SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Results and Discussions
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[1] We investigate the vertical correlations between temperature variations at 925 hPa, in the atmospheric boundary layer, and temperature variations in the free troposphere and lower stratosphere in the Tropics in daily and monthly averaged satellite and radiosonde measurements and in six General Circulation Models (GCMs). The results show generally positive correlations between the boundary layer temperatures and temperatures in the rest of the troposphere, with negative correlations occurring around the tropopause and in the lower stratosphere. In typically non-convective regions, the variations at the surface show little connection to mid and upper tropospheric temperature variations. In the convective Western Pacific, the correlations are low in the mid troposphere and much larger around 200 hPa. GCMs generally capture the temperature correlations, although as a group they tend to overpredict the coupling between the boundary layer and the rest of the troposphere. The basic correlation patterns of monthly temperature are found similar to the daily results.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Results and Discussions
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[2] General Circulation Models (GCMs) play a vital role in atmospheric research as well as in making predictions of our future climate for policymakers. Much effort has therefore been expended by the scientific community in testing the fidelity of GCM simulations. Despite this effort, questions about the abilities of GCMs remain. Over the past several years, there has been an ongoing discrepancy between the measured warming trend at the surface and in the lower troposphere, and this discrepancy had been used by some critics of CO2-mitigation policies to cast doubt on the accuracy of GCM projections of future warming. Although the cause of the discrepancy has been discussed in recent publications [Mears and Wentz, 2005; Karl et al., 2006], the more general question of how tightly the surface and boundary layer are coupled to the free troposphere remains an important one. In the real atmosphere, these regions are connected by convection: evaporation near the surface followed by condensation during convective ascent transports energy and momentum from the lower to the upper troposphere. Sun and Held [1996] and Sun et al. [2001] compared modeled and observed correlations of interannual monthly-averaged variations of water vapor and temperature between the surface and troposphere and found that GCMs tend to too tightly couple the surface/boundary layer with the rest of the troposphere. Bauer et al. [2002] and Lanzante [1996] investigated this discrepancy and argued that sampling biases in the Sun and Held and Sun et al. analyses were responsible for this conclusion. Based on radiosonde observations and twelve GCM simulations, Ross et al. [2002] examined annual and seasonal correlations between temperature and corresponding humidity at different altitudes of individual stations over the tropical Pacific Ocean and North America. The GCMs showed higher correlations in high latitude, varying degrees of success in mid latitudes and generally unsuccessful at low latitudes at simulating the observed longitudinal difference in the temperature-humidity relationship. By using soundings and radar data, satellite datasets, and ECMWF, Sobel et al. [2004] analyzed the large-scale behavior of the atmosphere and found that boundary layer moist static energy is strongly correlated with lower-tropospheric mean temperature but not with upper-tropospheric mean temperature. Most recently, Santer et al. [2005] showed that there is now general agreement between interannual monthly-averaged variations of tropical temperatures at the surface and in the troposphere in GCMs and in the measurements, although the comparison between GCMs and measurements shows considerable scatter.

[3] Here, we use measurements of temperature from the Atmospheric Infrared Sounder (AIRS) and from radiosondes to examine the vertical temperature correlations in the Tropics. We will also compare these to correlation derived from GCM runs.

2. Data and Method

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Results and Discussions
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[4] We use the AIRS level-3 V4.0 daily standard physical retrieval temperature product [Olsen et al., 2005] for our analysis. This product takes individual AIRS profiles and averages them onto a 1° × 1° grid. AIRS temperature profiles have been extensively validated (see the special section “Validation of Atmospheric Infrared Sounder Observations” in Journal of Geophysical Research, 111, 2006; in particular, see Divakarla et al. [2006], Fetzer [2006], and Tobin et al. [2006]), and have an estimated accuracy of 1° K. The measurements have a vertical resolution of ∼1 km. Data from March 1, 2003 to November 30, 2005 are used for this analysis.

[5] We also analyze radiosonde data from the Integrated Global Radiosonde Archive (IGRA) [Durre et al., 2006]. We use V1.02, which are available on the mandatory pressure levels (same levels as the AIRS, except there are no radiosonde data at 600 hPa). There are 423 IGRA stations in the Tropics (20°S–20°N); more than 80% are over land. Data from the same time period as the AIRS analysis are used.

[6] We will compare correlations derived from these data sources to correlations derived from simulations of the 20th-century climate from GCM model runs performed for the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (hereafter AR4). These models represent the state-of-the-art of climate modeling and are the scientific community's best simulations of the climate. We use the last 10 years (1991–2000) of daily temperature profiles from a subset of AR4 models; the models we used are listed in Table 1. These model runs include all climate forcings in the 20th century (detailed GCMs forcings are listed in the supplemental materials of Santer et al. [2005]). The model outputs are available at http://www-pcmdi.llnl.gov/.

Table 1. Models Used in the Analyses
NumberModel NameAbbreviationReference
1GFDL-CM2.0GFDL2.0Delworth et al. [2006]
2GFDL-CM2.1GFDL2.1Delworth et al. [2006]
3GISS-AOMGISSAOMRussell et al. [1995]
4CGCM3.1(T63)CCCMAMcFarlane et al. [1992]
5ECHAM5/MPI-OMMPIJungclaus et al. [2005]
6MRI-CGCM2.3.2MRIYukimoto et al. [2001]

[7] Most of our analysis will focus on daily temperature data. For each day of each data set, we first remove the temporal mean of that particular month to get daily anomaly data. This has the effect of removing variations in the data with time scales longer than about a month. We calculate Pearson correlation coefficients between the daily temperature at 925-hPa temperature, in the boundary layer, and each level above 925 hPa. All of our correlations shown in the paper are done without any time lag, because convection is rapid and we expect the atmosphere to react rapidly to surface temperature fluctuations. The correlations for other time lags have been calculated and their magnitudes are smaller than zero-day-lag correlations (not shown).

[8] We will also analyze the correlations in monthly averaged data. We first temporally average the daily data in each month to get monthly averaged temperature; then spatially averaging the monthly temperature at each level over the entire tropical region (20°S–20°N). Next, the annual and semi-annual seasonal cycles and the mean are removed. We then calculate lag-zero Pearson correlation coefficients between the monthly temperature anomaly at 925-hPa temperature and that at each level above 925 hPa.

3. Results and Discussions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Results and Discussions
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[9] We first look at the daily tropical average (20°S–20°N) correlations. The correlations are calculated by first spatially averaging the temperature at each level over the entire tropical region (20°S–20°N), and then performing the correlation calculations on the tropical average temperature profiles. Figure 1 presents the vertical correlations from the satellite (black solid) and radiosonde (black dashed) data sets as well as from six GCM outputs (gray). The agreement between the AIRS and radiosonde data is striking, especially considering the sparse sampling of the radiosonde data set, and adds great confidence in the fidelity of the result. Both AIRS and the radiosondes show that the correlations with 925-hPa temperature decrease linearly from 1.0 at 925 hPa to about 0.3 at 600 hPa. The value is then approximately constant to about 200 hPa. Above that, in the so-called tropical tropopause layer [e.g., Sherwood and Dessler, 2000, 2001], the correlation is −0.3 to −0.5.

image

Figure 1. Correlation between the tropical averaged (20°N–20°S) daily temperature at 925 hPa and the tropical averaged daily temperature at other levels of the troposphere. AIRS data are represented by the black solid line, radiosonde by the black dashed line, and GCMs by the gray lines. 95% confidence intervals at 850 hPa, 500 hPa, and 200 hPa are plotted.

Download figure to PowerPoint

[10] As discussed above, the mechanism coupling the boundary layer to the rest of the troposphere is convection, which is relatively fast (time scale< 1 day). The short time scale is supported by the fact that the zero-day-lag correlation gives us the biggest magnitude correlations. Negative correlation (and therefore cooling) around the tropopause above convection has been discussed before [e.g., Johnson and Kriete, 1982; Sherwood et al., 2003; Kuang and Bretherton, 2004; Tian et al., 2006; Holloway and Neelin, 2006], and our analysis is consistent with these previous analyses.

[11] Our correlations are different from those found by Sun and Oort [1995], who found an almost constant vertical temperature correlation of 0.7 with surface variations over the whole troposphere. However, they were analyzing interannual variations in monthly-averaged radiosonde temperatures, while here we are looking at sub-monthly variability in daily data. We note that Bauer et al. [2002] and Lanzante [1996] have argued that the analysis scheme used by Sun and Oort to account for their data set's poor spatial coverage resulted in significant biases in their calculated correlation coefficients.

[12] Figure 1 shows that about half of the GCMs produce correlations that are similar to the AIRS and radiosonde correlations, while the other half produces correlations that are larger in the 600–200 hPa range. The width of 95% confidence intervals for all lines is similar: about 0.05 at 850 hPa, 0.16 at 500 hPa, and 0.18 at 200 hPa. The confidence intervals are calculated using Fisher's z-transform with the effective degrees of freedom equal to N(1 − r1r2)/(1 + r1r2), where N is the original sample size, and r1 and r2 are the lag-1 autocorrelations of the two time series [Bretherton et al., 1999]. The difference in the mid-troposphere between the highest-correlation models and the measurements is about 0.3, statistically significant at the 95% confidence level. Thus, these high-correlation GCMs have too vigorous connection between the boundary layer and the free troposphere, consistent with previous results [Sun and Held, 1996; Sun et al., 2001].

[13] We next compare AIRS, radiosondes, and GCMs in six individual regions. Three boxes are located in the western Pacific warm pool region and the other three are in North America, North Africa, and South Africa (see Table 2 for a description of the boxes). The regional boxes are 6° of latitude by 20° of longitude; the size of the boxes is set to be small enough to be comparable to GCM grid boxes while being big enough to include a reasonable number of sonde sites. The three boxes in the Western Pacific (NP, MS, and PH) are in regions of frequent deep convection, while convection is much less frequent in the other three regions (MA, SA, and NA). The local vertical correlations for the six regional boxes are shown in Figure 2.

image

Figure 2. Correlation between the daily temperature at 925 hPa and the daily temperature at other levels of the troposphere for six regions. AIRS data are represented by the black solid line, radiosonde by the black dashed line, and GCMs by the gray lines. 95% confidence intervals at 850 hPa, 500 hPa, and 200 hPa are plotted. The locations of the boxes and their valid IGRA stations are listed in Table 2.

Download figure to PowerPoint

Table 2. Six Boxes Selected for the Analyses
BoxesCenterValid IGRA StationsGeographic Name
MA(10E, 13N)5Mid Africa
SA(30E, 17S)10South Africa
NA(100W, 17N)4North America
NP(160E, 7N)3North Pacific
MS(105E, 3N)19Malaysia, Sumatera
PH(120E, 8N)3Philippines

[14] In the regions where convection is less frequent (MA, SA, and NA), correlation coefficients decrease from 1.0 at 925 hPa to around zero by 600 hPa, and remain near zero at higher altitudes. This is not surprising—we expect little physical connection between the boundary layer and the free troposphere in non-convective regions. The GCMs generally simulate this disconnect well, although some scatter among the models is evident. The widths of 95% confidence intervals for all lines range about 0.04 at 850 hPa, 0.16 at 500 hPa, and 0.16 at 200 hPa.

[15] In the Western Pacific regions (NP, MS, and PH), both AIRS and radiosonde data show low correlation coefficients in the mid-troposphere and a sharp maximum in the upper troposphere around 200 hPa. The radiosonde data reveal higher correlations than the AIRS data in the mid-troposphere, where the AIRS data show slightly negative correlations. The widths of 95% confidence intervals range about 0.06 at 850 hPa, 0.13 at 500 hPa, 0.14 at 200 hPa.

[16] The correlation pattern in the convective Western Pacific by observations and GCMs is consistent with relatively low detrainment rates in the mid-troposphere and high detrainment rates in the upper troposphere [e.g., Folkins and Martin, 2005] and associated cloud layers [Dessler et al., 2006]. Detrainment leads to heating either by direct turbulent mixing of high θ air into the environment or through radiative heating by detrained clouds.

[17] The GCMs produce correlations that are roughly the same shape, with a mid-troposphere minimum and maximum in the upper troposphere. However, the mid-troposphere minimum shows larger correlations than in the AIRS or radiosonde data, and the upper-troposphere maximum is broader than in the observations. In other words, GCMs tend to couple the surface to the mid and upper troposphere more strongly than the observations indicate. Considering the complexity of hydrological cycle in the real atmosphere, such differences in the convective regions are not surprising.

[18] We have also investigated correlations in the monthly averaged temperature data. Figure 3a shows the vertical correlations using 33 months of AIRS and radiosonde data. The calculated correlation coefficients are similar to those from the daily data, with correlations between 0.1 and 0.4 over most of the troposphere, and with zero (radiosonde) or negative values (AIRS) around the tropopause.

image

Figure 3. Correlation between the tropical averaged (20°N–20°S) monthly temperature at 925 hPa and the tropical averaged monthly temperature at other levels of the troposphere: (a) without filtering variations with period >2 years for ten-year data and (b) after filtering variations with period >2 years for ten-year data. AIRS data are represented by the black solid line, radiosonde by the black dashed line (both calculated using 33-month data). The ten-year radiosonde calculation is represented by the black dotted line. GCMs results are the gray lines. 95% confidence intervals at 850 hPa, 500 hPa, and 200 hPa are plotted for GCMs (thin lines) and ten-year radiosonde (thick lines).

Download figure to PowerPoint

[19] We also show on Figure 3a correlations calculated using ten years of radiosonde data (1991–2000). The ten-year radiosonde calculation shows correlation coefficients of ∼0.6–0.7 throughout the free troposphere, similar to that found by Sun and Oort [1995]. Differences between the radiosonde and model are not significant at the 95% level.

[20] Further analysis of the data sets shows that the larger correlation coefficients in the 10-year data arise because of ENSO events (there are none in the 33-month data sets). During an ENSO, the surface and atmosphere warm more synchronously than the shorter-term variations. To verify this, we plot in Figure 3b the correlations obtained after additionally filtering the data for variations with time scales >2 years. The correlations obtained from the 10-year data sets are significantly reduced, and the 10-year and 33-month radiosonde lines agree well, as we would expect. Overall, the GCMs show slightly larger correlations than the observations, although the differences are not always statistically significant.

4. Conclusions

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Results and Discussions
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[21] In this paper, we have analyzed the correlations between temperature variations at 925 hPa, in the boundary layer, with temperature at other levels of the troposphere. Our analysis uses data from the Atmospheric Infrared Sounder (AIRS) and radiosondes, and output from general circulation models (GCMs).

[22] Most of our analysis was done with the daily data, filtered to isolate short term (time scale < 30 day) variations in temperature. As expected, we see evidence for a connection between variations at 925 hPa and free troposphere and lower stratosphere in the daily AIRS and radiosonde data. Over most of the troposphere, the correlation is positive. In the lower stratosphere, however, the temperature variations are anti-correlated with the temperature at 925 hPa. It is likely that convection plays an important role in transferring temperature fluctuations from the boundary layer to the rest of the atmosphere.

[23] In regions that generally see little deep convection, variations at 925 hPa in AIRS and radiosonde are essentially disconnected from variations in the mid- and upper-troposphere. In regions where convection is frequent, such as the equatorial Western Pacific, we see a strong coupling between 925 hPa and the upper troposphere, where outflow from deep convection heats the atmosphere through cloud radiative or large-scale dynamic processes, and weaker coupling between 925 hPa and the mid troposphere.

[24] Correlations derived from daily output of GCMs show reasonable agreement with AIRS and radiosondes, although GCMs do tend to show larger correlations between the lower and the mid and upper troposphere, particularly in convective regions.

[25] We also calculated correlations using monthly averaged data. Interestingly, we found that the correlations depend to some extent on the timescale of the fluctuations being investigated. We found that GCMs did a poorer job of reproducing the observed monthly-averaged correlations than they did for the daily analysis.

[26] The stronger coupling between the boundary layer and the troposphere we found in GCMs may have impacts in simulating the long-term temperature trends in the atmosphere. Clearly, more work on this topic is required.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Results and Discussions
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information

[27] We thank Steve Sherwood, Eric Fetzer, Annmarie Eldering, Baijun Tian, Achim Stoessel, R. Saravanan, Robert Stewart, Sun Wong, Tom Delworth, and Gary Russell for their help. This work was supported by NASA EOS/IDS grant NNG04GH67G and by NASA Aqua, Terra, ACRIM data analysis grant NNG04GL64G.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Results and Discussions
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information
  • Bauer, M., A. D. Del Genio, and J. R. Lanzante (2002), Observed and simulated temperature-humidity relationships: Sensitivity to sampling and analysis, J. Clim., 15, 203215.
  • Bretherton, C. S., M. Widmann, V. Dymnikov, J. Wallace, and I. Blade (1999), The effective number of spatial degrees of freedom of a time-varying field, J. Clim., 12, 19902009.
  • Delworth, T. L., et al. (2006), GFDL's CM2 global coupled climate models—part 1: Formulation and simulation characteristics, J. Clim., 19, 643674.
  • Dessler, A. E., S. P. Palm, and J. D. Spinhirne (2006), Tropical cloud-top height distributions revealed by the Ice, Cloud, and Land Elevation Satellite (ICESat)/Geoscience Laser Altimeter System (GLAS), J. Geophys. Res., 111, D12215, doi:10.1029/2005JD006705.
  • Divakarla, M. G., C. D. Barnet, M. D. Goldberg, L. M. McMillin, E. Maddy, W. Wolf, L. Zhou, and X. Liu (2006), Validation of Atmospheric Infrared Sounder temperature and water vapor retrievals with matched radiosonde measurements and forecasts, J. Geophys. Res., 111, D09S15, doi:10.1029/2005JD006116.
  • Durre, I., R. S. Vose, and D. B. Wuertz (2006), Overview of the integrated global radiosonde archive, J. Clim., 19, 5368.
  • Fetzer, E. J. (2006), Preface to special section: Validation of Atmospheric Infrared Sounder Observations, J. Geophys. Res., 111, D09S01, doi:10.1029/2005JD007020.
  • Folkins, I., and R. V. Martin (2005), The vertical structure of tropical convection and its impact on the budgets of water vapor and ozone, J. Atmos. Sci., 62, 15201573.
  • Holloway, C. E., and J. D. Neelin (2006), The convective cold top and quasi-equilibrium, J. Atmos. Sci., in press.
  • Johnson, R. H., and D. C. Kriete (1982), Thermodynamic and circulation characteristics of winter monsoon tropical mesoscale convection, Mon. Weather Rev., 110, 18981911.
  • Jungclaus, J. H., M. Botzet, H. Haak, N. Keenlyside, J.-J. Luo, M. Latif, J. Marotzke, U. Mikolajewicz, and E. Roeckner (2005), Ocean circulation and tropical variability in the coupled model ECHAM5/MPI-OM, J. Clim., 19, 39523972.
  • Karl, T. R., S. J. Hassol, C. D. Miller, and W. L. Murray (Eds.) (2006), Temperature trends in the lower atmosphere: Steps for understanding and reconciling differences, Clim. Change Sci. Program, 164 pp., Washington, D. C.
  • Kuang, Z., and C. S. Bretherton (2004), Convective influence on the heat budget of the tropical tropopause layer: A cloud-resolving model study, J. Atmos. Sci., 61, 29192927.
  • Lanzante, J. R. (1996), Resistant, robust and non-parametric techniques for the analysis of climate data: Theory and examples, including applications to historical radiosonde station, Int. J. Climatol., 16, 11971226.
  • McFarlane, N. A., G. J. Boer, J.-P. Blanchet, and M. Lazare (1992), The Canadian Climate Centre second-generation general circulation model and its equilibrium climate, J. Clim., 5, 10131044.
  • Mears, C. A., and F. J. Wentz (2005), The effect of diurnal correction on satellite-derived lower tropospheric temperature, Science, 309, 15481551.
  • Olsen, E., et al. (2005), AIRS/AMSU/HSB version 4.0 data release user guide, 72 pp., Jet. Propul. Lab., Calif. Inst. of Technol., Pasadena, Dec.
  • Ross, R. J., W. P. Elliott, and D. J. Seidel (2002), Lower-tropospheric humidity-temperature relationships in radiosonde observations and atmospheric general circulation models, J. Hydrometeorol., 3, 2638.
  • Russell, G. L., J. R. Miller, and D. Rind (1995), A coupled atmosphere ocean model for transient climate change studies, Atmos. Ocean, 33, 683730.
  • Santer, B. D., et al. (2005), Amplification of surface temperature trends and variability in the tropical atmosphere, Science, 309, 15511556.
  • Sherwood, S. C., and A. E. Dessler (2000), On the control of stratospheric humidity, Geophys. Res. Lett., 27, 25132516.
  • Sherwood, S. C., and A. E. Dessler (2001), A model for transport across the tropical tropopause, J. Atmos. Sci., 58, 765779.
  • Sherwood, S. C., T. Horinouchi, and H. A. Zeleznik (2003), Convective impact on temperatures observed near the tropical tropopause, J. Atmos. Sci., 60, 18471856.
  • Sobel, A. H., S. E. Yuter, C. S. Bretherton, and G. N. Kiladis (2004), Large-scale meteorology and deep convection during TRMM KWAJEX, Mon. Weather Rev., 132, 422444.
  • Sun, D. Z., and I. M. Held (1996), A comparison of modeled and observed relationships between interannual variations of water vapor and temperature, J. Clim., 9, 665675.
  • Sun, D. Z., and A. H. Oort (1995), Humidity-temperature relationships in the tropical troposphere, J. Clim., 8, 19741987.
  • Sun, D. Z., C. Covey, and R. S. Lindzen (2001), Vertical correlations of water vapor in GCM, Geophys. Res. Lett., 28, 259262.
  • Tian, B., D. E. Waliser, E. J. Fetzer, B. H. Lambrigtsen, Y. L. Yung, and B. Wang (2006), Vertical moist thermodynamic structure and spatial-temporal evolution of the MJO in AIRS observations, J. Atmos. Sci., in press.
  • Tobin, D. C., H. E. Revercomb, R. O. Knuteson, B. M. Lesht, L. L. Strow, S. E. Hannon, W. F. Feltz, L. A. Moy, E. J. Fetzer, and T. S. Cress (2006), Atmospheric Radiation Measurement site atmospheric state best estimates for Atmospheric Infrared Sounder temperature and water vapor retrieval validation, J. Geophys. Res., 111, D09S14, doi:10.1029/2005JD006103.
  • Yukimoto, S., A. Noda, A. Kitoh, M. Sugi, Y. Kitamura, M. Hosaka, K. Shibata, S. Maeda, and T. Uchiyama (2001), The new Meteorological Research Institute Coupled GCM (MRI-CGCM2) — Model climate and variability, Pap. Meteorol. Geophys., 51, 4788.

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Method
  5. 3. Results and Discussions
  6. 4. Conclusions
  7. Acknowledgments
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
grl21768-sup-0001-t01.txtplain text document0KTab-delimited Table 1.
grl21768-sup-0002-t02.txtplain text document0KTab-delimited Table 2.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.