• Bellenger, H., , and J. P. Duvel, 2007: Intraseasonal convective perturbations related to the seasonal march of the Indo-Pacific monsoons. J. Climate, 20 , 28532863.

    • Search Google Scholar
    • Export Citation
  • Bernie, D. J., , S. J. Woolnough, , J. M. Slingo, , and E. Guilyardi, 2005: Modeling diurnal and intraseasonal variability of the ocean mixed layer. J. Climate, 18 , 11901202.

    • Search Google Scholar
    • Export Citation
  • Bernie, D. J., , E. Guilyardi, , G. Madec, , S. Woolnough, , and J. Slingo, 2007: Impact of resolving the diurnal cycle in an ocean–atmosphere GCM. Part 1: A diurnally forced OGCM. Climate Dyn., 29 , 575590.

    • Search Google Scholar
    • Export Citation
  • Betts, A. K., , M. Zhao, , P. A. Dirmeyer, , and A. C. M. Beljaars, 2006: Comparison of ERA-40 and NCEP/DOE near-surface data sets with other ISLSCP-II data sets. J. Geophys. Res., 111 , D22S04. doi:10.1029/2006JD007174.

    • Search Google Scholar
    • Export Citation
  • Chou, M. D., , and W. Zhao, 1997: Estimation and model validation of surface solar radiation and cloud radiative forcing using TOGA COARE measurements. J. Climate, 10 , 610620.

    • Search Google Scholar
    • Export Citation
  • Clayson, C. A., , and D. Weitlich, 2007: Variability of tropical diurnal sea surface temperature. J. Climate, 20 , 334352.

  • Cornillon, P., , and L. Stramma, 1985: The distribution of diurnal sea surface warming events in the western Sargasso Sea. J. Geophys. Res., 90 , 1181111815.

    • Search Google Scholar
    • Export Citation
  • Cronin, M. F., , and M. J. McPhaden, 1997: The upper ocean heat balance in the western equatorial Pacific warm pool during September–December 1992. J. Geophys. Res., 102 , 85338553.

    • Search Google Scholar
    • Export Citation
  • de Boyer Montégut, C., , G. Madec, , A. S. Fischer, , A. Lazar, , and D. Iudicone, 2004: Mixed layer depth over the global ocean: An examination of profile data and a profile-based climatology. J. Geophys. Res., 109 , C12003. doi:10.1029/2004JC002378.

    • Search Google Scholar
    • Export Citation
  • Donlon, C., and Coauthors, 2007: The global ocean data assimilation experiment high-resolution sea surface temperature pilot project. Bull. Amer. Meteor. Soc., 88 , 11971213.

    • Search Google Scholar
    • Export Citation
  • Duvel, J. P., , and J. Vialard, 2007: Indo-Pacific sea surface temperature perturbations associated with intraseasonal oscillations of the tropical convection. J. Climate, 20 , 30563082.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., , E. F. Bradley, , J. S. Godfrey, , G. A. Wick, , J. B. Edson, , and G. S. Young, 1996: Cool-skin and warm-layer effects on sea surface temperature. J. Geophys. Res., 101 , 12951308.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., , E. F. Bradley, , J. E. Hare, , A. A. Grachev, , and J. B. Edson, 2003: Bulk parameterization of air–sea fluxes: Updates and verification for the COARE algorithm. J. Climate., 16 , 571591.

    • Search Google Scholar
    • Export Citation
  • Feng, L., , and J. Li, 2006: A comparison of latent heat fluxes over global oceans for ERA and NCEP with GSSTF2. Geophys. Res. Lett., 33 , L03810. doi:10.1029/2005GL024677.

    • Search Google Scholar
    • Export Citation
  • Feng, M., , R. Lukas, , P. Hacker, , R. A. Weller, , and S. P. Anderson, 2000: Upper-ocean heat and salt balances in the western equatorial Pacific in response to the intraseasonal oscillation during TOGA COARE. J. Climate, 13 , 24092427.

    • Search Google Scholar
    • Export Citation
  • Flament, P., , J. Firing, , M. Sawyer, , and C. Trefois, 1994: Amplitude and horizontal structure of a large diurnal sea surface warming event during the coastal ocean dynamics experiment. J. Phys. Oceanogr., 24 , 124139.

    • Search Google Scholar
    • Export Citation
  • Flatau, M. K., , P. J. Flatau, , and D. Rudnick, 2001: The dynamics of double monsoon onsets. J. Climate, 14 , 41304146.

  • Gentemann, C. L., 2007: Diurnal warming at the ocean surface. Ph.D. thesis, Dept. of Meteorology and Physical Oceanography, University of Miami, 173 pp.

  • Gentemann, C. L., , C. J. Donlon, , A. Stuart-Menteth, , and F. J. Wentz, 2003: Diurnal signals in satellite sea surface temperature measurements. Geophys. Res. Lett., 30 , 1140. doi:10.1029/2002GL016291.

    • Search Google Scholar
    • Export Citation
  • Hansen, D. V., , and P-M. Poulain, 1996: Quality control and interpolations of WOCE–TOGA drifter data. J. Atmos. Oceanic Technol., 13 , 900909.

    • Search Google Scholar
    • Export Citation
  • Horrocks, L. A., , B. Candy, , T. J. Nightingale, , R. W. Saunders, , A. O’Caroll, , and A. R. Harris, 2003: Parameterizations of the ocean skin effect and implications for satellite-based measurement of sea-surface temperature. J. Geophys. Res., 108 , 3096. doi:10.1029/2002JC001503.

    • Search Google Scholar
    • Export Citation
  • Kawai, Y., , and H. Kawamura, 2000: Study on a platform effect in the in situ sea surface temperature observations under weak wind and clear sky conditions using numerical models. J. Atmos. Oceanic Technol., 17 , 185196.

    • Search Google Scholar
    • Export Citation
  • Kawai, Y., , and H. Kawamura, 2002: Evaluation of the diurnal warming of sea surface temperature using satellite-derived marine meteorological data. J. Oceanogr., 58 , 805814.

    • Search Google Scholar
    • Export Citation
  • Kawai, Y., , and A. Wada, 2007: Diurnal sea surface temperature variation and its impact on the atmosphere and ocean: A review. J. Oceanogr., 63 , 721744.

    • Search Google Scholar
    • Export Citation
  • Kennedy, J. J., , P. Brohan, , and S. F. B. Tett, 2007: A global climatology of the diurnal variations in sea surface temperature and implications for MSU temperature trends. Geophys. Res. Lett., 34 , L05712. doi:10.1029/2006GL028920.

    • Search Google Scholar
    • Export Citation
  • Kondo, J., , Y. Sasano, , and T. Ishii, 1979: On wind-driven current and temperature profiles with diurnal period in the oceanic planetary boundary layer. J. Phys. Oceanogr., 9 , 360372.

    • Search Google Scholar
    • Export Citation
  • Large, W. G., , J. C. McWilliams, , and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parameterization. Rev. Geophys., 32 , 363403.

    • Search Google Scholar
    • Export Citation
  • Liebmann, B., , and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77 , 12751277.

    • Search Google Scholar
    • Export Citation
  • Mellor, G. L., , and T. Yamada, 1982: Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys., 20 , 851875.

    • Search Google Scholar
    • Export Citation
  • Price, J. F., , R. A. Weller, , and R. Pinkel, 1986: Diurnal cycling: Observations and models of the upper ocean response to diurnal heating, cooling, and wind mixing. J. Geophys. Res., 91 , 84118427.

    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., , P. Brohan, , D. E. Parker, , C. K. Folland, , J. J. Kennedy, , M. Vanicek, , T. J. Ansell, , and S. F. Tett, 2006: Improved analyses of changes and uncertainties in sea surface temperature measured in situ since the mid-nineteenth century: The HadSST2 data set. J. Climate, 19 , 446469.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate, 7 , 929948.

    • Search Google Scholar
    • Export Citation
  • Saunders, P. M., 1967: The temperature at the ocean–air interface. J. Atmos. Sci., 24 , 269273.

  • Schiller, A., , and J. S. Godfrey, 2005: A diagnostic model of the diurnal cycle of sea surface temperature for use in coupled ocean–atmosphere models. J. Geophys. Res., 110 , C11014. doi:10.1029/2005JC002975.

    • Search Google Scholar
    • Export Citation
  • Scott, J. D., , and M. A. Alexander, 1999: Net shortwave fluxes over the ocean. J. Phys. Oceanogr., 29 , 31673174.

  • Shinoda, T., 2005: Impact of the diurnal cycle of solar radiation on intraseasonal SST variability in the western equatorial Pacific. J. Climate, 18 , 26282636.

    • Search Google Scholar
    • Export Citation
  • Shinoda, T., , and H. H. Hendon, 1998: Mixed layer modeling of intraseasonal variability in the tropical western Pacific and Indian Oceans. J. Climate, 11 , 26682685.

    • Search Google Scholar
    • Export Citation
  • Soloviev, A., , and R. Lukas, 1997: Observation of large diurnal warming events in the near-surface layer of the western equatorial Pacific warm pool. Deep-Sea Res. I, 44 , 10551076.

    • Search Google Scholar
    • Export Citation
  • Stramma, L., , P. Cornillon, , R. A. Weller, , J. F. Price, , and M. G. Briscoe, 1986: Large diurnal sea surface temperature variability: Satellite and in situ measurements. J. Phys. Oceanogr., 16 , 827837.

    • Search Google Scholar
    • Export Citation
  • Stuart-Menteth, A. C., , I. S. Robinson, , and P. G. Challenor, 2003: A global study of diurnal warming using satellite-derived sea surface temperature. J. Geophys. Res., 108 , 3155. doi:10.1029/2002JC001534.

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

  • Ward, B., 2006: Near-surface ocean temperature. J. Geophys. Res., 111 , C02004. doi:10.1029/2004JC002689.

  • Webster, P. J., , C. A. Clayson, , and J. A. Curry, 1996: Clouds, radiation, and the diurnal cycle of sea surface temperature in the tropical western Pacific. J. Climate, 9 , 17121730.

    • Search Google Scholar
    • Export Citation
  • Weller, R., , and S. Anderson, 1996: Surface meteorology and air–sea fluxes in the western equatorial Pacific warm pool during the TOGA coupled ocean–atmosphere response experiment. J. Climate, 9 , 19591991.

    • Search Google Scholar
    • Export Citation
  • Woolnough, S. J., , F. Vitart, , and M. A. Balmaseda, 2007: The role of the ocean in the Madden–Julian Oscillation: Implications for MJO prediction. Quart. J. Roy. Meteor. Soc., 133 , 117128.

    • Search Google Scholar
    • Export Citation
  • Zeng, X., , and A. Beljaars, 2005: A prognostic scheme of sea surface skin temperature for modeling and data assimilation. Geophys. Res. Lett., 32 , L14605. doi:10.1029/2005GL023030.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Schematic of the upper ocean temperature profile for a diurnal warm layer. The different definitions of the upper ocean temperatures used in the article are reported following Donlon et al. (2007) and Kawai and Wada (2007). The cool skin base (dashed line) corresponds to a depth of around 0.1–1 mm.

  • View in gallery

    (top) Temperature measured by the SVP buoy 34158 the Indian Ocean around 6°S, 81°E for raw measurements (thin) and the Krig time series (bold). (bottom) Results of the interpolation method following the procedure described in the appendix (dotted) and Tdepth at 25 cm obtained using F96 (bold gray) and ZB05 (black) bilinearly interpolated at the buoy’s location.

  • View in gallery

    Number of days of SVP buoy measurements for each 2.5° × 2.5° region for the dataset (1993–2002) used for the validation of the F96 and ZB05 results.

  • View in gallery

    Distribution of 3-day average SVP buoy DSA as a function of a 3-day average of (top) ZB05 and (bottom) F96 DSA, where DSAs are computed for a depth of 25 cm and interpolated at SVP buoy locations for the (left) Indian, (middle) Pacific, and (right) Atlantic Oceans. The contours represent the number of points per 0.05 K × 0.05 K intervals. The contours represent 2, 10, 20, 100, and 200 points per bin. The correlations and the equation of the corresponding best linear fits (bold solid lines) are listed for each ocean basin. The best fit is computed following the vertical offset least squares fitting method.

  • View in gallery

    (a) Daily SVP buoy DSA distribution for 1993–2002 (bold solid) and the corresponding F96 (thin solid) and ZB05 DSA (dashed) computed for a depth of 25 cm and interpolated at the buoy locations. (b) SVP DSA distributions for F96 (thin solid) and ZB05 DSA = 0 (bold dashed).

  • View in gallery

    Daily F96 DSA for T0.25 (25-cm depth) as a function of F96 DSA for Tsubskin (dots) and the corresponding second-degree polynomial fit (light solid line). The corresponding relation for ZB05 is also shown (bold solid line).

  • View in gallery

    (bottom) Average and (top) standard deviation of F96 and ZB05 DSA for Tsubskin in given intervals of daily average surface insolation and surface (10 m) wind speed. (a),(b) The variation of the DSA with the surface wind speed for given insolation intervals. (c),(d) The variation of the DSA with the insolation for given surface wind speed intervals. The results from the F96 and the ZB05 models (solid lines) are compared to the empirical relation reported in Gentemann et al. (2003) (dotted lines). The top panels represent the standard deviation of the F96 and ZB05 results for each interval. Each interval is 10 W m−2 wide for insolation and 0.2 m s−1 wide for the surface wind speed. Only bins with at least 100 points are shown.

  • View in gallery

    Monthly average DSA (colors, K) and DWL depth (5 and 7.5 m, thin solid line; 10 m, dotted line) for the period 1979–2002.

  • View in gallery

    (a) Annual mean surface flux perturbation (surface cooling, W m−2) due to the DWL and (b) the corresponding annual mean cooling of the ocean mixed layer (K yr−1).

  • View in gallery

    Distribution of daily surface flux perturbations due to the DWL for all 2.5° × 2.5° oceanic regions.

  • View in gallery

    Average cumulative distribution (decreasing size) of the DWL equivalent radius for DSAs larger than 0.68 (solid), 1.05 (stripes), and 1.42 K (white).

  • View in gallery

    Occurrence (% of day) of DWLs with an equivalent radius larger than 1000 km for a DSA > 0.68 K for December–February, March–May, June–August (JJA), and September–November from 1979 to 2002. DWLs reaching 30° of latitude are not considered.

  • View in gallery

    Distribution of DWL episode durations for all 2.5° × 2.5° oceanic regions. For each region, the duration of a DWL episode is the number of consecutive days with a DSA larger than a given threshold. As in Fig. 10, the DSA threshold are 0.68 (solid), 1.05 (stripes), and 1.42 K (white).

  • View in gallery

    Occurrence (% of day) of DWLs with durations larger than 5 days for a DSA > 0.68 K for DJF, MAM, JJA, and SON from 1979 to 2002.

  • View in gallery

    Modulation of the DSA (colors) by the intraseasonal oscillation for the (top) November–April (18 ISO events) and (bottom) May–October seasons (21 ISO events) for the period 1979–2002. The four ISO phases are detected based on the filtered OLR signal around 0°, 90°E: (a),(e) maximum OLR; (c),(g) minimum OLR; (b),(d),(f),(h) intermediate phases. The corresponding OLR anomaly is also reported. Dashed (solid) contours indicate −(+)10 W m−2.

  • View in gallery

    Fig. A1. Average diurnal evolution of the temperature anomalies for raw SVP buoy 34158 measurements (bold solid) and the results of the F96 algorithm at the same locations as the buoy (thin solid). The cosine function (dashed) used by Hansen and Poulain (1996) to build the Krig dataset is also shown.

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

An Analysis of Tropical Ocean Diurnal Warm Layers

View More View Less
  • 1 CNRS, Laboratoire de Météorologie Dynamique, ENS, Paris, France
© Get Permissions
Full access

Abstract

During periods of light surface wind, a warm stable layer forms at the ocean surface with a maximum sea surface temperature (SST) in the early afternoon. The diurnal SST amplitude (DSA) associated with these diurnal warm layers (DWLs) can reach several degrees and impact the tropical climate variability. This paper first presents an approach to building a daily time series of the DSA over the tropics between 1979 and 2002. The DSA is computed over 2.5° of latitude–longitude regions using a simple DWL model forced by hourly-interpolated surface radiative and turbulent fluxes given by the 40-yr ECMWF Re-Analysis (ERA-40). One advantage of this approach is the homogeneity of the results given by the relative homogeneity of ERA-40. The approach is validated at the global scale using empirical DWL models reported in the literature and the Surface Velocity Program (SVP) drifters of the Marine Environmental Data Service (MEDS). For the SVP dataset, a new technique is introduced to derive the diurnal variation of the temperature from raw measurements.

This DWL time series is used to analyze the potential role of DWLs in the variability of the tropical climate. The perturbation of the surface fluxes by DWLs can give a cooling of the ocean mixed layer as large as 2.5 K yr−1 in some tropical regions. On a daily basis, this flux perturbation is often above 10 W m−2 and sometimes exceeds 50 W m−2. DWLs can be organized on regions up to a few thousand kilometers and can persist for more than 5 days. It is shown that strong DWLs develop above the equatorial Indian Ocean during the suppressed phase of the intraseasonal oscillation (ISO). DWLs may trigger large-scale convective events and favor the eastward propagation of the ISO convective perturbation during boreal winter. This study also suggests that the simple approach presented here may be used as a DWL parameterization for atmospheric general circulation models.

Corresponding author address: Dr. Hugo Bellenger, LMD, ENS, 24 Rue Lhomond, 75231, Paris CEDEX 05, France. Email: belleng@lmd.ens.fr

Abstract

During periods of light surface wind, a warm stable layer forms at the ocean surface with a maximum sea surface temperature (SST) in the early afternoon. The diurnal SST amplitude (DSA) associated with these diurnal warm layers (DWLs) can reach several degrees and impact the tropical climate variability. This paper first presents an approach to building a daily time series of the DSA over the tropics between 1979 and 2002. The DSA is computed over 2.5° of latitude–longitude regions using a simple DWL model forced by hourly-interpolated surface radiative and turbulent fluxes given by the 40-yr ECMWF Re-Analysis (ERA-40). One advantage of this approach is the homogeneity of the results given by the relative homogeneity of ERA-40. The approach is validated at the global scale using empirical DWL models reported in the literature and the Surface Velocity Program (SVP) drifters of the Marine Environmental Data Service (MEDS). For the SVP dataset, a new technique is introduced to derive the diurnal variation of the temperature from raw measurements.

This DWL time series is used to analyze the potential role of DWLs in the variability of the tropical climate. The perturbation of the surface fluxes by DWLs can give a cooling of the ocean mixed layer as large as 2.5 K yr−1 in some tropical regions. On a daily basis, this flux perturbation is often above 10 W m−2 and sometimes exceeds 50 W m−2. DWLs can be organized on regions up to a few thousand kilometers and can persist for more than 5 days. It is shown that strong DWLs develop above the equatorial Indian Ocean during the suppressed phase of the intraseasonal oscillation (ISO). DWLs may trigger large-scale convective events and favor the eastward propagation of the ISO convective perturbation during boreal winter. This study also suggests that the simple approach presented here may be used as a DWL parameterization for atmospheric general circulation models.

Corresponding author address: Dr. Hugo Bellenger, LMD, ENS, 24 Rue Lhomond, 75231, Paris CEDEX 05, France. Email: belleng@lmd.ens.fr

1. Introduction

In calm conditions, the small wind stress reduces the vertical mixing near the ocean surface and the solar radiation absorbed in the first meters of the ocean produces a stable stratification. A diurnal warm layer (DWL) forms that gives a daytime augmentation of the sea surface temperature (SST) that can reach several degrees (Stramma et al. 1986; Flament et al. 1994; Weller and Anderson 1996; Soloviev and Lukas 1997; Ward 2006). During night, mixing by oceanic convection generally destroys this surface stratification, giving an SST close to the bulk mixed-layer temperature. A DWL thus increases the daily average SST. Shinoda and Hendon (1998), Shinoda (2005), and Bernie et al. (2005) showed that DWLs indeed have a substantial impact on the day-to-day evolution of the SST. A larger daily-mean SST increases the infrared and turbulent heat fluxes and also results in a cooler ocean mixed layer. This effect is small on average. For example, this additional heat transfer is estimated to be 4 W m−2 on average for 70-day measurements in the west equatorial Pacific during the Tropical Ocean–Global Atmosphere (TOGA) Coupled Ocean–Atmosphere Response Experiment (COARE) (Fairall et al. 1996, hereafter F96). This effect is also counterbalanced by the decrease of the SST due to the cool skin effect. On the first order, considering both DWL and cool skin effects in a GCM should thus modify marginally the energy balance at the surface. However, the heat output due to DWL may be stronger for particular days and can reach high values (>50 W m−2) around noon (F96; Ward 2006). On a day-to-day and a regional basis, the reinforcement of the SST contrast between regions with or without DWL should thus impact the regional contrast in the atmospheric boundary layer temperature. This may influence the atmospheric variability from diurnal to intraseasonal time scales (Webster et al. 1996). Woolnough et al. (2007) gave a very good illustration for the intraseasonal time scales by showing that Madden–Julian oscillation (MJO) predictability is improved when DWLs are taken into account in a GCM.

Previous observational studies already analyzed DWL at regional to global scale. Stuart-Menteth et al. (2003) derived the first global climatology of the diurnal SST amplitude (DSA) and its seasonal and interannual variations from 6 yr of daytime and nighttime Advanced Very High Resolution Radiometer (AVHRR) infrared measurements. They show that large regions of the ocean were affected by diurnal warming. The monthly average DSA patterns show a strong seasonal variability linked to wind variations and the solar insolation cycle. They note the significant impact of DWL on the global SST field and the necessity of taking it into account in remote sensing and in global climate modeling studies. Using satellite data (both infrared and microwave) in clear-sky conditions, Gentemann et al. (2003) derived useful empirical formulations linking the DSA to daily average surface wind and insolation. This relation will be used below to evaluate our approach. Recently, Kennedy et al. (2007) constructed a global monthly DSA climatology based on the Surface Velocity Program (SVP) drifting buoy measurements from 1990 to 2004. Although the seasonal pattern and DSA amplitudes retrieved from SVP measurements are in good agreement with the satellite approaches used previously, they also used this climatology to show that satellite drift could account for more than 10% of the decadal near surface temperature trend observed in the tropics.

Other empirical approaches were developed to evaluate the DSA from daily statistics. Webster et al. (1996) used a 1D model validated with TOGA COARE in situ data to study the sensitivity of DSA to wind speed, insolation, and precipitation and to derive a formulation of the DSA as a function of daily mean surface wind and precipitation rate and of daily solar radiation maximum. Kawai and Kawamura (2002) developed a similar empirical formulation but neglected the precipitation rate data. They showed examples of DWL organized on the large scale in the western Pacific Ocean. Clayson and Weitlich (2007) used the Webster et al. (1996) empirical formulation to derive a 5-yr daily climatology from peak shortwave solar radiation—determined from International Satellite Cloud Climatology Project (ISCCP) data—and daily averaged wind speed determined from Special Sensor Microwave Imager (SSM/I) data. They did not consider the daily average precipitation rate because it was not available everywhere. Comparison of their results with Tropical Atmosphere Ocean (TAO)–Triangle Trans-Ocean Buoy Network (TRITON) and Prediction and Research Moored Array in the Tropical Atlantic (PIRATA) buoy measurements showed a correlation of 0.74. Then, this daily climatology was used to study seasonal and interannual variability of the DSA. They showed in particular that there is a higher (lower) DSA in the equatorial eastern (western) ocean for La Niña years compared to El Niño years.

Note that DWLs can be simulated using current ocean GCM (OGCM) formulations. An OGCM with a high vertical resolution is also necessary if one wants to take into account all the perturbations and feedbacks associated with DWLs, including perturbations of the ocean dynamics. Bernie et al. (2007) obtained a global estimate of the perturbations induced by the DWLs by forcing an OGCM of 300 vertical levels (first level of 1-m depth) with a daily climatology of fluxes from ERA-40. However, such an approach is costly and thus not relevant for long climate integrations of coupled climate models. In addition, some problems (e.g., an underestimate of the DSA) still exist in these OGCM simulations (Bernie et al. 2007). A parameterization could thus be useful if the goal is simply to calculate the perturbation of surface flux generated by DWLs. The only action of this parameterization is to modify the surface fluxes computed by the atmospheric GCM (AGCM) boundary layer scheme in the presence of DWLs. For daily averages, this will modify the surface fluxes identically for the atmosphere and the ocean models. The energy budget thus remains fully balanced for coupled GCMs using such a parameterization. Note that for a parameterized DWL, the same formulation is applicable on coupled as well as forced atmospheric GCM simulations, giving thus a homogeneous approach to compare forced and coupled simulations.

The aim of the present study is to report a single approach that can be used to (i) diagnose the main characteristics of the DWL from homogeneous surface meteorological field over the whole tropical zone and (ii) compute the perturbation of the surface fluxes related to the presence of DWL in an AGCM. The approach is based on simple DWL models forced by a large-scale surface meteorological field. The approach is evaluated by constructing a long time series with two DWL models forced by hourly-interpolated surface parameters given by the 40-yr European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-40; Uppala et al. 2005). This study uses the DWL models of both F96 and of Zeng and Beljaars (2005, hereafter ZB05) to test the sensitivity of the results to the scheme considered. The DSAs are computed for each region and each day by forcing these DWL models by hourly fields. This is more precise than evaluating the DSA from empirical relation based on daily statistics and gives an approach usable in GCMs.

The DWL models and the approach are presented section 2. A long (24 yr) time series of daily values of the DSA is constructed and the result is evaluated in section 3 using global in situ buoy measurements and satellite estimates reported in previous studies. These evaluations also test in what measure these DWL models, which are designed to be forced by local in situ measurements, can also be used at the spatial resolution of large-scale models (reanalyses and GCMs). In section 4, the long DSA time series is further used (i) to derive statistics on the DWL characteristics, (ii) to analyze the potential role of the DWLs in the perturbation of the surface fluxes, and (iii) to determine the relation between the intraseasonal perturbation of the convection and the distribution of the DWL. A summary and conclusions are presented in section 5.

2. The DWL algorithm

a. DWL models

Following Donlon et al. (2007) and Kawai and Wada (2007), the upper ocean structure can be divided into three layers (Fig. 1). Starting from the surface, there is first the skin layer, with a temperature Tskin, that incorporates the cool skin effect of the ocean’s first tenths of millimeter due to heat fluxes and surface longwave emission (Saunders 1967). This cooling is about 0.3 K on the average and can reach 0.6 K. Under this skin layer is the bulk layer (down to a few meters), where measurements from ship and buoys are typically performed. The bulk temperature Tdepth is either the temperature of the warm layer, which varies with depth, or the temperature of the mixed layer. At the top of the bulk layer, the subskin temperature Tsubskin is the higher temperature reached in the warm layer. The deeper layer corresponds to the nocturnal mixed layer with no diurnal variation of its temperature Tfnd.

Different DWL models such as diffusion (Kondo et al. 1979; Mellor and Yamada 1982; Large et al. 1994) or bulk models (Price et al. 1986) exist to diagnose the DWL characteristics using a surface forcing. Based on Price et al. (1986), F96 proposed a simple DWL model used to refine bulk surface turbulent fluxes computations. ZB05 and Schiller and Godfrey (2005) also proposed simple DWL models. These DWL models give an estimate of the evolution of the SST for each time step of the simulation and may thus be used as a parameterization to diagnose DWL and alter the surface flux accordingly in an AGCM.

In the DWL algorithm of F96, a DWL appears when the solar heat absorbed in the top first meter of the ocean compensates for the heat loss through the ocean surface. Briefly, the depth of the DWL is determined by a bulk Richardson number criterion (bottom of the DWL at Ri = 0.65) and by the values of the time integrals of momentum and heat associated with surface turbulent fluxes. The temperature profile in the DWL is linear, with a slope deduced from the heat integral and the DWL depth. The F96 algorithm also enables us to evaluate the cool skin effect that modifies the surface fluxes and thus also the DWL evolution. The F96 parameterization of the cool skin effect was validated during the Nauru cruise using in situ radiometer and Tdepth measurements during night (Horrocks et al. 2003).

The F96 algorithm is based on the hypotheses that changes in temperature and current shear in the upper ocean are only due to surface fluxes and are confined to the DWL. Advection effects, which may affect the stability of the layer, are hence not taken into account. This can be justified in the tropics because horizontal SST gradients are rather weak (see, e.g., Bernie et al. 2005). The horizontal advection thus modifies the SST mostly at longer time scales (Cronin and McPhaden 1997; Feng et al. 2000). The F96 algorithm also ignores the salinity, although it may have an impact on the stratification (Webster et al. 1996). Under the DWL, current and temperature profiles are assumed to be constant. In particular, the diurnal warming of the water under the DWL by shortwave radiation is neglected. Then, when the DWL deepens during the day, its temperature evolution is not affected by solar heat trapped underneath. Because the DWL generally deepens throughout the day, this could lead to a slight underestimation of the DWL heat content and hence the DSA. Conversely, if the DWL shrinks, the time-integrated momentum and heat remain in the resulting shallower layer. These integrals are thus not computed in a conservative way in the F96 algorithm. Also, the DWL parameters and thus the time integrals of heat and momentum are reinitialized at midnight. The F96 algorithm is thus a parameterization designed to diagnose the diurnal SST evolution, but it is not a precise model of the evolution of the ocean surface layers; this is why, when used in an AGCM, it should be considered only as a parameterization used to correct the surface fluxes in presence of a DWL.

The algorithm proposed by ZB05 was developed to give a prognostic SST scheme for modeling and data assimilation. Compared to F96, the main interest of the ZB05 scheme is to compute the diurnal evolution of Tsubskin in a conservative fashion. In ZB05, the depth of the DWL is assumed to be constant and equal to 3 m. According to ZB05, changing the DWL depth by 0.5 m changes the average DSA for less than 0.15 K. We will not test such sensitivity further here, but the impact of the fixed depth and the fixed temperature profile in ZB05 will be considered in the comparison with the F96 scheme. In ZB05, Tsubskin evolves as a function of the surface heat flux, the solar radiation absorbed in the three first meters, and the ocean vertical mixing parameterized based on the Monin–Obukhov theory. Although there is no initialization of the DWL at midnight in ZB05, such an initialization is done here for an easier comparison with F96. Moreover, the same oceanic shortwave radiation absorption profile as in F96 is used with ZB05. This reduces the ZB05-diagnosed DSA of about 15% with regard to the original version of the model. Apart from the changes mentioned above, both codes have been run with their original set of parameters. Because ZB05 does not consider the effect of rain, this field is also set to zero in F96.

b. The ERA-40 hourly forcing

The forcing of F96 and ZB05 requires hourly data of surface radiative and turbulent fluxes. These hourly data are computed from the 6-h integrals of the surface fluxes given in ERA-40. These fields are used because they are more reliable over tropical oceans than other meteorological reanalysis products, in particular for the incident solar flux (Scott and Alexander 1999; Betts et al. 2006). Although more reliable, ERA-40 still underestimates the incident solar flux and overestimates the latent heat flux compared to observations (Feng and Li 2006). This overestimation (up to 40 W m−2 in annual mean) will tend to reduce the DSA in the tropical oceans except for the central and southwestern Pacific, where ERA-40 actually underestimates the latent heat flux. The large latent heat flux results from a too-large surface humidity (about 0.8 g kg−1), but the annual mean surface wind is in fact underestimated by around 1 m s−1 in the tropics. This last point should give a smaller DWL depth and a larger DSA. There is thus no evident impact of known ERA-40 bias on the DSA.

The 6-h integrals given in ERA-40 are interpolated to hourly values. Turbulent and net infrared fluxes are simply assumed to be constant during the 6-h period. However, the incident solar flux must obviously be estimated, considering the earth rotation. For each 6-h period Δh, we assume a constant albedo and a constant atmospheric transmission τh), defined as
i1520-0442-22-13-3629-e1
where S0 is the solar constant, Sdh) is the 6-h average incident solar flux (W m−2), h and h + Δh are the beginning and the end of the 6-h period, and φ is the solar zenith angle for hour h′, day d, and latitude λ. The hourly average downward solar flux Sd(h) is thus
i1520-0442-22-13-3629-e2
For the oceanic mixed layer temperature Tfnd, we use linearly interpolated values of the SST given by the weekly blended SST analysis product of Reynolds and Smith (1994). This product is partly affected by the presence of DWL but, as already noted in Gentemann et al. (2003) and ZB05, this bias is reduced by the higher weight given to nighttime measurement. In the F96 algorithm, we suppose that the Reynolds SST is the temperature Tfnd at the base of the DWL. As already stated above, this depth is constant and equal to 3 m in the ZB05 algorithm.

Because the ERA-40 reanalyses do not consider DWLs, the surface fluxes and surface parameters are not fully consistent with the diagnosed evolution of the SST. However, the determining factors for the development of a DWL are the incident solar flux, which does not depend strongly on the existence of the DWL (even if shallow convection can develop in the afternoon following the DWL formation), and the surface wind, which only slightly depends on it. The diurnal perturbation of turbulent and infrared flux by the DWL is small in comparison with the diurnal variation of the incident solar flux and should have a marginal impact on the DSA computation. Because the DWL tends to increase the turbulent and radiative fluxes (the ocean cooling effect noted in the introduction), the use of unperturbed ERA-40 fluxes will tend to slightly increase the DSA. With the F96 algorithm, it is also possible to compute these fluxes using the ERA-40 surface parameters (Fairall et al. 2003). However, this is not considered here because it may lead to physical inconsistencies, in particular between the SST and the 2-m atmospheric temperature (note that a test of these two approaches, however, gives small differences in the average DSA distributions, confirming the marginal impact of the diurnal variations of the turbulent and infrared fluxes).

3. Validation of the approach

a. SVP buoys measurements

The drifters of the Surface Velocity Program of the Marine Environmental Data Service (MEDS) measure the ocean temperature at a depth of about 25 cm. In this section, the DSA will refer to the diurnal amplitude of Tdepth at 25 cm (denoted T0.25). The so-called Krig product developed by Hansen and Poulain (1996) gives a very manageable dataset. The Krig algorithm is a projection of the raw SVP data onto a regular 6-h UTC time frame. However, this algorithm uses a 1-day cosine function peaking at noon for the SST instead of the observed nonsinusoidal shape peaking at 1400 LST for DWL (F96). This leads to underestimation of the DSA. In addition, the UTC time frame projection increases this underestimation, particularly for those longitudes where the 6-h time sampling gives points far from 1400 LST. The comparison between raw and Krig data (Fig. 2) clearly shows this underestimate of the DSA, especially for strong DWL episodes. We thus developed a new algorithm (see the appendix) that aims to better estimate the measured DSA from the raw dataset. The principle of this algorithm is to suppress the effect of wrong measurements by finding the best fit with an idealized diurnal shape. The result reported in Fig. 2 illustrates the good match between the raw and the processed datasets, giving good confidence in the derived DSA.

The comparison between SVP and F96/ZB05 is not straightforward because of different error sources in the two datasets. For buoy measurements, the absorption of the solar radiation by the mechanical structure of the buoy may lead to overestimation of the maximum temperature. Also, the water mixing induced by the buoy changes the vertical temperature profile by transferring heat downward, leading to a reduction in the equivalent depth of the temperature measurement. This platform effect (Kawai and Kawamura 2000) depends on the buoy geometry and on the wave characteristics and is not well quantified for SVP buoys. Also, the DSA may strongly vary on scales of a few kilometers (Soloviev and Lukas 1997). The comparison between DSA measured by a buoy and diagnosed for a 2.5° region will thus lead to a relatively strong dispersion. A bilinear interpolation of F96/ZB05 temperatures at the location of the buoy improves the comparison, but it also tends to further reduce the dynamics of the DSA with regard to SVP. As an example, evolutions of T0.25 derived from interpolated F96/ZB05 models are shown in Fig. 2. The DSA is smaller for F96/ZB05, but periods of weak and strong DSA are well captured. The smaller DSA for F96/ZB05 is expected because of the local buoy measurement compared to the 2.5° averages that serve as a basis for F96 and ZB05 estimates of T0.25 at the location of the buoy. Note that the Reynolds SST used as Tfnd overestimates the nighttime T0.25 for F96/ZB05. This is also expected because daylight values of the satellite SST estimate are used in the Reynolds analysis. This should, however, have a small impact on the value of the DSA because the ERA-40 fluxes are computed on the basis of this value of Tfnd.

To moderate the contrast between local and large-scale estimates, 3-day running means of the DSA are considered for both SVP measurements and F96/ZB05 models. These comparisons are made using ∼50 000 DSA values. These points, however, are not regularly distributed over the tropical oceans (Fig. 3). The surface drifters tend to be trapped in the subtropical highs, in regions with strong low-level trade winds that prevent DWL formation. The SVP DSA distribution will be then shifted toward low values, reducing the weight of strong DWL for the validation.

The correlation between measured and modeled DSA for T0.25 varies between 0.68 and 0.75 depending on the ocean basin and the DWL model considered (Fig. 4). This correlation is comparable for the two DWL models, certainly because both computations use the same ERA-40 atmospheric and radiative forcings. Linear regressions of buoys as a function of the modeled DSA have offsets of about 0.1 K and slopes between 0.86 and 1.1. The small positive offset could be caused by the buoy warming due to solar radiation.

Both F96 and ZB05 tend to overestimate the number of DSA smaller than 1.5 K and underestimate the number of DSA larger than 1.5 K (Fig. 5a). This tendency is stronger for ZB05. There are only few cases in which the buoys measure large DSA; F96/ZB05 models forced by ERA-40 identify no DWL (Fig. 5b). In this case, there are only a few SVP measurements with a DSA larger than 0.5 K (Fig. 5b). Hence, despite the expected shortcomings of this type of comparison (large dispersion, more extreme values for local measurements), there is a relatively good agreement in the distribution of the DSA values between buoy observations and models.

b. Satellite measurements

In certain conditions, especially for large DSAs, there may be a relatively strong temperature gradient between the surface and 25 cm. Thus, a good agreement between observed and modeled DSA at 25 cm does not guarantee such an agreement for Tsubskin. Indeed, ZB05 uses a fixed mathematical formulation for the temperature profile, giving a DSA for Tsubskin roughly twice as large as the DSA for T0.25. In the F96 algorithm, the DWL depth and the vertical temperature gradient are variable from one event to another. This gradient tends to increase for large DSAs and the relation between Tsubskin and T0.25 DSA is not linear (Fig. 6). In their Fig. 5, Soloviev and Lukas (1997) show that only the strongest DWLs should correspond to the sharp temperature gradient in the first 3 m that is actually used in the ZB05 scheme. For weaker DWLs, the temperature is homogeneous on a deeper layer and ZB05 certainly overestimates the Tsubskin DSA. For the F96 scheme, the gradient between Tsubskin and T0.25 could be too small for strong DWL (Fig. 6), leading to underestimation of Tsubskin DSA compared to the Soloviev and Lukas (1997) results.

To verify the DSA distribution given by F96 and ZB05 for Tsubskin, a comparison is made using the empirical relation deduced from Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) observations by Gentemann et al. (2003). This empirical relation gives the DWL perturbation of Tsubskin at a given local time as a function of the daily average insolation at the top of the atmosphere (TOA) and of the daily average surface wind speed. This relation was obtained and is valid for clear-sky conditions only. However, radiative transfer calculations (see, e.g., Chou and Zhao 1997) show that surface clear-sky SW fluxes are weakly sensitive to water vapor and ozone amounts in the tropical atmosphere. Using their results for precipitable water between 40 and 55 mm, typical of tropical regions, gives a ratio between the surface and the TOA insolation of around 0.75. The Gentemann et al. (2003) empirical relation can thus be extended to estimate the diurnal amplitude of the SST in all sky conditions by considering that their TOA insolation indeed represents a surface insolation divided by 0.75. We compare here the DSA obtained for Tsubskin with F96 and ZB05 to the amplitude deduced from the Gentemann et al. (2003) relation at 1400 LST. The agreement between the observed and the modeled DSA for Tsubskin is good for the F96 algorithm, with similar shape and values for the variation of the DSA as a function of daily average surface wind and insolation (Figs. 7a,c). However, these relations are biased for the ZB05 algorithm (Figs. 7b,d), which strongly overestimates the DSA for surface wind speeds larger than 1 m s−1. This may be attributed to the fixed and sharp temperature profile used in ZB05 that is realistic only at very low wind speed (see Soloviev and Lukas 1997).

The dispersion of the DSA for Tsubskin in each interval of daily average surface wind and insolation reflects nonlinearity in the dependency of the DSA on the hourly variability of the atmospheric state (surface wind, cloudiness). Also, part of this dispersion is due to near-surface vertical gradients of temperature and humidity that modulate the surface turbulent fluxes, even for a given surface wind. For F96 (Figs. 7a,c), the standard deviation of the DSA is up to 0.8 K for low wind speed and up to 0.4 K for wind speed around 4 m s−1. This enables us to evaluate the error caused by computing the DSA using empirical relations based on daily averaged parameters. Note that for ZB05, the dispersion is maximal for average wind speeds of around 4 m s−1 (Figs. 7b,d), which can be attributed to the high sensitivity of ZB05 (constant profile shape) and to the larger standard deviation of the wind speed for these wind interval.

4. The DWL climatology

a. DSA monthly climatology

Only F96 will be considered in the following section because it appears to perform better than the ZB05 algorithm for weak and medium DWL (these DWLs dominate the distribution; Fig. 5). Monthly mean maps of the DSAs obtained using F96 are shown in Fig. 8. The results are in good agreement with previous observational and modeling studies (Stuart-Menteth et al. 2003; Clayson and Weitlich 2007; Kennedy et al. 2007; Bernie et al. 2007; Kawai and Wada 2007). The largest average DSAs are obtained over the northern Indian Ocean in March and April, over the Timor and Arafura Seas (north of Australia) from September to November, over the eastern equatorial Pacific from February to April, and near the Gulf of California almost throughout the year (with average values up to 1.8 K in July). The strongest seasonal variations of DSA are observed in the Indian Ocean in relation to the seasonal march of the monsoon. In particular, the strong DSA over the whole northern Indian Ocean decays in May south of the Bay of Bengal in association with low-level wind burst related to premonsoon intraseasonal events (Flatau et al. 2001; Bellenger and Duvel 2007). Between June and August, the strong low-level monsoon jet prevents the formation of DWL in the whole northern Indian Ocean, but a relative DSA maximum appears over the northwestern Pacific Ocean, where the low-level wind is still weak (Bellenger and Duvel 2007). During boreal winter, the DSA is maximal south of the equator in the Indian Ocean and north of Australia. The low-level wind bursts associated with the intraseasonal variability over these regions, however, is likely to strongly influence DWL formation (see below). Other interesting details appear on these monthly maps, such as the effect of the seasonal migration of the ITCZ over the tropical Atlantic, with calm conditions during boreal summer over a band oriented southwest–northeast between the Amazon and the African coast. Also, some DWL activity remains over the equatorial Indian Ocean during boreal summer (June to September), with calm conditions between the trade winds to the south and the monsoon jet to the north. There are also calm conditions giving large DSA in the Mozambique Channel between October and January.

A by-product of the F96 algorithm is the estimate of the DWL depth for each time step. This depth can be used to better understand the origin of the DWL and also as a basis of comparison for DWL simulated by high-resolution OGCMs. The good agreement with both SVP and satellite observations for the DSAs computed by F96 (for T0.25 and Tsubskin, respectively) gives some confidence in these DWL depths. As an illustration, the monthly mean depth corresponding to the mean DSA is also reported in Fig. 8. Only days with an actual DWL are considered. Monthly mean DWL depth is clearly related to the monthly average DSA, with a depth generally smaller than 5 m for DSAs larger than 0.8–1.0 K. The effect of the solar flux is also clearly evident by comparing the Northern and Southern Hemispheres in June and December. For identical average DWL depths, the DSA is obviously larger in the summer hemisphere. The relation between the DWL depth and the DSA, however, is not simple and the relation between monthly average values should be considered with care. For daily values (not shown), the dispersion of the DSA values increases for smaller DWL depth, with DSA values between 0 and 3.5 K for a depth of 1 m, between 0 and 1 K for a depth of 5 m, and between 0 and 0.5 K for a depth of 10 m.

b. DWL impact on surface fluxes

As shown by Cornillon and Stramma (1985), a larger SST during the day increases the surface infrared and the turbulent heat fluxes. This decreases the net heat flux into the ocean and cools the ocean mixed layer. Cornillon and Stramma (1985) found an average reduction at around 5 W m−2 of the net surface heat flux for a large region of the north Atlantic, in agreement with the 4 W m−2 reported by F96 for the TOGA COARE region. This could be considered as small in regard to the uncertainty in both the measurement and the calculation of these fluxes. However, as noted by Cornillon and Stramma (1985), this energy loss is always of the same sign and occurs over large regions (see also the section below).

This section presents an evaluation of the perturbation in the surface flux (infrared cooling, latent and sensible heat fluxes) and in the ocean mixed layer temperature due to the development of DWLs. The fluxes are computed using temperature and moisture at 2 m and wind at 10 m given by ERA-40. The turbulent heat fluxes are determined using the bulk algorithm of Fairall et al. (2003). Upward surface infrared radiation is simply computed by assuming a surface emissivity of 0.97. The flux perturbation due to the DWL is evaluated every hour by the difference between surface fluxes with and without the diurnal SST perturbation given by F96. Considering that the SST used in ERA-40 is closer to the actual nighttime SST, the temperature at 2 m in ERA-40 is cooler during daytime than the actual temperature with the DWL. This tends to overestimate the DWL heat flux perturbation. It is nevertheless interesting because it gives an upper limit of the ocean cooling due to DWLs if one considers only their direct effect on the surface fluxes (i.e., without dynamical ocean processes and indirect effect related to, e.g., the diurnal variation of cloudiness on the incident shortwave radiation).

The annual mean surface flux out of the ocean (Fig. 9) increases up to 9 W m−2 in regions of strong DWL (the eastern Pacific Ocean, the eastern Indian Ocean northwest of Australia, and the western equatorial Indian Ocean). The latent heat flux accounts for two thirds of this difference, and the infrared and sensible heat fluxes give a comparable contribution of one sixth. The associated mixed layer cooling is evaluated using the de Boyer Montégut et al. (2004) monthly mixed layer depth climatology. Regions strongly affected by DWL tend to be regions with a shallow mixed layer because it corresponds to weak surface wind. This augments the local effect of the DWL on the mixed layer temperature. Using our evaluation of the DWL surface flux perturbation (Fig. 9a), this leads to an average mixed layer cooling up to 2.5 K yr−1 in some regions, such as the tropics (Fig. 9b). These values can reach 0.5 K month−1 over northwest Australia in October and 0.8 K month−1 over the eastern Pacific in March. Thus this an important tendency that should be taken into account for a correct representation of the SST field in coupled models. Note that the cool skin effect will tend to reduce this effect by increasing the net flux into the ocean (F96). However, because the cool skin effect is more homogeneously distributed, a large part of the regional mixed layer temperature contrast given by the DWL is still applicable and may impact the tropical climate variability at different time scales. The distribution of the increase of the daily average surface flux due to the DWL (Fig. 10) shows that there are many cases with an augmentation larger than 10 W m−2 due to the DWL that counterbalance, for example, the effect of the cool skin (11 W m−2) observed during TOGA COARE (F96).

c. Horizontal extension and persistence of the DWLs

Whereas Soloviev and Lukas (1997) showed that DWL characteristics may vary for distances as short as a few kilometers, Stramma et al. (1986) and Webster et al. (1996) also showed DWLs organized at a scale of a few hundred kilometers. The DWL diagnostic using ERA-40 can be used to provide further statistics on the large-scale organization of DWLs. The size of a DWL is addressed by detecting adjacent grid points having a DSA larger than a given threshold. This size is expressed as an equivalent radius computed assuming a disc shape for the surface covered by these adjacent grid points. If a DWL reached our limit of 30° of latitude, it was not considered because its size is not representative (a part is missing). This affects in particular DWLs developing poleward of the main oceanic subtropical high during the summer season (see Fig. 8).

Figure 11 shows the annual average equivalent radius cumulative distribution for different DSA thresholds. The first threshold of 0.68 K corresponds to the average value of the DSA obtained with F96 (0.31 K), augmented by a standard deviation (0.37 K). For this threshold, there are some DWLs with equivalent radii larger than 1800 km. To give an idea of the temporal occurrence of these large DWLs, an equivalent radius larger than 1000 km appears between 2 and 3 times a year for a threshold of 1.42 K and between 2 and 3 times a week for a threshold of 0.68 K. Although this requires further studies, DWLs larger than 1000 km could be sufficiently large to initiate large-scale convective perturbations associated, for example, with the initiation of intraseasonal events.

After attributing a DWL size for each region affected by a particular DWL episode, it is possible to compute the probability that a region is affected by large DWLs. Such large DWLs could have an impact on the triggering of large-scale organized convection and should be represented in GCMs. For a DSA threshold of 0.68 K, we consider here that large DWLs are defined by equivalent radii larger than 1000 km. These large DWLs (Fig. 12) occur mostly in regions and seasons of high average DSA (Fig. 8), in particular in the equatorial eastern Pacific Ocean and the northern Indian Ocean during spring and north of Australia during fall. For these regions and seasons, these large DWLs occur up to about 20% of the time.

The persistence of a DWL for a 2.5° × 2.5° region is simply the number of consecutive days with a DSA larger than a given threshold. As expected, the number of DWL episodes (Fig. 13) decreases very rapidly as the duration increases. There are, however, some regions with durations larger than 40 days (up to 90 days) for DSAs larger than 0.68 K and durations larger than 15 days (up to 40 days) for DSAs larger than 1.42 K. For a threshold of 0.68 K, the average duration is 2.6 days and the standard deviation is 2 days. For each region, it is possible to compute the probability that a region is affected by a persistent DWL. For a DSA threshold of 0.68 K, we consider here that persistent DWLs are defined by durations larger than 5 days. As for large DWLs (Fig. 12), persistent DWLs (Fig. 14) occur mostly in regions and seasons of high average DSA (Fig. 8). For these regions and seasons, these persistent DWLs occur more than 20% of the time and up to 45%–50% in the Timor Sea in September–November (SON) and in the eastern equatorial Pacific in March–May (MAM). For these two regions, DWLs with durations larger than 10 days (not shown) occur about 20% of the time.

Regions and seasons with large average DSA are thus also regions and seasons with persistent and extended DWLs. This must be attributed to the atmospheric forcing causing relatively long periods of suppressed convection and weak surface wind conditions over relatively large regions in relation to the seasonal migration of the ITCZ. The existence and frequency of occurrence of large and persistent DWLs illustrate the fact that DWLs should be represented in GCM even at coarse horizontal resolutions. In addition, DWLs can be sufficiently large to help initiate large-scale convective perturbation. It is striking that extended and persistent DWLs occur mostly over monsoon regions (the northern Indian Ocean and the Timor Sea), just prior to monsoon onsets (April and November, respectively). These seasons also precede the large intraseasonal events associated with these onsets (Bellenger and Duvel 2007). During the summer monsoon season, the DWLs nearly disappear over the northern Indian Ocean but still persist over the northwestern Pacific, in agreement with the seasonal migration of the intraseasonal signal. During December–February (DJF), the season with the largest intraseasonal signal south of the equator, persistent and extended DWLs are present over the western Indian Ocean and over the Timor Sea.

d. Intraseasonal variability

As already mentioned in the introduction, these DWLs can be related to suppressed wind conditions preceding the triggering of the convection and produce an enhanced variability of the SST at intraseasonal time scales. In a case study, Woolnough et al. (2007) showed that MJO predictability is significantly improved when DWL-related SST diurnal variations are taken into account in GCM simulations. The role of these organized DWL episodes in the intraseasonal oscillation, however, is still unclear and requires further investigation. A simple approach is proposed here to give a first estimate of the relation between the intraseasonal convective variability and the DSA. To this end, composite maps of the DSA are constructed for four different phases of the intraseasonal oscillation (ISO). The ISO phases are detected using the National Oceanic and Atmospheric Administration’s (NOAA’s) outgoing longwave radiation (OLR) time series (Liebmann and Smith 1996) as a proxy for tropical convective activity. The OLR time series for the equatorial region at 90°E is filtered between 20 and 90 days using the method described in Bellenger and Duvel (2007) and ISO events with a filtered signal larger than 80 W m−2 are considered. This region is selected because it is strongly affected by the ISO during both boreal summer and winter seasons (see, e.g., Duvel and Vialard 2007). Using this approach, we retain 21 strong ISO events during boreal summer months (May–October) and 18 strong ISO events during boreal winter months (November–April).

The perturbation of the DSA is maximal during the boreal winter (Fig. 15, top) in association with a larger average DSA (Fig. 8) for the corresponding months (especially March and April). This perturbation is especially strong in the Indian Ocean (partly but not exclusively due to the choice of the reference region for the composite), with a clear increase of the DSA during the convectively suppressed phase of the ISO (positive OLR anomaly over the basin; Fig. 15a). The DSA decreases progressively as the convection develops first to the west (Fig. 15b) and then over the entire basin (Fig. 15c). The DSA begins to increase over the western Indian Ocean in the following phase (Fig. 15d). The DSA is at a maximum northwest of Australia in a phase corresponding to low surface wind conditions when the convection is maximal over the eastern Indian Ocean (Fig. 15c). This strong DSA to the east of the convective perturbation will favor its eastward propagation because a strong DSA is associated with a higher SST, which increases the convective instability.

During boreal summer months (Fig. 15, bottom) the perturbation of the DSA by the ISO is smaller. This is expected because of the strong trade winds and the monsoon flux, which respectively blow south and north of the equator. However, it is interesting to note that the larger DSA at the equator (Fig. 8) is strongly modulated by the ISO of the convection, with a maximum value for the suppressed phase (Fig. 15e). The development of a DWL in this equatorial region could be an important factor in the development of boreal summer ISO over the Indian Ocean. The lack of DSA modulation in the northwestern Pacific is due to the simple ISO index that is used here and to the fact that the ISO phase relation between the Indian and Pacific Oceans during boreal summer is not well reproducible from one event to another (Duvel and Vialard 2007).

5. Summary and conclusions

A long time series of daily DWL characteristics is constructed using simple DWL models forced hourly using ERA-40 fields between 1979 and 2002. This approach is applied using the F96 and the ZB05 DWL models. Daily values of the DSA obtained by this approach are compared globally to in situ buoy measurements. Despite some expected differences due to the different spatial scales, the DSAs at a depth of 25 cm deduced from the buoy measurements and from our approach are well correlated for both DWL models. The average DSAs of the skin temperature in given intervals of daily average wind and solar flux are also compared to previous satellite estimates obtained by Gentemann et al. (2003). The F96 DWL model gives a DSA distribution, consistent with the satellite observation, but the ZB05 model gives a too-large DSA for medium wind and solar flux values. This difference may be attributed to the constant DWL depth of 3 m and a constant temperature profile shape for ZB05, which tends to overestimate the DSA in medium wind speed conditions. In contrast, the F96 model tends to underestimate the DSA for very low wind speed condition. In the future, this approach could be refined using a more consistent DWL model for all conditions, such as the Profiles of Surface Heating (POSH) model of Gentemann (2007).

The spatial distribution of monthly-mean DSA obtained with F96 is also well comparable with previous estimates from satellite data. This result can be attributed in a large part to the good representation of the low-wind conditions over the tropical ocean by ERA-40. This daily climatology, adapted to 2.5° regions, is used to analyze the potential role of the DWLs in the variability of the tropical climate. The annual average perturbation of the surface fluxes by the DWL can be as large as 9 W m−2, giving a cooling of the ocean mixed layer as large as 2.5 K yr−1 in some regions (e.g., the tropics). On a daily basis, this flux perturbation is often larger than 10 W m−2 and larger than the cool skin effect estimated by F96.

Some examples are also shown here on the size and the duration of the DWLs. The DWLs associated with relatively strong DSAs (more than 0.68 K for F96) are organized by region up to a few thousand kilometers and can persist for more than 5 days. Considering monthly mean averages, the stronger, more extended, and more persistent DWLs are interestingly located in regions of strong intraseasonal oscillation of the convection. This deserves further study, taking into account different intraseasonal events and their relation with DWL formations. A first simple approach shows a very clear relation between DWL and the phase of the ISO, with stronger DWLs developing during the suppressed phase of the ISO in regions that later experience large convective perturbations.

This study suggests that the same approach could be used as a parameterization of DWLs for forced atmospheric models or for coupled models using OGCM with a low vertical resolution (i.e., 10 m) near the surface. For the F96 scheme, the correct DSA distribution obtained using ERA-40 surface forcing shows that this algorithm, although designed for local in situ measurements, is indeed adapted for large-scale forcing typically given by atmospheric GCMs. For this purpose, as explained in the introduction, F96 may be considered as a parameterization giving the diurnal evolution of the SST for a more precise flux computation in the atmospheric GCM. These DWLs modify the daily mean turbulent flux of heat, water, and momentum and may thus have an influence on the variability of the simulated tropical climate at different time scales. Using GCM simulations with and without a representation of the DWL will thus allow us to test the sensitivity of the tropical climate variability to this phenomenon.

Acknowledgments

We cordially thank Frederic Vitart for positive discussions and Anton Beljaars for useful comments and help with the use of his warm layer model. The comments of four anonymous reviewers were very useful to improve this article.

REFERENCES

  • Bellenger, H., , and J. P. Duvel, 2007: Intraseasonal convective perturbations related to the seasonal march of the Indo-Pacific monsoons. J. Climate, 20 , 28532863.

    • Search Google Scholar
    • Export Citation
  • Bernie, D. J., , S. J. Woolnough, , J. M. Slingo, , and E. Guilyardi, 2005: Modeling diurnal and intraseasonal variability of the ocean mixed layer. J. Climate, 18 , 11901202.

    • Search Google Scholar
    • Export Citation
  • Bernie, D. J., , E. Guilyardi, , G. Madec, , S. Woolnough, , and J. Slingo, 2007: Impact of resolving the diurnal cycle in an ocean–atmosphere GCM. Part 1: A diurnally forced OGCM. Climate Dyn., 29 , 575590.

    • Search Google Scholar
    • Export Citation
  • Betts, A. K., , M. Zhao, , P. A. Dirmeyer, , and A. C. M. Beljaars, 2006: Comparison of ERA-40 and NCEP/DOE near-surface data sets with other ISLSCP-II data sets. J. Geophys. Res., 111 , D22S04. doi:10.1029/2006JD007174.

    • Search Google Scholar
    • Export Citation
  • Chou, M. D., , and W. Zhao, 1997: Estimation and model validation of surface solar radiation and cloud radiative forcing using TOGA COARE measurements. J. Climate, 10 , 610620.

    • Search Google Scholar
    • Export Citation
  • Clayson, C. A., , and D. Weitlich, 2007: Variability of tropical diurnal sea surface temperature. J. Climate, 20 , 334352.

  • Cornillon, P., , and L. Stramma, 1985: The distribution of diurnal sea surface warming events in the western Sargasso Sea. J. Geophys. Res., 90 , 1181111815.

    • Search Google Scholar
    • Export Citation
  • Cronin, M. F., , and M. J. McPhaden, 1997: The upper ocean heat balance in the western equatorial Pacific warm pool during September–December 1992. J. Geophys. Res., 102 , 85338553.

    • Search Google Scholar
    • Export Citation
  • de Boyer Montégut, C., , G. Madec, , A. S. Fischer, , A. Lazar, , and D. Iudicone, 2004: Mixed layer depth over the global ocean: An examination of profile data and a profile-based climatology. J. Geophys. Res., 109 , C12003. doi:10.1029/2004JC002378.

    • Search Google Scholar
    • Export Citation
  • Donlon, C., and Coauthors, 2007: The global ocean data assimilation experiment high-resolution sea surface temperature pilot project. Bull. Amer. Meteor. Soc., 88 , 11971213.

    • Search Google Scholar
    • Export Citation
  • Duvel, J. P., , and J. Vialard, 2007: Indo-Pacific sea surface temperature perturbations associated with intraseasonal oscillations of the tropical convection. J. Climate, 20 , 30563082.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., , E. F. Bradley, , J. S. Godfrey, , G. A. Wick, , J. B. Edson, , and G. S. Young, 1996: Cool-skin and warm-layer effects on sea surface temperature. J. Geophys. Res., 101 , 12951308.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., , E. F. Bradley, , J. E. Hare, , A. A. Grachev, , and J. B. Edson, 2003: Bulk parameterization of air–sea fluxes: Updates and verification for the COARE algorithm. J. Climate., 16 , 571591.

    • Search Google Scholar
    • Export Citation
  • Feng, L., , and J. Li, 2006: A comparison of latent heat fluxes over global oceans for ERA and NCEP with GSSTF2. Geophys. Res. Lett., 33 , L03810. doi:10.1029/2005GL024677.

    • Search Google Scholar
    • Export Citation
  • Feng, M., , R. Lukas, , P. Hacker, , R. A. Weller, , and S. P. Anderson, 2000: Upper-ocean heat and salt balances in the western equatorial Pacific in response to the intraseasonal oscillation during TOGA COARE. J. Climate, 13 , 24092427.

    • Search Google Scholar
    • Export Citation
  • Flament, P., , J. Firing, , M. Sawyer, , and C. Trefois, 1994: Amplitude and horizontal structure of a large diurnal sea surface warming event during the coastal ocean dynamics experiment. J. Phys. Oceanogr., 24 , 124139.

    • Search Google Scholar
    • Export Citation
  • Flatau, M. K., , P. J. Flatau, , and D. Rudnick, 2001: The dynamics of double monsoon onsets. J. Climate, 14 , 41304146.

  • Gentemann, C. L., 2007: Diurnal warming at the ocean surface. Ph.D. thesis, Dept. of Meteorology and Physical Oceanography, University of Miami, 173 pp.

  • Gentemann, C. L., , C. J. Donlon, , A. Stuart-Menteth, , and F. J. Wentz, 2003: Diurnal signals in satellite sea surface temperature measurements. Geophys. Res. Lett., 30 , 1140. doi:10.1029/2002GL016291.

    • Search Google Scholar
    • Export Citation
  • Hansen, D. V., , and P-M. Poulain, 1996: Quality control and interpolations of WOCE–TOGA drifter data. J. Atmos. Oceanic Technol., 13 , 900909.

    • Search Google Scholar
    • Export Citation
  • Horrocks, L. A., , B. Candy, , T. J. Nightingale, , R. W. Saunders, , A. O’Caroll, , and A. R. Harris, 2003: Parameterizations of the ocean skin effect and implications for satellite-based measurement of sea-surface temperature. J. Geophys. Res., 108 , 3096. doi:10.1029/2002JC001503.

    • Search Google Scholar
    • Export Citation
  • Kawai, Y., , and H. Kawamura, 2000: Study on a platform effect in the in situ sea surface temperature observations under weak wind and clear sky conditions using numerical models. J. Atmos. Oceanic Technol., 17 , 185196.

    • Search Google Scholar
    • Export Citation
  • Kawai, Y., , and H. Kawamura, 2002: Evaluation of the diurnal warming of sea surface temperature using satellite-derived marine meteorological data. J. Oceanogr., 58 , 805814.

    • Search Google Scholar
    • Export Citation
  • Kawai, Y., , and A. Wada, 2007: Diurnal sea surface temperature variation and its impact on the atmosphere and ocean: A review. J. Oceanogr., 63 , 721744.

    • Search Google Scholar
    • Export Citation
  • Kennedy, J. J., , P. Brohan, , and S. F. B. Tett, 2007: A global climatology of the diurnal variations in sea surface temperature and implications for MSU temperature trends. Geophys. Res. Lett., 34 , L05712. doi:10.1029/2006GL028920.

    • Search Google Scholar
    • Export Citation
  • Kondo, J., , Y. Sasano, , and T. Ishii, 1979: On wind-driven current and temperature profiles with diurnal period in the oceanic planetary boundary layer. J. Phys. Oceanogr., 9 , 360372.

    • Search Google Scholar
    • Export Citation
  • Large, W. G., , J. C. McWilliams, , and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parameterization. Rev. Geophys., 32 , 363403.

    • Search Google Scholar
    • Export Citation
  • Liebmann, B., , and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77 , 12751277.

    • Search Google Scholar
    • Export Citation
  • Mellor, G. L., , and T. Yamada, 1982: Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys., 20 , 851875.

    • Search Google Scholar
    • Export Citation
  • Price, J. F., , R. A. Weller, , and R. Pinkel, 1986: Diurnal cycling: Observations and models of the upper ocean response to diurnal heating, cooling, and wind mixing. J. Geophys. Res., 91 , 84118427.

    • Search Google Scholar
    • Export Citation
  • Rayner, N. A., , P. Brohan, , D. E. Parker, , C. K. Folland, , J. J. Kennedy, , M. Vanicek, , T. J. Ansell, , and S. F. Tett, 2006: Improved analyses of changes and uncertainties in sea surface temperature measured in situ since the mid-nineteenth century: The HadSST2 data set. J. Climate, 19 , 446469.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , and T. M. Smith, 1994: Improved global sea surface temperature analyses using optimum interpolation. J. Climate, 7 , 929948.

    • Search Google Scholar
    • Export Citation
  • Saunders, P. M., 1967: The temperature at the ocean–air interface. J. Atmos. Sci., 24 , 269273.

  • Schiller, A., , and J. S. Godfrey, 2005: A diagnostic model of the diurnal cycle of sea surface temperature for use in coupled ocean–atmosphere models. J. Geophys. Res., 110 , C11014. doi:10.1029/2005JC002975.

    • Search Google Scholar
    • Export Citation
  • Scott, J. D., , and M. A. Alexander, 1999: Net shortwave fluxes over the ocean. J. Phys. Oceanogr., 29 , 31673174.

  • Shinoda, T., 2005: Impact of the diurnal cycle of solar radiation on intraseasonal SST variability in the western equatorial Pacific. J. Climate, 18 , 26282636.

    • Search Google Scholar
    • Export Citation
  • Shinoda, T., , and H. H. Hendon, 1998: Mixed layer modeling of intraseasonal variability in the tropical western Pacific and Indian Oceans. J. Climate, 11 , 26682685.

    • Search Google Scholar
    • Export Citation
  • Soloviev, A., , and R. Lukas, 1997: Observation of large diurnal warming events in the near-surface layer of the western equatorial Pacific warm pool. Deep-Sea Res. I, 44 , 10551076.

    • Search Google Scholar
    • Export Citation
  • Stramma, L., , P. Cornillon, , R. A. Weller, , J. F. Price, , and M. G. Briscoe, 1986: Large diurnal sea surface temperature variability: Satellite and in situ measurements. J. Phys. Oceanogr., 16 , 827837.

    • Search Google Scholar
    • Export Citation
  • Stuart-Menteth, A. C., , I. S. Robinson, , and P. G. Challenor, 2003: A global study of diurnal warming using satellite-derived sea surface temperature. J. Geophys. Res., 108 , 3155. doi:10.1029/2002JC001534.

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

  • Ward, B., 2006: Near-surface ocean temperature. J. Geophys. Res., 111 , C02004. doi:10.1029/2004JC002689.

  • Webster, P. J., , C. A. Clayson, , and J. A. Curry, 1996: Clouds, radiation, and the diurnal cycle of sea surface temperature in the tropical western Pacific. J. Climate, 9 , 17121730.

    • Search Google Scholar
    • Export Citation
  • Weller, R., , and S. Anderson, 1996: Surface meteorology and air–sea fluxes in the western equatorial Pacific warm pool during the TOGA coupled ocean–atmosphere response experiment. J. Climate, 9 , 19591991.

    • Search Google Scholar
    • Export Citation
  • Woolnough, S. J., , F. Vitart, , and M. A. Balmaseda, 2007: The role of the ocean in the Madden–Julian Oscillation: Implications for MJO prediction. Quart. J. Roy. Meteor. Soc., 133 , 117128.

    • Search Google Scholar
    • Export Citation
  • Zeng, X., , and A. Beljaars, 2005: A prognostic scheme of sea surface skin temperature for modeling and data assimilation. Geophys. Res. Lett., 32 , L14605. doi:10.1029/2005GL023030.

    • Search Google Scholar
    • Export Citation

APPENDIX

Estimate of the Diurnal Amplitude from the Raw SVP Dataset

The dataset of raw SVP buoy temperature measurements is available from the Marine Environmental Data Service (MEDS). These temperatures are used as a proxy for Tsubskin and Tdepth at 25 cm. These raw data need to be cleaned and interpolated along the diurnal cycle. To this end, the following procedure is applied.

The average temperature between 0100 and 0800 LST is assumed to represent the nocturnal temperature minimum (Fig. A1). This average value and the standard deviation are computed for each day. Following the procedure of Rayner et al. (2006) and Kennedy et al. (2007), every point that differs from this average by more than 1.25 times the standard deviation is removed and the nocturnal average is computed again. This procedure suppresses wrong nocturnal points. Using these nocturnal averages, the trend between the previous and corresponding nights is then computed and subtracted from the buoy temperature time series of the corresponding day in order to remove the interdiurnal evolution.

The shape of the diurnal temperature evolution is given by a canonical function. This canonical function is constructed using the first three diurnal harmonics of a typical average hourly temperature evolution. This average hourly temperature evolution is computed using the F96 scheme (at a depth of 0.25 m), forced by ERA-40, for the locations of the SVP buoy 34158 for the first 2 months of measurement (November–December 2001 in the near-equatorial Indian Ocean). The canonical function is
i1520-0442-22-13-3629-ea1
where the coefficients ak and bk are given in Table A1. This function reproduces the shape of the mean observed diurnal SVP temperature evolution, with a sharp increase in the morning and a smooth decrease during late afternoon and early night (Fig. A1). At this depth, the temperature maximum diagnosed by F96 is reached around 1400 LST, in agreement with the average daily evolution of the raw SVP data (Fig. A1). This maximum occurs only slightly sooner when considering the surface temperature. The cosine function used by Hansen and Poulain (1996) for the Krig method certainly tends to underestimate the DSA. The DSA for each day and each buoy is computed using a least squares fit between the canonical function and the daily SVP measurements. The daily temperature evolution is then supposed to be DSA × f (t). Again, measurements that differ from the fitted function by more than 1.25 times the standard deviation are removed and the regression is redone. To have sufficient information for the least squares approach, this method is applied provided that at least three correct measurements remains for each night and for the period between 0900 to 1900 LST. This leads us to consider around 25% (105 days) of the original dataset.
Fig. 1.
Fig. 1.

Schematic of the upper ocean temperature profile for a diurnal warm layer. The different definitions of the upper ocean temperatures used in the article are reported following Donlon et al. (2007) and Kawai and Wada (2007). The cool skin base (dashed line) corresponds to a depth of around 0.1–1 mm.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 2.
Fig. 2.

(top) Temperature measured by the SVP buoy 34158 the Indian Ocean around 6°S, 81°E for raw measurements (thin) and the Krig time series (bold). (bottom) Results of the interpolation method following the procedure described in the appendix (dotted) and Tdepth at 25 cm obtained using F96 (bold gray) and ZB05 (black) bilinearly interpolated at the buoy’s location.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 3.
Fig. 3.

Number of days of SVP buoy measurements for each 2.5° × 2.5° region for the dataset (1993–2002) used for the validation of the F96 and ZB05 results.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 4.
Fig. 4.

Distribution of 3-day average SVP buoy DSA as a function of a 3-day average of (top) ZB05 and (bottom) F96 DSA, where DSAs are computed for a depth of 25 cm and interpolated at SVP buoy locations for the (left) Indian, (middle) Pacific, and (right) Atlantic Oceans. The contours represent the number of points per 0.05 K × 0.05 K intervals. The contours represent 2, 10, 20, 100, and 200 points per bin. The correlations and the equation of the corresponding best linear fits (bold solid lines) are listed for each ocean basin. The best fit is computed following the vertical offset least squares fitting method.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 5.
Fig. 5.

(a) Daily SVP buoy DSA distribution for 1993–2002 (bold solid) and the corresponding F96 (thin solid) and ZB05 DSA (dashed) computed for a depth of 25 cm and interpolated at the buoy locations. (b) SVP DSA distributions for F96 (thin solid) and ZB05 DSA = 0 (bold dashed).

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 6.
Fig. 6.

Daily F96 DSA for T0.25 (25-cm depth) as a function of F96 DSA for Tsubskin (dots) and the corresponding second-degree polynomial fit (light solid line). The corresponding relation for ZB05 is also shown (bold solid line).

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 7.
Fig. 7.

(bottom) Average and (top) standard deviation of F96 and ZB05 DSA for Tsubskin in given intervals of daily average surface insolation and surface (10 m) wind speed. (a),(b) The variation of the DSA with the surface wind speed for given insolation intervals. (c),(d) The variation of the DSA with the insolation for given surface wind speed intervals. The results from the F96 and the ZB05 models (solid lines) are compared to the empirical relation reported in Gentemann et al. (2003) (dotted lines). The top panels represent the standard deviation of the F96 and ZB05 results for each interval. Each interval is 10 W m−2 wide for insolation and 0.2 m s−1 wide for the surface wind speed. Only bins with at least 100 points are shown.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 8.
Fig. 8.

Monthly average DSA (colors, K) and DWL depth (5 and 7.5 m, thin solid line; 10 m, dotted line) for the period 1979–2002.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 9.
Fig. 9.

(a) Annual mean surface flux perturbation (surface cooling, W m−2) due to the DWL and (b) the corresponding annual mean cooling of the ocean mixed layer (K yr−1).

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 10.
Fig. 10.

Distribution of daily surface flux perturbations due to the DWL for all 2.5° × 2.5° oceanic regions.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 11.
Fig. 11.

Average cumulative distribution (decreasing size) of the DWL equivalent radius for DSAs larger than 0.68 (solid), 1.05 (stripes), and 1.42 K (white).

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 12.
Fig. 12.

Occurrence (% of day) of DWLs with an equivalent radius larger than 1000 km for a DSA > 0.68 K for December–February, March–May, June–August (JJA), and September–November from 1979 to 2002. DWLs reaching 30° of latitude are not considered.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 13.
Fig. 13.

Distribution of DWL episode durations for all 2.5° × 2.5° oceanic regions. For each region, the duration of a DWL episode is the number of consecutive days with a DSA larger than a given threshold. As in Fig. 10, the DSA threshold are 0.68 (solid), 1.05 (stripes), and 1.42 K (white).

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 14.
Fig. 14.

Occurrence (% of day) of DWLs with durations larger than 5 days for a DSA > 0.68 K for DJF, MAM, JJA, and SON from 1979 to 2002.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Fig. 15.
Fig. 15.

Modulation of the DSA (colors) by the intraseasonal oscillation for the (top) November–April (18 ISO events) and (bottom) May–October seasons (21 ISO events) for the period 1979–2002. The four ISO phases are detected based on the filtered OLR signal around 0°, 90°E: (a),(e) maximum OLR; (c),(g) minimum OLR; (b),(d),(f),(h) intermediate phases. The corresponding OLR anomaly is also reported. Dashed (solid) contours indicate −(+)10 W m−2.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

i1520-0442-22-13-3629-fa01

Fig. A1. Average diurnal evolution of the temperature anomalies for raw SVP buoy 34158 measurements (bold solid) and the results of the F96 algorithm at the same locations as the buoy (thin solid). The cosine function (dashed) used by Hansen and Poulain (1996) to build the Krig dataset is also shown.

Citation: Journal of Climate 22, 13; 10.1175/2008JCLI2598.1

Table A1. Coefficients of the three-harmonic function f (t) (A2) used for the evaluation of the diurnal variation of Tdepth (SVP DSA) observed by SVP buoys.

i1520-0442-22-13-3629-ta01
Save