 25 Aug 2006. The easterly u max (m s−1) is shown at the bottom of the left panel. In both panels, solid lines with whole numbers represent NOA25 and the IAA25 − NOA25 differences are color shaded. In the right panel, the solid line with decimal numbers shows the corresponding dust mixing ratio (10−6 kg kg−1).

  • View in gallery

    As in Fig. 7, but for 10°E.

  • View in gallery

    As in Fig. 7, but for the average between 10°W and 10°E.

  • View in gallery

    (top) The 84-h forecast (initialized at 0000 UTC 10 Sep 2006) of the extinction optical thickness computed from the GEOS-5 at 550 nm, and (bottom) the observed MODIS optical depth at 1200 UTC 13 Sep 2006.

  • View in gallery

    (top) The 108-h forecast (initialized at 0000 UTC 10 Sep 2006) of the extinction optical thickness computed from the GEOS-5 at 550 nm, and (bottom) the observed MODIS optical depth at 1200 UTC 14 Sep 2006.

  • View in gallery

    As in Fig. 6, but for 1200 UTC 14 Sep 2006 (84-h forecast, initialized at 0000 UTC 11 Sep 2006).

  • View in gallery

    Vertical profiles of zonal wind forecast error variance (m2 s−2), validated against the Cape Verde (14.92°N, 23.49°W) soundings. The lines represent the GEOS-5 CLA50 (blue), IAA50 (red), and IAA25 (orange) for the (left) 84- and (right) 108-h forecast error profiles, for 7 and 9 simulations, respectively.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 42 42 8
PDF Downloads 34 34 7

Impact of Interactive Aerosol on the African Easterly Jet in the NASA GEOS-5 Global Forecasting System

View More View Less
  • 1 Laboratory for Atmospheres, NASA Goddard Space Flight Center, Greenbelt, Maryland
  • 2 Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, Maryland
© Get Permissions
Full access

Abstract

The real-time treatment of interactive, realistically varying aerosols in a global operational forecasting system, as opposed to prescribed (fixed or climatologically varying) aerosols, is a very difficult challenge that has only recently begun to be addressed. Experiment results from a recent version of the NASA’s Goddard Earth Observing System (GEOS-5) forecasting system, inclusive of interactive-aerosol direct effects, are presented in this work. Five sets of 30 five-day forecasts are initialized from a high quality set of analyses previously produced and documented, to cover the period from 15 August to 16 September 2006, which corresponds to the NASA African Monsoon Multidisciplinary Analysis (NAMMA) observing campaign. Four forecast sets are at two different horizontal resolutions, with and without interactive-aerosol treatment. A fifth forecast set is performed with climatologically varying aerosols. The net impact of the interactive aerosol, associated with a strong Saharan dust outbreak, is a temperature increase at the dust level, and a decrease in the near-surface levels, in agreement with previous observational and modeling studies. Moreover, forecasts in which interactive aerosols are included depict an African easterly jet (AEJ) at slightly higher elevation, and slightly displaced northward, with respect to the forecasts in which aerosols are not included. The shift in the AEJ position goes in the direction of the observations and agrees with previous results.

Additional affiliation: Universities Space Research Association, Columbia, Maryland.

Corresponding author address: Oreste Reale, Laboratory for Atmospheres, NASA Goddard Space Flight Center, Greenbelt, MD 20771. E-mail: oreste.reale-1@nasa.gov

Abstract

The real-time treatment of interactive, realistically varying aerosols in a global operational forecasting system, as opposed to prescribed (fixed or climatologically varying) aerosols, is a very difficult challenge that has only recently begun to be addressed. Experiment results from a recent version of the NASA’s Goddard Earth Observing System (GEOS-5) forecasting system, inclusive of interactive-aerosol direct effects, are presented in this work. Five sets of 30 five-day forecasts are initialized from a high quality set of analyses previously produced and documented, to cover the period from 15 August to 16 September 2006, which corresponds to the NASA African Monsoon Multidisciplinary Analysis (NAMMA) observing campaign. Four forecast sets are at two different horizontal resolutions, with and without interactive-aerosol treatment. A fifth forecast set is performed with climatologically varying aerosols. The net impact of the interactive aerosol, associated with a strong Saharan dust outbreak, is a temperature increase at the dust level, and a decrease in the near-surface levels, in agreement with previous observational and modeling studies. Moreover, forecasts in which interactive aerosols are included depict an African easterly jet (AEJ) at slightly higher elevation, and slightly displaced northward, with respect to the forecasts in which aerosols are not included. The shift in the AEJ position goes in the direction of the observations and agrees with previous results.

Additional affiliation: Universities Space Research Association, Columbia, Maryland.

Corresponding author address: Oreste Reale, Laboratory for Atmospheres, NASA Goddard Space Flight Center, Greenbelt, MD 20771. E-mail: oreste.reale-1@nasa.gov

1. Introduction

The role of the Saharan air layer (SAL) has been suggested to be relevant to weather forecasting over the tropical Atlantic since at least the early 1970s (Carlson and Prospero 1972). In addition to the thermal effects due to its intrinsically high heat content, the role of dust in terms of direct radiative effects has been intensely investigated for several decades (e.g., Carlson and Benjamin 1980). Surface- and space-based measurements of atmospheric optical thickness (AOT) go back to the late 1970s (e.g., Carlson and Wendling 1977) but the possibilities of merging modern-era satellites with data such as those from the Moderate Resolution Imaging Spectroradiometer (MODIS) or Cloud-Aerosol lidar and Infrared Pathfinder Satellite Observation (CALIPSO) give us a much more accurate understanding of the aerosols’ optical properties (e.g., Remer et al. 2008; Omar et al. 2009).

Among the various effects of dust over the tropical atmosphere, Dunion and Velden (2004) suggested a role of dust unfavorable to tropical cyclogenesis. One point of their argument, among other issues, is the increased static stability induced by midtropospheric warming and surface cooling. Wong and Dessler (2005) investigate the thermodynamic structure and the dust content of the SAL, showing that the warmer and drier air is located below 700 hPa, with the largest temperature anomalies around 850 hPa. This structure raises the lifting condensation level and level of free convection, increasing the energetic barrier to convection and thus leading to reduced occurrences of deep convection. Consistent findings are also obtained by Lau and Kim (2007) and by Sun et al. (2008) on a seasonal time scale. Wong et al. (2009), in a comprehensive study that makes use of both MODIS data and modeling results, confirm that positive temperature anomalies are associated with the SAL around and below 600 hPa, with a cooling effect below 925 hPa. Reale et al. (2009a, hereafter RA09a) and Reale and Lau (2010), relying upon a global data assimilation and modeling effort and the production of high quality analyses, also suggest that high dust content is associated with a thermal dipole: relatively warm at 600–700 hPa and cooler at about 900 hPa or below. RA09a’s findings are supported by the fact that the finite-volume dynamics of version 5 of the National Aeronautics and Space Administration’s (NASA) Goddard Earth Observing System (GEOS-5) is particularly suitable to maintaining fine thermal features avoiding unrealistic dispersions.

Recent climate sensitivity experiments using general circulation models have shown that the radiative effects of Saharan dust tend to draw the Atlantic ITCZ northward toward the southern edge of the dust layer, and induce large-scale north–south- and east–west-oriented divergent circulations that affect the African easterly waves and rainfall variability of the West African monsoon and climate of the Atlantic (Lau et al. 2009; Kim et al. 2010). Wilcox et al. (2010) have shown from observations that the temperature structure associated with Saharan air outbreaks is consistent with previous work, and that one of the effects of dust outbreaks is a northward shift of the intertropical convergence zone (ITCZ). Also within the operational forecasting modeling framework, evidence has been provided by Tompkins et al. (2005) indicating that an improved representation of aerosols (i.e., seasonally varying instead of fixed) leads to a more correct representation of the African easterly jet (namely, a northward and upward shift, in agreement with observations).

In this article, a further step is made in this direction: climatologically varying aerosols are replaced with interactive aerosols that are fully consistent with the meteorology at every time. This is one of the first successful attempts to insert the radiative effects of realistically varying aerosols into a global operational framework, designed to produce real-time weather forecasts. Section 2 describes the model used and the aerosol treatment, and section 3 describes the experiments performed. Section 4 presents the results, in terms of aerosol impacts on forecasting, first verifying the global skill, and then focusing on two major dust outbreaks observed over the western subtropical Africa and the northern tropical Atlantic during the third special observing phase (SOP-3) of the NASA African Monsoon Multidisciplinary Analysis (NAMMA). The impacts of interactive-aerosol treatments in these two cases are found to be very relevant on a regional scale and represent further improvement with respect to the climatologically varying aerosols. Section 5 aims at a dynamical understanding of the results, and section 6 states the conclusions of this work. It will be shown in this article that the inclusion of the radiative effects of interactive aerosols (and particularly Saharan dust) within a global modeling framework substantially improves the representation of the African easterly jet and may therefore impact the forecast skill of tropical cyclogenesis in the eastern Atlantic. The results are consistent with those of Lau et al. (2009), Kim et al. (2010), and Wilcox et al. (2010), and confirm that the direction started by the pioneering work by Tompkins et al. (2005) needs to be further pursued.

2. The model

The global data assimilation and forecasting system used is the NASA GEOS-5, documented in Rienecker et al. (2008). The GEOS-5 combines the gridpoint statistical interpolation (GSI) analysis algorithm, developed by the National Centers for Environmental Predictions (NCEP) (e.g., Wu et al. 2002) and modified by the NASA Global Modeling Assimilation Office (GMAO), with the NASA atmospheric global forecast model. The forecast model shares the same dynamical core (Lin 2004) with the so-called finite-volume general circulation model (fvGCM), also known as the GEOS-4, which has performed very well in studies focused on tropical cyclones (e.g., Atlas et al. 2005), but contains a different set of physical parameterizations, partly developed by the NASA GMAO. The GEOS-5 version used in this study is very similar to the one that has been used to produce the Modern Era Retrospective-analysis for Research and Applications (MERRA), documented by, among others, Bosilovich et al. (2010).

However, the GEOS-5 used for this work differs primarily from the MERRA version because of the aerosol interactive component, which is a new feature currently implemented also in the most recent operational version of the GEOS-5. The aerosol component used here also represents a further step with respect to the previous transport model described in Colarco et al. (2010), which merged the Goddard Chemistry, Aerosol, Radiation and Transport Model (GOCART) with the GEOS-4. The transport model described by Colarco et al. (2010) allows the treatment of dust, sea salt, and carbonaceous and sulfate aerosols, producing realistic aerosol distributions (validated against ground- and satellite-based measurements), which are consistent with the dynamics and the meteorological fields produced by the GEOS-4. In the version used in this study, aerosol transport processes (advection, diffusion, convection) are provided by the host GEOS-5 model, with the aerosol-specific processes (emission, deposition, simplified sulfate chemistry) as in Colarco et al. (2010). The direct aerosol radiative effect is included in the GEOS-5 while indirect effects on cloud–precipitation processes are not included at the present time.

3. The experiments

Five sets of 5-day forecast experiments (of 30 forecasts each, for a total of 120 five-day forecasts) at two horizontal global resolutions of ½° or ¼° (more precisely, 0.5° × 0.67° and 0.25° × 0.33°, both with 72 vertical levels) are performed. For all sets of forecasts, each individual 5-day forecast is initialized at 0000 UTC. The first forecast of the series is initialized at 0000 UTC 15 August 2006, the last at 0000 UTC 14 September 2006. All forecasts are initialized from the same set of analyses used in RA09a. These were produced by assimilating all conventional and satellite observations used operationally at that time, but with a much denser level of coverage from the Atmospheric Infrared Sounder (AIRS), obtained by ingesting quality controlled cloudy retrievals instead of clear-sky radiances, following the methodology discussed in Susskind et al. (2006, 2007). As discussed in Reale et al. (2008, 2009b) and in RA09a, AIRS-derived information in partly cloudy regions substantially improves the analyses, particularly in the tropics.

All of the experiments described in this article are therefore initialized with the same analyses but differ among themselves in their resolutions of the forecasts, and in their inclusion or exclusion of interactive aerosols. The fifth additional experiment includes climatologically prescribed aerosols, with a previously produced climatological aerosol distribution employed as in Colarco et al. (2010). The experiments are named NOAxx (no aerosol), IAAxx (interactive aerosol), and CLAxx (climatological aerosol) with the suffix xx indicating the 0.50° or 0.25° forecasts, respectively. For the climatological aerosol case, only the experiment CLA50 is performed. The period covered by the experiments includes, among others, two strong dust outbreaks associated with SAL that affected western Africa and the eastern tropical Atlantic: one occurred on about 24–28 August 2006 and the other on 10–14 September. These are the two SAL outbreaks that are the focus of the impact study discussed in this article.

One important aspect of the experiments’ formulation is that sea surface temperatures (SSTs) are prescribed and are the same (Reynolds optimum interpolation, O/I v.2) in all NOAxx and IAAxx experiments. The implication is that dust-induced near-surface temperature changes over ocean may be underestimated in these experiments, since the ocean does not have the possibility to adjust.

4. Results

a. Global impact

The role of interactive aerosols in general circulation models has been shown to be very important in a number of studies (e.g., Kim et al. 2006; Masaru et al. 2007; Kim et al. 2010). Changes in the thermal structure of the atmosphere due to the proper treatment of the aerosol radiative direct effect can lead to major changes in the circulation and may positively impact our ability to model a variety of phenomena ranging from subseasonal variability to climate change.

However, no operational center has yet implemented interactive aerosols in a global weather forecasting system. Being a state-of-the-art model improvement, it is important that the benefits of such major change are carefully considered. The 500-hPa geopotential anomaly correlation is a standard metric for assessing the impacts of an observational innovation in a data assimilation system or a change in a model. While making sure that modifications to the system do not cause a deterioration of the forecast skill according to this metric is crucial, it is also important not to discard such modifications just because they do not bring major improvements in the anomaly correlation score.

This appears specifically to be the case for the implementation of interactive aerosols in the operational global weather forecasting system GEOS-5. The 500-hPa anomaly correlations for all four sets of 30 five-day forecasts are computed against operational NCEP analyses: the impact, according to this metric, is virtually zero. In other words, the insertion of interactive aerosols does not affect the globally averaged skill of the system. In Fig. 1 the anomaly correlations of NOA50 and IAA50 are compared; the 0.25° forecasts provide identical results (not shown). The lack of global impact of the interactive aerosol is understandable because the production of dust is confined to few areas of the world, and does not occur at all times even in the dust-producing regions. At the same time, it is also important to notice that the forecast skill does not degrade as a consequence of the insertion of aerosols, which represents a major change in the model’s physics. Moreover, the lack of impact on the 500-hPa anomaly correlation computed globally does not mean that there is no impact on other quantities such as surface temperatures, wind shear, etc.

Fig. 1.
Fig. 1.

Anomaly correlation plots for 500-hPa height, averaged from 90°S to 90°N: black line, interactive aerosol; red line, no aerosol. Virtually no global impact can be detected from the insertion of interactive aerosol treatment into the model.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

In fact, it is crucial to emphasize that, for states of the atmospheres associated with specific weather systems, at times in which dust production is strong, and selected regions are analyzed, the impacts of aerosol insertion are very large and dynamically relevant. In this work we focus on the impact on the African easterly jet (AEJ), which flows, with a predominantly easterly component, at about 650 hPa and about 15°–20°N, and which is partly controlled by meridional temperature gradients over the Sahelian region. The importance of the AEJ in controlling a variety of atmospheric events spanning from weather to climate scales is well recognized (e.g., Asnani 2005). In particular, it is accepted that a mechanism of the Charney–Stern kind, which include both barotropic and baroclinic instabilities (e.g., Thorncroft and Hoskins 1994; Hsieh and Cook 2005), is one of the causes controlling the growth of disturbances extracting energy from the AEJ. These instabilities are strongly constrained by vertical and horizontal shears, and as such the position, intensity, and confinement of the AEJ in a model are crucial requirements for any meaningful skill in tropical cyclogenesis. For a modern-era comprehensive discussion of previous and contemporary works on the AEJ structure, representation, and maintenance, see, among others, Wu et al. (2009), which also emphasizes the differences in AEJ representation still existing among different reanalysis datasets.

Since the effects of aerosols in weather forecasting cannot be assessed with a global metric such as 500-hPa anomaly correlation, one must focus on specific important atmospheric features such as the AEJ. Moreover, to detect and understand the impacts of interactive aerosols, in this case Saharan dust, is not an easy or obvious task, because of (a) the intrinsically noisy nature of the radiative forcing associated with Saharan dust, (b) the overlapping with strong thermal signal due to the diurnal cycle, and (c) the inhomogeneous distribution of dust, rapidly varying in space and time.

Before designing meaningful metrics, it is therefore crucial to understand the aerosol impacts from a synoptic perspective. To this end, two prominent outbreaks observed during the SOP-3 NAMMA campaign, notable for their large scale, relatively sharp boundaries, and clear signal, are selected as a focus for this work, to show examples of a strong impact associated with dust.

b. The 25–29 August dust outbreak

The first event discussed in this article is a strong dust outbreak observed during SOP-3, which moved from Africa to the Atlantic between 25 and 28 August. RA09a and Reale and Lau (2010) suggest that the amount of dust associated with this event produced a temperature dipole (cool at about 900 hPa and warm at about 600 hPa), whose effects propagated well into the Atlantic. However, RA09a did not yet have the capability of interactive aerosols and their study is based on a comparison between the MODIS aerosol optical depths (AODs), and on the temperature structure in the GEOS-5 analyses and integrations (Reale and Lau 2010). In this work the presence of a thermal dipole possibly associated with dust is reviewed in light of this new NASA tool.

As a first step, it is important to verify that the above-referred dust outbreak is represented well in the simulations. Figure 2 shows the outbreak as represented by the 36-h forecast (the atmospheric optical depth is computed by the GEOS-5 from the aerosol total mass column, via extinction coefficient) initialized at 0000 UTC 24 August, and a composition of AOTs obtained from MODIS instruments on board both the Aqua and Terra satellites, corresponding to the verification time of 1200 UTC 25 August. Figure 2 shows that the GEOS-5 reproduces a strong dust outbreak intercepting the African coastline at about 20°–30°N and then recurving southward at about 20°W and northwestward at about 25°–35°W.

Fig. 2.
Fig. 2.

(top) The 36-h forecast (initialized at 0000 UTC 24 Aug 2006) of extinction optical thickness computed from the GEOS-5 at 550 nm and (bottom) observed MODIS optical depth at 1200 UTC 25 Aug 2006.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

Figure 3 shows the same outbreak, as captured by a 36-h forecast initialized on the following day, at 0000 UTC 25 August 2006. By comparing the observed AOD from 1200 UTC 26 August with the forecast, it can be seen that GEOS-5 again captures well the westward propagation of the dust edge, and even produces a very thin low-dust channel between the two major high dust regions centered at about 15°N, 30–40°W and 25°N, 20–30°W, respectively. On these grounds, it can be safely stated that there is a major dust outbreak in the GEOS-5, and that its timing and scale correspond well with the MODIS data.

Fig. 3.
Fig. 3.

As in Fig. 2, but for next day.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

The second step is to investigate the impact of this dust outbreak on the temperature structure, as represented by the GEOS-5. Figure 4 shows the temperature impact of aerosols at 10°W, from the same simulations in Fig. 2, and the corresponding aerosol vertical distribution. An evident strong thermal dipole, very similar to the one discussed in RA09a and Reale and Lau (2010), and previously by Dunion and Velden (2004), can be observed: a heating of about 1°–1.5°C between 850 and 500 hPa, and a cooling in the lower, near-surface, levels. This finding is also consistent with Wong et al.’s (2009) results, which explain how the dry air in the SAL can have a cooling effect below the dust layer and, to a lesser extent, in the cases with higher AOTs, also above the dust layer. It is interesting to note that some level of cooling at the top of the dust layer can also be seen in Fig. 4 at about 450–500 hPa and 22°–28°N, in agreement with Wong et al. (2009). A very similar pattern in the IAA25 minus NOA25 temperature structure across the same latitude range and at 10°W can be seen also in Fig. 5, where the corresponding aerosol impact on temperature is computed from the 36-h forecast initialized on the following day (0000 UTC 25 August) and verified at 1200 UTC 26 August. The simulations at 0.50° provide consistent results (not shown) but with diluted values and smoother gradients. As discussed in RA09a and Reale and Lau (2010), the resolution of 0.25° is substantially better than 0.50° when investigating the fine structure of the AEJ. It is particularly important to note the very good correspondence between the aerosol mixing ratio vertical distribution and the induced temperature anomaly in both Figs. 4 and 5.

Fig. 4.
Fig. 4.

Vertical meridional cross section of temperature (°C) at 10°W in the GEOS-5 NOA25 simulation at 1200 UTC 25 Aug 2006 (36-h forecast, initialized at 0000 UTC 24 Aug) (solid lines with whole numbers); IAA25 − NOA25 difference (color shading); and corresponding dust mixing ratio (10−6 kg kg−1) (solid lines with decimal numbers).

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

Fig. 5.
Fig. 5.

As in Fig. 4, but for the 36-h forecast initialized at 0000 UTC 25 Aug and verified at 1200 UTC 26 Aug 2006.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

Because the AEJ and the wind over the region are strongly controlled by lower-level temperature gradients, a change in the thermal structure must have an impact on the AEJ and more generally on the circulation. In Fig. 6 the vertical profiles extracted from a 108-h forecast in the GEOS-5 integrations (NOA25 and IAA25) are compared with the corresponding profiles of the temperature and zonal wind component, obtained from the vertical ground-based soundings from the Cape Verde Islands, at 14.9°N, 23.5°W (radiosonde data available online at http://namma.msfc.nasa.gov). The most evident impacts on temperature are a warming of up to 1°C between 875 and 750 hPa, and at about 550 hPa, and a slight cooling from the near-surface levels up to 900 hPa. The observations are colder than the NOA25 at 900 hPa, and warmer than NOA25 from 875 to about 600 hPa, indicating that the radiative effect of the aerosol in the IAA25 case forces the model in the right direction. The temperature anomaly profiles correspond well to the profiles in Wong et al. (2009) that are computed for higher AOTs (their Fig. 2). The climatological run CLA50 does not have any cooling at the 900-hPa level, but produces some relative warming between 875 and 650 hPa. The overall CLA50 minus NOA50 temperature curve is smoother and more regular than the IAA25 minus NOA25, which is as expected from a climatologically varying forcing and from the lower resolution.

Fig. 6.
Fig. 6.

Vertical profiles of (left) temperature (°C) and (right) zonal wind u (m s−1) at Cape Verde (14.92°N, 23.49°W) in the GEOS-5 NOA25 (green) and CLA50 (blue) simulations at 1200 UTC 29 Aug 2006 (108-h forecast, initialized at 0000 UTC 25 Aug), verifying analyses (black, open circles), compared with observed soundings (purple, open circles) at Cape Verde from the SOP-3 campaign. (left) The orange and blue lines represent the IAA25 − NOA25 and CLA50 − NOA50 differences, respectively (values are provided along the lower axis); (right) the orange and blue lines represent the actual IAA25 and CLA50 u components of the wind, respectively.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

The zonal wind panel shows that the verifying analysis and the observational profile match very well, confirming the overall good quality of the analyses, as previously discussed by RA09a. When it comes to the forecasts, however, the intrinsic difficulty of weather forecasting in a tropical environment becomes evident. In fact, the 108-h NOA25 forecast has a negative bias with respect to the observations, of about 4 m s−1, from the surface to 800 hPa. This is a large bias for an important feature such as the wind on the southern flank of the AEJ.

However, the quality of the forecast improves thanks to the insertion of the aerosol treatment. In fact, the aerosol impact is very clear in the lower levels, reducing the bias of about 40%. The impact is virtually zero (not shown) above 400 hPa, where dust is almost absent. It is important to stress that when there is no or minimal dust production, the impacts on temperature and zonal wind are totally negligible (not shown). This is a relevant aspect to keep in mind when designing metrics to validate the forecast impacts of aerosols in a model. In addition, one must consider that the radiative effects of dust are controlled by the diurnal cycle and the distribution of aerosols is very noisy. The wind profile forecast obtained from the run with climatologically varying aerosol is very smooth, and fits the observations better than the other runs only in the lowest levels, up to 850 hPa. However, at the jet level, the CLA50 forecast substantially underestimates the AEJ strength (of about 6 m s−1) and cannot produce the observed double maximum, which is instead captured quite well in the IAA25 forecast.

To obtain a clearer signal on the wind field and of the AEJ structure, a 5-day average across the forecast initialized on 25 August is performed at 10°W (Fig. 7). There is an evident impact in the zonal wind structure, indicating a northward and upward shift in the AEJ, consequent to the change in the temperature fields, as expected from the thermal wind relationship. A similar thermal anomaly at 10°E, and a corresponding northward shift in the AEJ, can be seen in Fig. 8. Again, a remarkable correspondence can be observed between the aerosol distribution (also averaged through time) and the induced thermal anomaly. The AEJ northward displacement is also quite evident at any longitude between 20°W and 20°E (not shown) and, to a lesser extent, also in the IAA50 experiments (not shown).

Fig. 7.
Fig. 7.

Vertical meridional cross sections of (left) 5-day-averaged zonal wind (m s−1) and (right) temperature (°C) at 10°W in the GEOS-5 NOA25 simulation initialized at 0000 UTC 25 Aug 2006. The easterly u max (m s−1) is shown at the bottom of the left panel. In both panels, solid lines with whole numbers represent NOA25 and the IAA25 − NOA25 differences are color shaded. In the right panel, the solid line with decimal numbers shows the corresponding dust mixing ratio (10−6 kg kg−1).

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

Fig. 8.
Fig. 8.

As in Fig. 7, but for 10°E.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

The northward–upward shift of the AEJ becomes even more evident if a zonal average is performed, spanning a longitude range through which the AEJ flows. We choose the longitudes between 10°W and 10°E to illustrate the point clearly; Fig. 9 represents a 5-day average across a longitude range that is affected by dust at almost all times during the outbreak. The signal in the meridional cross section is very clear: a warming of about 0.3°C north of 22°N between 850 and 600 hPa, a cooling below 850 hPa, a northward and upward shift of the AEJ, and a northward shift of the low-level westerly flow at latitudes south of 22°N. Since these features are obtained through a 5-day average across a longitude range of 20°, they represent dynamically meaningful impacts of the radiative effects of interactive Saharan dust. As seen in the previous figures, the thermal anomaly matches very well the aerosol distribution, even if the dust mixing ratios are reduced by the averaging through time and longitudes. A good level of correspondence with the results by Wong et al. (2009) can be noted.

Fig. 9.
Fig. 9.

As in Fig. 7, but for the average between 10°W and 10°E.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

As for the impacts on the AEJ consequent to the thermal anomalies, the findings in this work are consistent with those of Tompkins et al. (2005). Prior to that work, the European Centre for Medium-Range Weather Forecasts (ECMWF) operational model tended to represent the AEJ at a lower elevation and slightly more to the south than observed. The insertion of climatologically varying aerosols in the ECMWF model moved the AEJ in the right direction. This reflects the fact that dust production is not constant throughout the year, and a closer-to-reality dust distribution affects the AEJ when dust production is higher and leaves it unchanged when dust production is small or negligible. In this article, it is shown that similar changes in AEJ height and latitude as found by Tompkins et al. (2005) are being produced in the NASA global model, when the aerosol distribution is made more realistic. This work represents a further step in the direction started by Tompkins, because the improvement comes by inserting realistic aerosols that are consistent with the meteorology at each time, instead of climatologically varying aerosols. The impacts however are observed only and exclusively at those specific times at which a strong dust outbreak exists. When dust production is absent, the radiative effects of dust are not operative, and the dynamics is thus unaffected.

c. The outbreak of 10–14 September

During the SOP-3 campaign, another dust outbreak was observed, among others, toward the end of the observing period, transiting along the coast of Africa on 10 September and moving far into the Atlantic in the following days. As discussed in RA09b, this outbreak was detected clearly but was weaker than the previous. As done for the first outbreak, it is necessary to verify how the GEOS-5 reproduces it: in Fig. 10 the 84-h GEOS-5 IAA25 forecast of the optical thickness is compared with the merged AOTs from both the Aqua and Terra instruments. The structure of the dust distribution corresponds very well: in both the GEOS-5 forecast and the observations, the western edge of the dust is at about 45°–50°W, and an evident corridor of high dust values, elongated from ENE to WSW, intercepting the African coastline at about 20°N, can be also recognized in both the model and the observations. The forecast remains skillful up to 108 h. In Fig. 11, the western dust edge appears at about 55°W in the observations, and is well reproduced in the model forecast. In addition, the latitude range of the dust outbreak in its maximum width, ranging from approximately 5° to 30°N at about 40°W, is captured remarkably well in the GEOS-5 simulation. The elongated channel that connects the dust front with its source and intersects the African coast at about 20° is also reproduced properly. As in the case of the previous outbreak, the Cape Verde Islands, located at about 15°N, are on the southern edge of the high-dust corridor.

Fig. 10.
Fig. 10.

(top) The 84-h forecast (initialized at 0000 UTC 10 Sep 2006) of the extinction optical thickness computed from the GEOS-5 at 550 nm, and (bottom) the observed MODIS optical depth at 1200 UTC 13 Sep 2006.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

Fig. 11.
Fig. 11.

(top) The 108-h forecast (initialized at 0000 UTC 10 Sep 2006) of the extinction optical thickness computed from the GEOS-5 at 550 nm, and (bottom) the observed MODIS optical depth at 1200 UTC 14 Sep 2006.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

Figure 12 shows an 84-h forecast of zonal wind and temperature for the location of the Cape Verde vertical sounding, validated against the actual observations taken during SOP-3. The impact of the aerosol on the wind forecast is quite remarkable, even more so than for the case in August. The analyses match quite well the observed zonal wind profile but the NOA25 forecast has a negative bias with respect to the observations, from the surface to 550 hPa. The insertion of the aerosol reduces the bias by about 25% in the lower levels, less in the midlevels. The impact is negligible above 500 hPa (not shown) and the IAA25 and NOA25 are virtually indistinguishable due to the absence of dust in the mid- to upper troposphere. Figure 12 shows clearly that the insertion of aerosols modifies the lower-tropospheric temperature structure by warming the layer between 875 and 700 hPa by up to 0.5°C. The departure between the observations and NOA25 is mostly positive up to 600 hPa, indicating that the effects of the aerosol are in the same direction as the observations, albeit of a smaller magnitude and vertical extent. In the same plot, the results from the forecast with climatological aerosols CLA50 are also shown. The induced temperature anomaly mostly consists in an homogeneous warming of about 0.2°C from 800 to 500 hPa, and produces a wind profile that is very smooth and better than the simulation with interactive aerosols at 750, 700, and 600 hPa; however, the overall simulation with an interactive aerosol leads to a substantially better wind profile both in the lower levels and at 650 hPa, which is very close to the level in which the AEJ is maximum. The simulation with the interactive aerosol produces an improved representation of the wind structure also above 500 hPa.

Fig. 12.
Fig. 12.

As in Fig. 6, but for 1200 UTC 14 Sep 2006 (84-h forecast, initialized at 0000 UTC 11 Sep 2006).

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

The 10–14 September outbreak is long lived, and the area around Cape Verde is affected by it for several days. Additionally, in the 108-h forecast initialized at 0000 UTC 10 September and verified for the same time (1200 UTC 14 September), the NOA25 wind 108-h forecast has a negative bias with respect to observations up to 750 hPa, and a positive bias from 700 to 550 hPa (not shown). The IAA25 108-h forecast profile is closer to the observations at all levels, decreasing the NOA25 negative bias in the lower levels and the positive bias in the midlevels (not shown) and modifying the temperature structure in the right direction (not shown).

Overall, Figs. 6 and 12 indicate that the insertion of interactive aerosols in the GEOS-5 system produces a remarkable improvement in the wind forecast for a tropical environment.

In Fig. 13, the error variances of the 84- and 108-h forecasts for the CLA50, IAA50, and IAA25 simulations with respect to the observed wind at the Cape Verde observing location are presented. The purpose of this plot is to assess the improvement (if any) resulting from the interactive aerosol treatment at two different resolutions, with respect to a radiative forcing that is derived from a climatological aerosol distribution. To assess this improvement in a statistically rigorous way is not simple because the interactive-aerosol routine is not invoked when there is no aerosol. Therefore, only cases in which some aerosol production affects the station area, or occurs to the north and east of the station, should be used. It should be noted that aerosol production may not specifically occur at the validation point, and no other sites are available at this time. In addition, there are missing data from the 1200 UTC soundings at Cape Verde, which limit the number of days in which the runs can be validated (nine simulations are chosen in our case). So, the comparison is indicative and cannot be taken as being conclusive evidence, also considering the limited sample size. However, Fig. 13 shows that the IAA50 84-h forecast outperforms the CLA50 at all levels except at 600 and 650 hPa. The higher-resolution IAA25 is the best of the three at 600 hPa, and outperforms the CLA50 from 700 to 450 hPa. However, the IAA25 has the largest errors in the lower levels.

Fig. 13.
Fig. 13.

Vertical profiles of zonal wind forecast error variance (m2 s−2), validated against the Cape Verde (14.92°N, 23.49°W) soundings. The lines represent the GEOS-5 CLA50 (blue), IAA50 (red), and IAA25 (orange) for the (left) 84- and (right) 108-h forecast error profiles, for 7 and 9 simulations, respectively.

Citation: Weather and Forecasting 26, 4; 10.1175/WAF-D-10-05025.1

In the 108-h forecast, the deficiency of the IAA25 in the lower levels becomes even more evident. However, the low-resolution forecast with interactive aerosols outperforms the simulation with climatological aerosols at almost all levels. Since the CLA50 and IAA50 are strictly comparable (unlike CLA25 and CLA50, which differ in aerosol treatment and resolution), it is reasonable to state that, in these experiments, the use of interactive aerosols has produced better results than the use of climatological aerosols. In addition, in both forecasts, the largest errors are produced by the CLA50 simulations in the midtroposphere. The deficiency of the IAA25 in the lower levels may be explained by the fact that the absence of SST adjustments (due to prescribed SST) may penalize the higher-resolution simulation more than it does the IAA50. Also, the boundary layer parameterizations may have not been optimized in the higher-resolution simulations.

However, despite the relative high errors of the IAA25 in the lower levels, Fig. 13 shows clearly that the IAA50 errors are smaller than the CLA50, and that both IAA50 and IAA25 outperform the forecasts with climatological aerosols at the most critical levels in which the AEJ is located.

5. Discussion

The inclusion of the radiative effects of interactive aerosols in a global forecasting system is a complex model development problem, requiring a major effort in terms of human and computational resources. For this reason, it is very important to properly assess the potential benefits.

In the first part of this work, it is shown that the inclusion of the radiative effects of interactive aerosols in the NASA GEOS-5 global models brings no improvement in the global forecasting skill, if assessed through the conventional metric of global anomaly correlation for 500-hPa geopotential. However, caution should be exercised in using this metric alone to accept or dismiss modeling innovations. In fact, this article shows that even if the anomaly correlation is unaffected, the major change in model physics brought by interactive aerosols induces a substantial modification of the atmospheric thermal structure in one of the most meteorologically sensitive areas of the world, the African monsoon and the northern tropical Atlantic region. Any change in the low- to midtroposphere thermal structure over the African monsoon region affects drastically the AEJ and the low-level monsoonal flow. The importance of an improved representation of the African easterly jet cannot be overestimated. However, the scarcity of observations over the region, often stressed to support observational experiment such as JET2000 (Thorncroft et al. 2003) and NAMMA SOP-3, is still considered a major problem that affects the quality of the representation of the AEJ.

Wu et al. (2009) have emphasized how even the most advanced, state-of-the-art reanalyses, such as the 40-yr ECMWF Re-Analysis (ERA-40; Uppala et al. 2005), the NCEP–Department of Energy (NCEP–DOE) Reanalysis-2 (NCEP R2; Kanamitsu et al. 2002), and the Japanese Meteorological Agency 25-yr Reanalyses (JRA-25; Onogi et al. 2007), differ substantially in their representations of the AEJ. Interestingly, no reanalysis is produced from a system containing a realistic, interactive aerosol at this time. Moreover, the scarcity of observations from the area (which prevents us from gaining a clear depiction of the AEJ from observations only) does not allow the validation of this northward shift in a more rigorous way.

However, the experiments described in this article demonstrate that the radiative effects of interactive aerosols in the model, without degrading the global forecasting skill, alter significantly the circulation over the dust-affected region of the tropical eastern Atlantic and West Africa and provide a substantially improved vertical profile at one validation point.

The introduction of interactive aerosols, varying consistently with dynamics, is a serious challenge requiring consideration. In this work it is shown that a common metric used to assess improvement in forecasting skill, such as 500-hPa anomaly correlation, is not suitable for evaluating the impacts of the inclusion of interactive aerosols. At the same time, the results from this paper demonstrate the need to make the scientific community aware of the large improvements that arise out of aerosol implementation for specific areas and times.

Therefore, for areas and events affected by dust aerosols, it is important to design meaningful, event-focused, metrics. To properly design metrics, it is first necessary to appreciate the events from a synoptic point of view. To this end, two dust outbreaks observed during the NAMMA campaign during August 2006 are investigated with the aid of the GEOS-5 forecasting system, inclusive of the interactive-aerosol component.

It is first shown that the forecast of aerosol optical depths qualitatively agrees with the MODIS-merged AOTs from both Aqua and Terra, and then that the effects of the aerosol on the temperature structure are completely consistent with previous observational and modeling studies (e.g., Wong et al. 2009), producing a thermal dipole (warmer at the aerosol level, cooler below). In particular, it is shown that the effects of this thermal anomaly projects into the dynamics and improves the 84- and 108-h forecasts of temperature and wind profiles, validated against existing soundings. Meridional vertical sections at different longitudes, performed on 5-day averages across the forecast to dampen the noisy structure of the dust-induced signal, show that the net effect of the aerosol is a northward and upward shift of the AEJ. Similarly, a zonally averaged vertical section, computed along the longitudes most affected by the dust outbreak (spanning 20° in longitude), also confirms that the net effect of the aerosol simulation is a northward (and, to a lesser extent, upward) shift of the AEJ, in agreement with other modeling studies. In addition, Fig. 9 suggests a northward displacement of the westerly low-level monsoonal flow, in agreement with Wilcox et al. (2010) and related modeling results (e.g., Lau et al. 2009).

In summary, we can state the following outcome. The inclusion of radiative effects of realistically varying aerosols in the atmospheric GEOS-5 model (with prescribed observed sea surface temperatures) produces the following results:

  • no loss of global forecasting skill;
  • negligible impacts in areas not affected by dust;
  • a reasonable simulation of two strong dust outbreaks over western Africa and the northern tropical Atlantic observed during SOP-3, validated against MODIS data;
  • warming at the dust levels (between 800 and 500 hPa) in correspondence to the observed dust outbreaks;
  • no change or slight cooling in the lower, near-surface, levels;
  • a northward and upward shift of the AEJ when dust is present; a northward shift of the low-level monsoonal flow;
  • improved wind forecast up to 108 h, validated against observed vertical soundings at Cape Verde, which is on the southern edge of the dust-affected areas during both outbreaks;
  • no degradation of the forecast occurs by inclusion of interactive aerosols in the model; and
  • the low-resolution simulations in which interactive aerosols are included show consistently better results than the corresponding resolution experiments in which climatological aerosols are used.
Two caveats must be inserted for this study. The first is that the low-level cooling noted in observational studies (e.g., Dunion and Velden 2004) is substantially larger over the ocean, compared to the one reproduced in these experiments. The reason is that sea surface temperature (SST) is prescribed in both NOAxx and IAAxx experiments, because the atmospheric model is not coupled with the ocean. Therefore, the SST forces the lower levels to remain unchanged. In addition, the SST cooling consequent to high-dust production, as discussed by Lau and Kim (2007), cannot be reproduced in these experiments.

The second caveat refers to the fact that the warming of the dust-affected levels is also somehow smaller than in observational studies. This may be due to the limitations intrinsic in a gridded representation of a very finescale feature, which may need even higher resolutions to be properly represented. Finally, it is important to emphasize that the aerosols are completely modeled, from production to transport. No observed aerosol information is inserted in this version of the model. As a consequence, it is likely that dust concentrations are inferior than reality, despite the fact that the modeled dust distribution, as seen in Figs. 2, 3, 10, and 11, qualitatively matches MODIS-observed distribution. This aspect can be improved only by the next-generation effort, which will attempt to assimilate observed AOTs.

6. Concluding remarks

In this article we have shown that the use of interactive aerosols in the GEOS-5 data assimilation and forecasting system does not produce any loss of global skill but substantially changes the thermal structure over northwestern tropical Africa, during two observed dust outbreaks (one occurred around 24–29 August, and the other around 10–14 September), affecting the circulation and particularly the AEJ representation. In both cases wind and temperature anomalies induced by aerosols are consistent with the vertical aerosol distribution, and agree with previous observational and modeling studies. Temperature and wind forecasts produced during both outbreaks are validated with vertical soundings at Cape Verde, showing that the insertion of interactive aerosols improves the forecast up to day 5 for both events, at one validation point. These changes occur only when dust is present: in the cases where dust is negligible, the insertion of aerosols in the model does not degrade or affect the forecast.

The conclusion of this work is that the insertion of aerosols is definitely a worthy subject in a global forecasting model, and warns against dismissing the issue because the standard metric of the 500-hPa anomaly correlation does not show a global impact. In fact, interactive realistically varying aerosols affect prominent atmospheric features, namely the African easterly jet, the low-level monsoonal flow, and the overall thermal structure over western Africa, which are immensely important for tropical forecasting. An improved thermal representation of the area leads to improved forecasts of African easterly waves and possibly of the transformation of some of them into tropical cyclones.

The issue of the sensitivity of tropical cyclogenesis with respect to SAL is still an open one partly because of models that are unable to fully capture the detrimental effects of the SAL, according to Pratt and Evans (2009). In this sense, a more accurate thermal representation of the air mass surrounding a developing tropical low, which would be a consequence of an accurate aerosol treatment, is likely to lead to a better genesis forecast. One goal of this article is therefore to advocate for operational models dealing with Atlantic tropical weather to insert some treatment of aerosols.

In addition, it is likely that the radiative properties of dust, mixing with black carbon from biomass burning, could even increase the absorption and strengthen the effects described in this article. In particular, the choice of the single scatter albedo (SSA) is very important. In this work, the SSA for dust is 0.93 for the mean dust particle size distribution (following Colarco et al. 2010), quite close to values found by Jeong et al. (2008) over northern Africa (ranging from 0.915 to 0.959 for two events in September 2006). However, SSA is a sensitive quantity, likely to change when dust is mixed with black carbon or other mineral constituents. Over Asian deserts, for example, the range of SSA has been shown to vary from 0.87 to 0.94 at 412 nm, according to different source regions, with variations possibly linked to differing levels of iron oxide content (Hsu et al. 2006).

The limitation of this work is to rely upon modeled aerosol production and distribution. The next step, namely real-time use of satellite-derived aerosol information in global models, is a very recent development of the most advanced data assimilation systems. In this regard, the NASA GMAO is currently attempting to assimilate real-time AOTs from MODIS, which may represent the future direction for this kind of research.

Acknowledgments

The authors thank Dr. Ramesh Kakar for support through a NAMMA grant and Dr. Tsengdar Lee for allocations on NASA High-End Computing Systems. Thanks are also due to two anonymous reviewers for their constructive and insightful comments. Finally, the authors thank Mr. Ravi Govindaraju for his valuable help with the modeling experiments.

REFERENCES

  • Asnani, G. C., 2005: Tropical Meteorology. 3 Vols. Indian Institute of Tropical Meteorology, 2241 pp.

  • Atlas, R., and Coauthors, 2005: Hurricane forecasting with the high-resolution NASA finite-volume general circulation model. Geophys. Res. Lett., 32, L03807, doi:10.1029/2004GL021513.

    • Search Google Scholar
    • Export Citation
  • Bosilovich, M. G., and Coauthors, cited 2010: NASA’s modern era retrospective-analysis for research and applications: Integrating earth observations. Earthzine. [Available online at http://www.earthzine.org/2008/09/26/nasas-modern-era-retrospective-analysis.]

    • Search Google Scholar
    • Export Citation
  • Carlson, T. N., , and Prospero J. M. , 1972: The large-scale movement of Saharan air outbreaks over the northern equatorial Atlantic. J. Appl. Meteor., 11, 283297.

    • Search Google Scholar
    • Export Citation
  • Carlson, T. N., , and Wendling P. , 1977: Reflected radiance measured by NOAA 3 VHRR as a function of optical depth for Saharan dust. J. Appl. Meteor., 16, 13681371.

    • Search Google Scholar
    • Export Citation
  • Carlson, T. N., , and Benjamin S. G. , 1980: Radiative heating rates for Saharan dust. J. Atmos. Sci., 37, 193213.

  • Colarco, P., , da Silva A. , , Chin M. , , and Diehl T. , 2010: Online simulations of global aerosol distributions in the NASA GEOS-4 model and comparison to satellite and ground-based aerosol optical depth. J. Geophys. Res., 115, D14207, doi:10.1029/2009JD012820.

    • Search Google Scholar
    • Export Citation
  • Dunion, J., , and Velden C. S. , 2004: The impact of the Saharan air layer on Atlantic tropical cyclone activity. Bull. Amer. Meteor. Soc., 85, 353365.

    • Search Google Scholar
    • Export Citation
  • Hsieh, J.-S., , and Cook K. , 2005: Generation of African easterly wave disturbances: Relationship to the African easterly jet. Mon. Wea. Rev., 133, 13111327.

    • Search Google Scholar
    • Export Citation
  • Hsu, N.-C., , Tsay S.-C. , , King M. D. , , and Herman J. R. , 2006: Deep-blue retrievals of Asian aerosol properties during ACE-Asia. IEEE Trans. Geosci. Remote Sens., 44, 31803195.

    • Search Google Scholar
    • Export Citation
  • Jeong, M. J., , Tsay S.-C. , , Ji Q. , , Hsu N. C. , , Hansell R. A. , , and Lee J. , 2008: Ground-based measurements of airborne Saharan dust in marine environment during the NAMMA field experiment. Geophys. Res. Lett., 35, L20805, doi:10.1029/2008GL035587.

    • Search Google Scholar
    • Export Citation
  • Kanamitsu, M., , Ebisuzaki W. , , Woollen J. , , Yang S.-K. , , Hnilo J. J. , , Fiorino M. , , and Potter G. L. , 2002: NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 16311643.

    • Search Google Scholar
    • Export Citation
  • Kim, K.-M., , Lau W. K. M. , , Sud Y. C. , , and Walker G. K. , 2010: Influence of aerosol-radiative forcings on the diurnal and seasonal cycles of rainfall over West Africa and eastern Atlantic Ocean using GCM simulations. Climate Dyn., 35, 115126, doi:10.1007/s00382-010-0750-1.

    • Search Google Scholar
    • Export Citation
  • Kim, M.-K., , Lau W. K. M. , , Chin M. , , Kim K.-M. , , Sud Y. C. , , and Walker G. K. , 2006: Atmospheric teleconnection over Eurasia induced by aerosol radiative forcing during boreal spring. J. Climate, 19, 47004718.

    • Search Google Scholar
    • Export Citation
  • Lau, K. M., , and Kim K.-M. , 2007: How nature foiled the 2006 hurricane forecasts. Eos, Trans. Amer. Geophys. Union, 88, 105117.

  • Lau, K. M., , Kim K.-M. , , Sud Y. C. , , and Walker G. K. , 2009: A GCM study of the response of the atmospheric water cycle of West Africa and the Atlantic to Saharan dust radiative forcing. Ann. Geophys., 27, 40234037.

    • Search Google Scholar
    • Export Citation
  • Lin, S.-J., 2004: A “vertically Lagrangian” finite-volume dynamical core for global models. Mon. Wea. Rev., 132, 22932307.

  • Masaru, Y., , Mahowald N. M. , , Conley A. J. , , Collins W. D. , , Fillmore D. W. , , Zender C. S. , , and Coleman D. B. , 2007: Impact of desert dust radiative forcing on Sahel precipitation: Relative importance of dust compared to sea surface temperature variations, vegetation changes, and greenhouse gas warming. J. Climate, 20, 14451467.

    • Search Google Scholar
    • Export Citation
  • Omar, A. H., and Coauthors, 2009: The CALIPSO automated aerosol classification and lidar ratio selection algorithm. J. Atmos. Oceanic Technol., 26, 19942014.

    • Search Google Scholar
    • Export Citation
  • Onogi, K., and Coauthors, 2007: The JRA-25 Reanalysis. J. Meteor. Soc. Japan, 85, 369432.

  • Pratt, A. S., , and Evans J. L. , 2009: Potential impacts of the Saharan air layer on numerical model forecasts of North Atlantic tropical cyclogenesis. Wea. Forecasting, 24, 420435.

    • Search Google Scholar
    • Export Citation
  • Reale, O., , and Lau W. K. , 2010: Reply. J. Atmos. Sci., 67, 24112415.

  • Reale, O., , Susskind J. , , Rosenberg R. , , Brin E. , , Liu E. , , Riishojgaard L. P. , , Terry J. , , and Jusem J. C. , 2008: Improving forecast skill by assimilation of quality-controlled AIRS temperature retrievals under partially cloudy conditions. Geophys. Res. Lett., 35, L08809, doi:10.1029/2007GL033002.

    • Search Google Scholar
    • Export Citation
  • Reale, O., , Lau W. K. , , Kim K.-M. , , and Brin E. , 2009a: Atlantic tropical cyclogenetic processes during SOP-3 NAMMA in the GEOS-5 global data assimilation and forecast system. J. Atmos. Sci., 66, 35633578.

    • Search Google Scholar
    • Export Citation
  • Reale, O., , Lau W. K. , , Susskind J. , , Brin E. , , Liu E. , , Riishojgaard L. P. , , Fuentes M. , , and Rosenberg R. , 2009b: AIRS impact on the analysis and forecast track of Tropical Cyclone Nargis in a global data assimilation and forecasting system. Geophys. Res. Lett., 36, L06812, doi:10.1029/2008GL037122.

    • Search Google Scholar
    • Export Citation
  • Remer, L. A., and Coauthors, 2008: Global aerosol climatology from the MODIS satellite sensors. J. Geophys. Res., 113, D14S07, doi:10.1029/2007JD009661.

    • Search Google Scholar
    • Export Citation
  • Rienecker, M. M., and Coauthors, 2008: The GEOS-5 Data Assimilation System: Documentation versions 5.0.1, 5.1.0 and 5.20. NASA Tech. Rep. Series on Global Modeling and Data Assimilation, Vol. 27, NASA/TM-2008-104606, 118 pp. [Available online at http://gmao.gsfc.nasa.gov/pubs/tm/.]

    • Search Google Scholar
    • Export Citation
  • Sun, D. L., , Lau K. M. , , and Kafatos M. , 2008: Contrasting the 2007 and 2005 hurricane seasons: Evidence of possible impacts of Saharan dry air and dust on tropical cyclone activity in the Atlantic basin. Geophys. Res. Lett., 35, L15405, doi:10.1029/2008GL034529.

    • Search Google Scholar
    • Export Citation
  • Susskind, J., , Barnet C. , , Blaisdell J. , , Iredell L. , , Keita F. , , Kouvaris L. , , Molnar G. , , and Chahine M. , 2006: Accuracy of geophysical parameters derived from Atmospheric Infrared Sounder/Advanced Microwave Sounding Unit as a function of fractional cloud cover. J. Geophys. Res., 111, D09S17, doi:10.1029/2005JD006272.

    • Search Google Scholar
    • Export Citation
  • Susskind, J., , Barnet C. , , Blaisdell J. , , Iredell L. , , Keita F. , , Kouvaris L. , , Molnar G. , , and Chahine M. , 2007: Improved atmospheric soundings and error estimates from analysis of AIRS/AMSU data. Atmospheric and Environmental Remote Sensing Data Processing and Utilization III: Readiness for GEOSS, M. D. Goldberg et al., Eds., International Society for Optical Engineering (SPIE Proceedings, Vol. 6684), doi:10.1117/12.734336.

    • Search Google Scholar
    • Export Citation
  • Thorncroft, C. D., , and Hoskins B. J. , 1994: An idealized study of African easterly waves. Part I: A linear view. Quart. J. Roy. Meteor. Soc., 120, 953982.

    • Search Google Scholar
    • Export Citation
  • Thorncroft, C. D., and Coauthors, 2003: The JET2000 Project: Aircraft observations of the African easterly jet and African easterly waves. Bull. Amer. Meteor. Soc., 84, 337351.

    • Search Google Scholar
    • Export Citation
  • Tompkins, A. M., , Cardinali C. , , Morcrette J.-J. , , and Rodwell M. , 2005: Influence of aerosol climatology on forecasts of the African easterly jet. Geophys. Res. Lett., 32, L10801, doi:10.1029/2004GL022189.

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

  • Wilcox, E. M., , Lau K. M. , , and Kim K.-M. , 2010: A northward shift of the North Atlantic Ocean intertropical convergence zone in response to summertime Saharan dust outbreaks. Geophys. Res. Lett., 37, L04804, doi:10.1029/2009GL041774.

    • Search Google Scholar
    • Export Citation
  • Wong, S., , and Dessler A. E. , 2005: Suppression of deep convection over the tropical North Atlantic by the Saharan air layer. Geophys. Res. Lett., 32, L09808, doi:10.1029/2004GL022295.

    • Search Google Scholar
    • Export Citation
  • Wong, S., , Dessler A. E. , , Mahowald N. M. , , Yang P. , , and Feng Q. , 2009: Maintenance of lower-tropospheric temperature inversion in the Saharan air layer by dust and dry anomaly. J. Climate, 22, 51495162.

    • Search Google Scholar
    • Export Citation
  • Wu, M.-L., , Reale O. , , Schubert S. , , Suarez M. J. , , Koster R. , , and Pegion P. , 2009: African easterly jet: Structure and maintenance. J. Climate, 22, 44594480.

    • Search Google Scholar
    • Export Citation
  • Wu, W.-S., , Purser R. J. , , and Parrish D. F. , 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 29052916.

    • Search Google Scholar
    • Export Citation
Save