The 2006 NASA African Monsoon Multidisciplinary Analyses (NAMMA) field campaign was conducted to investigate the Saharan Air Layer (SAL) activity and its influence on Atlantic tropical cyclones. During the field campaign, 197 dropsonde soundings were collected from 13 research flights over the eastern Atlantic (10°N–22°N, 16°W–34°W) between 19 August and 12 September 2006 which provide a valuable data set to examine the uncertainty of the temperature and relative humidity profiles retrieved from the Atmospheric Infrared Sounder (AIRS) data in the SAL region. Comparing AIRS retrievals with the dropsonde observations indicates that AIRS profiles have a RMS difference of 1.35°C in temperature at the eight standard pressure levels and a RMS difference of 13.8% in relative humidity on seven standard pressure layers from 1000 hPa to 300 hPa. In particular, the RMS differences of temperature and relative humidity range from 0.79°C to 1.61°C and from 10.1% to 15.3%, respectively, from 1000 hPa to 300 hPa. Note that the uncertainty is estimated under sky conditions with active cloud activity because many NAMMA dropsondes were released in the cloud clusters or in the circulation of named tropical cyclones. This study suggests that in the SAL or the region with cloud activity the AIRS relative humidity/temperature has the RMS error within/approaching the design expectation for clear sky conditions.
 In general, two types of observational data are currently available for understanding of the SAL activity and its influence. One is those from a series of field projects such as the Barbados Oceanographic and Meteorological experiment (BOMEX) [Carlson and Prospero, 1972], the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE) [Carlson and Caverly, 1977], the National Aeronautics and Space Administration (NASA) Global Tropospheric Experiment/Atlantic Boundary Layer Experiment [Talbot et al., 1986], and the NASA Lidar In-space Technology Experiment (LITE) [Karyampudi et al., 1999]. In August 2006, the NASA African Monsoon Multidisciplinary Analyses (NAMMA) field campaign was conducted in a special effort to investigate the influence of the SAL on tropical cyclones over the Atlantic. These campaign data have limited temporal and spatial coverage. The other is remotely sensed profile retrievals from the Atmospheric Infrared Sounder (AIRS)/Advanced Microwave Sounding Unit (AMSU)/Humidity Sounder for Brazil (HSB) instrument suite, which were launched on the NASA Aqua satellite in 2002. By now, 7 years of AIRS data are available with more than 300,000 profiles of atmospheric temperature and humidity each day and the AIRS science team continues to improve the quality of the data [Chahine et al., 2006].
 AIRS data products have been used in various studies [Wu et al., 2006a, 2006b; Tian et al., 2006; Fu et al., 2006]. The retrieved profiles of temperature and humidity have been validated under a wide range of conditions. Comparing with aircraft observations from the Preliminary Aura Validation Experiment (PreAVE), Gettelman et al.  found that AIRS temperature is generally within ±1.5 K and AIRS water vapor is within ±25% of aircraft observations for pressures above 150 hPa. Divakarla et al.  found that AIRS temperature and water vapor are in very good agreement with operational radiosonde measurements. Using best estimates from the Department of Energy Atmospheric Radiation Measurement (ARM) program and other observations, Tobin et al.  showed that AIRS retrievals over the tropical ocean have very good accuracies for both temperature and water vapor for clear sky conditions.
 However, studies revealed that the dust aerosol in the SAL could adversely impact AIRS retrievals by introducing a systematically varying perturbation to upwelling infrared spectral radiances [Nalli and Stowe, 2002]. This issue was not addressed in the aforementioned validation studies probably because of lack of observational data in the SAL region. In order to improve the data quality and better use these valuable profiles, it is necessary to know how reliable AIRS retrievals are in the SAL region, which is the objective of this study. In this study the dropsonde observations from the NAMMA field campaign are used to validate the AIRS retrieved profiles. The data and analysis approach are described in section 2 and the results are presented in section 3. A brief summary is given in section 4.
2. Data and Methodology
 Aqua is in a Sun-synchronous polar orbit from a 705-km altitude, with an equatorial crossing of ∼1330 local time, covering the earth twice a day. The AIRS suite (AIRS, AMSU and HSB) on Aqua is a nadir scanning sounder, viewing the atmospheric infrared spectrum in 2378 channels and four visible and near-infrared channels. The three accompanied microwave instruments are AMSU A1 and A2 operating in the 23- to 90-GHz spectral range and HSB with three channels near 183 GHz and one channel at 150 GHz [Chahine et al., 2006]. (The HSB ceased functioning after 5 February 2003. The temperature and humidity profiles used in this study were retrieved without the HSB measurements.) HSB's designed goal was to provide temperature and humidity profiles with accuracies of the retrieved temperature and humidity profiles approaching those of conventional radiosondes [Fetzer et al., 2004]; that is, the retrieved temperature has an absolute accuracy of 1°C in layers 1 km thick and humidity has an absolute accuracy of 20% in layers 2 km thick in the troposphere.
 The retrieved AIRS level-2 standard products of atmospheric temperature and humidity have a nadir horizontal resolution of 45 km at 28 pressure levels from 1100 to 0.1 hPa under clear and partly cloudy conditions (for scenes with cloud fraction up to 80%). The latest AIRS products are version 5. In this version, two characteristic pressures “PBest” and “PGood” are introduced for level-2 temperature quality control. The “PBest” flag indicates that the temperature profile from the top of the atmospheric to this pressure is of best quality. The “PGood” flag indicates that the temperature profile below the level of “PBest” down to this pressure level is of good quality. “PGood” is often equal to “PBest” over ocean. Since the SAL is in the lower troposphere (850–600 hPa), the retrieved temperature is required to be of good or best quality above 700 hPa in this study; that is, “PGood” is set to be larger than 700 hPa. The retrieved water vapor products are used here when the flag “Qual_H2O” is 0 or 1, meaning that the retrieved moisture is of good or best quality above 700 hPa.
2.2. NAMMA Dropsonde
 The 2006 NAMMA field campaign was based in the Cape Verde Islands, 350 miles off the coast of Senegal in West Africa. The major research topics of this campaign is to examine the formation and evolution of tropical cyclones in the eastern and central Atlantic, and the composition and structure of the SAL and its influence on tropical cyclone development. Thirteen research flights were conducted and 197 dropsonde soundings were collected over the eastern Atlantic (10°N–22°N, 16°W–34°W) between 19 August and 12 September 2006 (Figure 1a). Figure 1b shows the AIRS 700-hPa mean temperature field averaged over the NAMMA campaign period from 15 August to 15 September 2006. In agreement with previous studies [Carlson and Prospero, 1972; Diaz et al., 1976; Karyampudi and Carlson, 1988], the SAL warm air extended westward between 10°N and 35°N. The NAMMA dropsondes were mainly released in the southern part of the mean SAL, where tropical cyclogenesis often takes place possibly because of the barotropic/baroclinic instability associated with the strong horizontal temperature gradients of the SAL [Karyampudi and Carlson, 1988].
 The dropsonde observations include dewpoint, temperature, relative humidity, wind speed and direction, and geopotential height. The quality control of these dropsonde observations is described in the document of the NAMMA dropsonde data set, which can be found in the Website at http://namma.msfc.nasa.gov/flighttracks.html. The raw data of temperature and relative humidity were first examined visually to ensure that nothing looked suspicious. The raw sounding data were then run through the Atmospheric Sounding Processing Environment software, which analyzes the data, performs smoothing, and removes suspect data points.
 As an example, Figure 2 shows one of these research flight tracks. The dropsonde release locations are plotted with close dots on the MODIS/Aqua true color image and the AIRS relative humidity field between 600 hPa and 700 hPa. The MODIS image indicates a dust storm that occurred off the coast of western Africa on 19 August 2006. Strong convective activity can be observed to the south of the SAL between 3°N and 10°N. Since most tropical cyclones emerge from such convective activities, a research flight was conducted and 15 dropsondes were released from 1441 UT to 1813 UT to collect prestorm environmental data. The aircraft first tracked southward into the convective region and then turned northward into the SAL. In Figure 2b, it can be seen that four dropsonde observations do not match AIRS data. Usually there are two reasons for these missing AIRS data: the location either with cloud fraction more than 80% or in the gaps of satellite tracks. The NAMMA field campaign used two different versions of the dropsonde. The new version has better tracking performance and reliability.
2.3. Comparing Procedure
 Although the NAMMA field campaign was not designed specifically for AIRS validation, all of the dropsonde soundings were collected within a few hours of the AIRS daytime overpasses in the eastern tropical Atlantic (Table 1). The dropsondes were launched primarily from 1200 UT to 1834 UT and the corresponding AIRS overpass time was generally from 1353 UT to 1605 UT. The maximum time difference between these AIRS overpasses and dropsonde soundings is about 4.5 h, which is comparable to the time window of ±3 h used by Divakarla et al. .
Table 1. List of Dropsonde Launch Time and Area With the Corresponding AIRS Swathes
Launch Time (UT)
AIRS Swath Time (UT)
 The NAMMA profile is reported on high-resolution pressure levels usually extending from about 300 hPa to the surface. In order to compare with AIRS profiles, the NAMMA temperature needs to be interpolated to the nine AIRS standard pressure levels: 1100, 1000, 925, 850, 700, 600, 500, 400, 300 while the NAMMA relative humidity is averaged over the standard AIRS pressure layers between 1100 and 300 hPa. Meantime, the AIRS temperature and relative humidity are collocated with the NAMMA profiles by averaging the AIRS data over a 0.5° latitude × 0.5° longitude box centered on the dropsonde locations. The drift of the dropsonde position with height is not taken into consideration in this study.
 Examples are first presented to show how large the individual differences of relative humidity and temperature between AIRS retrievals and NAMMA dropsondes can be. The dropsonde profiles of relative humidity and temperature are shown in Figures 3 and 4 at selected locations during the 19 August flight. With the very high resolution (less than 1 hPa) of these dropsondes in the vertical, the complicated structure of relative humidity is well revealed. At location 1 (Figure 2b), the dropsonde detected a minimum of 40% relative humidity around 800 hPa (Figure 3a), indicating the existence of the SAL dry air. As the aircraft tracked southward and reached locations 3 and 7 (Figure 2b), the SAL dry layer disappears and a deep layer of about 80% relative humidity was detected in the lower and middle troposphere (Figures 3b and 3c). The dry layer was detected again at location 15 (Figure 3d) when the aircraft tracked northward and reentered the SAL. Despite the relatively coarse resolution in the vertical, the AIRS retrieved relative humidity also shows a dry layer between 700 and 900 hPa (Figures 3a and 3d), indicating that the AIRS data can be used to detect the SAL. It can be seen that the AIRS profiles look like a smoothed version of the dropsonde profiles. Note that the water vapor profile is retrieved as the mean mixing ratio between two pressure levels in the standard AIRS products and it is reported on the lower-pressure level bounding the layer. In Figure 3, the AIRS relative humidity is plotted on the midpoint of the pressure layers.
 The AIRS temperature profile is reported on the pressure levels. Four NAMMA temperature profiles are selected to compare with the corresponding AIRS profiles. As shown in Figure 2, the profiles at locations 1 and 15 were collected in the SAL. As a result, we can see that a dry layer between 700 and 900 hPa (Figures 3a and 3d) is associated with a relatively stable layer near the SAL base (∼900 hPa) in Figure 4, which is a typical feature of the SAL [Dunion and Velden, 2004]. As shown in Figure 4, the AIRS temperature is in very good agreement with the NAMMA observations.
Figure 5 shows the differences of individual profiles for the 20 August and 8 September flights. The different vertical extent is due to the initial altitude of dropsonde data. The two research flights were conducted to focus on two tropical cloud clusters. One developed into tropical storm Debby (2006) and the other failed to develop into a named tropical storm. The accuracies of the AIRS temperature and relative humidity can be roughly estimated in Figure 5. For relative humidity, the difference is within ±30% for the 20 August flight while it is within ±20% for the 8 September flight (Figures 5a and 5b). The temperature difference above 850 hPa is within ±2°C, but it can be as large as −6°C in the layer below 850 hPa (Figures 5c and 5d). It seems that the convective activity associated with the developing disturbance may have affected the AIRS retrievals. Moreover the difference in the dust activity can be seen from the MODIS images and the associated aerosol optical depth (figures not included). On 20 August, a dust storm moved over the ocean, which may also have affected the AIRS retrievals [Nalli and Stowe, 2002]. On 8 September, the dust activity over the ocean is much weaker owing to the break between two easterly waves.
 The absolute mean (bias) and root mean square (RMS) differences for relative humidity and temperature are shown in Tables 2 and 3, respectively. The comparison is only possible for the seven pressure layers between 1000 hPa and 300 hPa. Table 2 suggests that the AIRS relative humidity is slightly wetter than the NAMMA soundings except in the 1000–925 hPa layer. Note that the sample size is small for this layer. The RMS difference of relative humidity ranges from 10.21% to 15.31%. The temperature statistics are computed at eight pressure levels from 1000 hPa to 300 hPa. Table 3 shows that the mean difference ranges from −0.24°C to 0.39°C at levels from 925 hPa to 400 hPa with relatively large bias at 1000 hPa and 300 hPa. The RMS difference of AIRS temperature ranges from 0.79°C to 1.51°C.
Table 2. Summary of the Difference of Relative Humidity Between NAMMA Dropsondes and AIRS Productsa
RH Bias (%)
RH RMS (%)
RH, relative humidity; RMS, root-mean-square.
Table 3. Summary of the Difference of Temperature Between NAMMA Dropsondes and AIRS Products
Temperature Bias (°C)
Temperature RMS (°C)
 A linear fit was computed for the data below 300 hPa (Figures 6 and 7) . For temperature, the AIRS retrievals are in very good agreement with the NAMMA soundings with a correlation coefficient of 0.995 and a RMS difference of 1.35°C. The RMS difference of AIRS temperature is comparable to that found by Gettelman et al. , who validated AIRS temperature and relative humidity in the upper troposphere and lower stratosphere using aircraft observations. The results indicate wide scatter for relative humidity with a small positive bias of 2.6%. The correlation coefficient between the AIRS and NAMMA relative humidity is 0.85 with a RMS difference of 13.8%. Figure 7 indicates that AIRS retrievals are to too wet as relative humidity is less than 55% while AIRS retrievals are too dry as relative humidity is larger than 55%. The RMS difference of relative humidity is smaller than that suggested by Gettelman et al.  possibly because they focused on the upper troposphere and lower stratosphere.
 The uncertainty of AIRS temperature and relative humidity profiles in the SAL region is examined using the dropsonde observations from the 2006 NAMMA campaign. The comparisons between the AIRS data and dropsonde observations are made on the eight standard pressure levels for AIRS temperature and seven standard pressure layers for AIRS relative humidity from 1000 hPa to 300 hPa. Compared with the NAMMA observations, the AIRS profiles have a RMS difference of 1.35°C in temperature and a RMS difference of 13.8% in relative temperature. In particular, the RMS differences of temperature and relative humidity range from 0.79°C to 1.61°C and from 10.1% to 15.3%, respectively, from 1000 hPa to 300 hPa. The uncertainty should be taken into account when the retrieved AIRS profiles are used in the study of the SAL activity and its influence on Atlantic tropical cyclones.
 It should be mentioned that the reported uncertainty in the retrieved AIRS profiles in the SAL is mostly associated with sky conditions with active cloud activity because the main objective of the NAMMA research flights was to investigate tropical cyclogenesis in the eastern tropical Atlantic. For this reason, many dropsondes were released in the cloud clusters that were thought to have potential to develop into tropical cyclones or in the circulation of named tropical cyclones. It is well known that cloud activity can degrade the accuracy of the AIRS retrievals [Divakarla et al., 2006; Tobin et al., 2006]. This study suggests that in the SAL or the region with cloud activity the AIRS relative humidity/temperature has the RMS error within/approaching the design expectation for clear sky conditions.
 This research was supported by the NAMMA project under NASA grant NNX07AM30G. The author gratefully acknowledges the NAMMA science team and AIRS science team for providing the valuable in situ dropsonde data and AIRS retrieved products. This study is also supported by the National Science Foundation of China (NSFC 40875038).