• Anderberg, M. R., 1973: Cluster Analysis for Applications. Academic Press, 359 pp.

  • Barkstrom, B., and Coauthors, 1989: Earth Radiation Budget Experiment (ERBE) archival and April 1985 results. Bull. Amer. Meteor. Soc., 70 , 12541262.

    • Search Google Scholar
    • Export Citation
  • Cess, R. D., and Coauthors, 1990: Intercomparison and interpretation of climate feedback processes in 19 atmospheric general circulation models. J. Geophys. Res., 95 , 1660116615.

    • Search Google Scholar
    • Export Citation
  • Cullen, M. J. P., 1993: The unified forecast/climate model. Meteor. Mag., 122 , 8194.

  • Forbes, R. M., and P. A. Clarke, 2003: Sensitivity of extratropical cyclone mesoscale structure to the parametrization of ice microphysical processes. Quart. J. Roy. Meteor. Soc., 129 , 11231148.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., and G. Tselioudis, 2003: Objective identification of cloud regimes in the tropical western Pacific. Geophys. Res. Lett., 30 .2082, doi:10.1029/2003GL018367.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., G. Tselioudis, and T. Hume, 2005: The radiative, cloud, and thermodynamic properties of the major tropical western Pacific cloud regimes. J. Climate, 18 , 12031215.

    • Search Google Scholar
    • Export Citation
  • Jones, A., D. L. Roberts, and A. Slingo, 1994: A climate model study of indirect radiative forcing by anthropogenic sulphate aerosols. Nature, 370 , 450453.

    • Search Google Scholar
    • Export Citation
  • Klein, S. A., and C. Jakob, 1999: Validation and sensitivities of frontal clouds simulated by the ECMWF model. Mon. Wea. Rev., 127 , 25142531.

    • Search Google Scholar
    • Export Citation
  • Klein, S. A., X. Jiang, J. Boyle, S. Malyshev, and S. Xie, 2006: Diagnosis of the summertime warm and dry bias over the U.S. Southern Great Plains in the GFDL climate model using a weather forecasting approach. Geophys. Res. Lett., 33 .L18805, doi:10.1029/2006GL027567.

    • Search Google Scholar
    • Export Citation
  • Klinker, E., and P. D. Sardeshmukh, 1992: The diagnosis of mechanical dissipation in the atmosphere from large-scale balance requirements. J. Atmos. Sci., 49 , 608627.

    • Search Google Scholar
    • Export Citation
  • Liou, K-N., 1986: Influence of cirrus clouds on weather and climate processes: A global perspective. Mon. Wea. Rev., 114 , 11671199.

  • Liou, K-N., and Q. Zheng, 1984: A numerical experiment on the interactions of radiation, clouds, and dynamic processes in a general circulation model. J. Atmos. Sci., 41 , 15131536.

    • Search Google Scholar
    • Export Citation
  • Martin, G. M., M. A. Ringer, V. D. Pope, A. Jones, C. Dearden, and T. J. Hinton, 2006: The physical properties of the atmosphere in the new Hadley Centre Global Environmental Model (HadGEM1). Part I: Model description and global climatology. J. Climate, 19 , 12741301.

    • Search Google Scholar
    • Export Citation
  • Phillips, T. J., and Coauthors, 2004: Evaluating parameterizations in general circulation models: Climate simulation meets weather prediction. Bull. Amer. Meteor. Soc., 85 , 19031915.

    • Search Google Scholar
    • Export Citation
  • Rawlins, F., S. P. Ballard, K. J. Bovis, A. M. Clayton, D. Li, G. W. Inverarity, A. C. Lorenc, and T. J. Payne, 2007: The Met Office global 4-dimensional data assimilation system. Quart. J. Roy. Meteor. Soc., 133 , 347362.

    • Search Google Scholar
    • Export Citation
  • Ringer, M. A., and Coauthors, 2006: Global mean cloud feedbacks in idealized climate change experiments. Geophys. Res. Lett., 33 .L07718, doi:10.1029/2005GL025370.

    • Search Google Scholar
    • Export Citation
  • Rodwell, M. J., and T. N. Palmer, 2007: Using numerical weather prediction to assess climate models. Quart. J. Roy. Meteor. Soc., 133 , 129146.

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

  • Senior, C. A., and J. F. B. Mitchell, 1993: Carbon dioxide and climate: The impact of cloud parameterization. J. Climate, 6 , 393418.

  • Soden, B. J., and I. M. Held, 2006: An assessment of climate feedbacks in coupled ocean–atmosphere models. J. Climate, 19 , 33543360.

    • Search Google Scholar
    • Export Citation
  • Strachan, J., 2007: Understanding and modelling the climate of the Maritime Continent. Ph.D. thesis, University of Reading.

  • WCRP, 2005: The World Climate Research Programme strategic framework 2005–2015: Coordinated Observation and Prediction of the Earth System (COPES). WCRP-123, WMO/TD 1291, 44 pp.

  • Webb, M. J., C. Senior, S. Bony, and J-J. Morcrette, 2001: Combining ERBE and ISCCP data to assess clouds in the Hadley Centre, ECMWF, and LMD atmospheric climate models. Climate Dyn., 17 , 905922.

    • Search Google Scholar
    • Export Citation
  • Webb, M. J., and Coauthors, 2006: On the contribution of local feedback mechanisms to the range of climate sensitivity in two GCM ensembles. Climate Dyn., 27 , 1738.

    • Search Google Scholar
    • Export Citation
  • Williams, K. D., and G. Tselioudis, 2007: GCM intercomparison of global cloud regimes: Present-day evaluation and climate change response. Climate Dyn., 29 , 231250.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., W. B. Rossow, A. A. Lacis, V. Oinas, and M. I. Mishchenko, 2004: Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets: Refinements of the radiative transfer model and input data. J. Geophys. Res., 109 .D19105, doi:10.1029/2003JD004457.

    • Search Google Scholar
    • Export Citation
  • View in gallery
    Fig. 1.

    Mean CTP–τ regime histograms for the principal cloud regimes over the tropics (20°N–20°S). Shading indicates the cloud amount (%) in each CTP–τ category. (top) Observed climatology by ISCCP (note ISCCP does not observe cloud in the thinnest optical depth category, τ < 0.3); (middle) mean over the 5-day forecast from the N96 model; and (bottom) mean over the 5-day forecast from the N216 model. The model does not simulate the midlevel convection regime.

  • View in gallery
    Fig. 2.

    As in Fig. 1 but for the ice-free extratropics. The model does not simulate the midlevel cloud regime.

  • View in gallery
    Fig. 3.

    Evolution of regime characteristics through the model forecast. N96 is shown solid; N216 is shown dashed. Observed climatology from ISCCP is shown with an asterisk; from ERBE is shown with a diamond; HadGAM1 climatology is shown with a triangle. Note that the ISCCP tropical stratocumulus and deep convective RFO are identical.

  • View in gallery
    Fig. 4.

    Initial temperature tendency profiles for each regime and the total for the whole region over the first 6 h of the N96 forecast.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 119 51 10
PDF Downloads 62 22 2

Initial Tendencies of Cloud Regimes in the Met Office Unified Model

K. D. WilliamsMet Office Hadley Centre, Exeter, United Kingdom

Search for other papers by K. D. Williams in
Current site
Google Scholar
PubMed
Close
and
M. E. BrooksMet Office, Exeter, United Kingdom

Search for other papers by M. E. Brooks in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

The Met Office unified forecast–climate model is used to compare the properties of simulated climatological cloud regimes with those produced in short-range forecasts initialized from operational analyses. The regimes are defined as principal clusters of joint cloud-top pressure–optical depth histograms. In general, the cloud regime properties are found to be similar at all forecast times, including the climatological mean. This suggests that weaknesses in the representation of fast local processes are responsible for errors in the simulation of the cloud regimes. The increased horizontal resolution of the model used for numerical weather prediction generally has little impact on the cloud regimes, although the simulation of tropical shallow cumulus is improved, while the relative frequency of tropical deep convection and cirrus compare less favorably with observations. Analysis of the initial temperature tendency profiles for each cloud regime indicates that some of the initial temperature tendency, which leads to a systematic bias in the model climatology, is associated with a particular cloud regime.

Corresponding author address: Keith Williams, Met Office Hadley Centre, FitzRoy Road, Exeter EX1 3PB, United Kingdom. Email: keith.williams@metoffice.gov.uk

Abstract

The Met Office unified forecast–climate model is used to compare the properties of simulated climatological cloud regimes with those produced in short-range forecasts initialized from operational analyses. The regimes are defined as principal clusters of joint cloud-top pressure–optical depth histograms. In general, the cloud regime properties are found to be similar at all forecast times, including the climatological mean. This suggests that weaknesses in the representation of fast local processes are responsible for errors in the simulation of the cloud regimes. The increased horizontal resolution of the model used for numerical weather prediction generally has little impact on the cloud regimes, although the simulation of tropical shallow cumulus is improved, while the relative frequency of tropical deep convection and cirrus compare less favorably with observations. Analysis of the initial temperature tendency profiles for each cloud regime indicates that some of the initial temperature tendency, which leads to a systematic bias in the model climatology, is associated with a particular cloud regime.

Corresponding author address: Keith Williams, Met Office Hadley Centre, FitzRoy Road, Exeter EX1 3PB, United Kingdom. Email: keith.williams@metoffice.gov.uk

1. Introduction

Differences in the radiative response from clouds account for much of the variation in climate sensitivity between global general circulation models (GCMs) used for climate change projection (e.g., Cess et al. 1990; Senior and Mitchell 1993; Webb et al. 2006; Soden and Held 2006; Ringer et al. 2006). Accurate simulation of cloud is also crucial for numerical weather prediction (NWP) models due to the direct effect on key forecast products, such as surface temperature, precipitation, and visibility, and also the large impact that diabatic processes exert on the evolution of the forecast (Liou and Zheng 1984; Forbes and Clarke 2003). Since individual cloud processes act on relatively short time scales, it is appropriate to consider cloud errors in climate models on NWP time scales and address the cloud simulation in both sets of models in a consistent framework.

Recently, several groups have investigated running climate models in “weather forecast mode” (i.e., initializing from an analysis produced by a data assimilation system) to diagnose the cause of systematic errors (Phillips et al. 2004; Klein et al. 2006; Strachan 2007). Meanwhile, Rodwell and Palmer (2007, hereafter RP07) analyze initial tendencies in state variables (e.g., Klinker and Sardeshmukh 1992) over the first few hours of a weather forecast. They suggest that these initial tendencies may be used to form a metric for the representation of fast processes in a climate model. The Met Office has the unique asset of a unified weather forecast–climate model (UM; Cullen 1993); that is, a GCM with the same physical parameterizations of the atmosphere is used for operational weather forecasts and climate change projection. This enables the climate model to be initialized from an analysis that has been produced by a data assimilation system developed around the same physical model. This “seamless forecasting” approach is recognized by the World Climate Research Program (WCRP) as being important to the continued development of GCMs (WCRP 2005).

Using the concept of “regimes” in the evaluation of GCMs can be useful for both NWP and climate models since they provide a summary of model performance across a range of synoptic situations and tend to be broadly aligned with some of the large-scale atmospheric processes. Jakob and Tselioudis (2003) use a clustering technique on International Satellite Cloud Climatology Project (ISCCP) joint cloud-top pressure–optical depth histograms to objectively identify cloud regimes. Williams and Tselioudis (2007, hereafter WT07) apply the technique to ISCCP simulator output from an ensemble of climate GCMs and demonstrate that a reduction in the range of climate sensitivity amongst the GCMs might be achieved if the simulations of the mean present-day regime characteristics were closer to those observed. In this study, one of the models analyzed by WT07 (the Met Office UM) is used to investigate the simulation of cloud regimes in weather forecast mode. The primary aims are to compare the cloud regimes in the model climatology and in a short-range forecast, to investigate whether the increased resolution of the operational forecast model improves the simulation, and to investigate whether initial tendencies in the model’s state variables may be attributed to regime-specific errors.

2. Model, observations, and methodology

a. Model simulations

The current climate version of the Met Office UM is known as the Hadley Centre Global Atmospheric Model version 1 (HadGAM1; Martin et al. 2006). The standard climate resolution for this model is N96 (1.875° × 1.25°). For this study, the model has been run from a series of operational analyses generated using the Met Office’s four-dimensional variational assimilation system (Rawlins et al. 2007). These analyses are produced at operational resolution; hence, the initial conditions are regridded onto the coarser N96 grid at the start of the first time step. The only difference between the model used for these short-range forecasts and the standard HadGAM1 atmosphere model is that interactive aerosols used in HadGAM1 have been replaced by aerosol climatologies. This alteration was necessary as the operational NWP model does not include interactive aerosols (due to their expense); hence, their initial conditions are not available in the analyses. The NWP model has also been rerun from the same set of analyses using the configuration that was operational from January to December 2005. The unified nature of the model means that the primary difference between the NWP and climate models used here is the horizontal resolution; the NWP model is run at N216 (0.833° × 0.556°). Other differences are mostly related to the resolution change (e.g., diffusion settings, the time scale for the dissipation of convective available potential energy (CAPE) by the convection scheme, etc.).

Twenty-four simulations of both the N96 and N216 models have been run from analyses between 1 November 2005 and 31 October 2006. Each forecast is run for 5 days. ISCCP simulator diagnostics, which aim to emulate the ISCCP observational data (Klein and Jakob 1999; Webb et al. 2001; http://gcss-dime.giss.nasa.gov/simulator.html), and top-of-atmosphere (TOA) fluxes are saved at the end of the first time step and every 3 h thereafter (which is equal to the radiation time step) for the first day and then as daily means for days 2–5. Although 24 forecasts do not provide a large ensemble, the results were found to be very similar when only half of the forecasts are analyzed. The 24 analyses are evenly distributed through the year so that results can be compared with the HadGAM1 climatology and are distributed 6 hourly through the diurnal cycle (i.e., six forecasts starting 0000, 6000, 1200, and 1800 UTC), so that when the forecasts are combined, all regions have some sunlit points for each forecast validation time. This is necessary as ISCCP simulator diagnostics are only available at sunlit points.

b. Observational data

The cloud regimes produced from the UM are compared with ISCCP observational data (Rossow and Schiffer 1999). The ISCCP D1 product is used, which contains cloud amount in six optical depth (τ) and seven cloud-top pressure (CTP) categories on a 2.5° grid (i.e., the dataset is formed of a τ–CTP histogram for each grid point). TOA fluxes from the ISCCP FD product (Zhang et al. 2004) and the S4G product from the Earth Radiation Budget Experiment (ERBE; Barkstrom et al. 1990) are used for observations of cloud radiative forcing (CRF; e.g., Cess et al. 1990).

c. Generation of cloud regimes

The cloud regimes are obtained following Jakob and Tselioudis (2003). The KMEANS clustering algorithm (Anderberg 1973) is applied to the ISCCP observational data and ISCCP simulator output from HadGAM1 forced with observed sea surface temperatures. In both cases daily mean data for the period March 1985–February 1990 are used (as this is the period when ISCCP and ERBE overlap), and the HadGAM1 data are regridded onto the ISCCP observational grid prior to clustering. As noted by WT07, very similar clusters are obtained when using 3-hourly or daily mean data. The relative frequency of occurrence (RFO) and mean shortwave CRF (SCRF) and longwave CRF (LCRF) are calculated for each cluster. The method of WT07 is used to identify the principal cloud regimes in the tropics (20°N–20°S) and the ice-free extratropics separately. Ice- and snow-covered regions are not considered here since the observational data are considered less reliable there.

For each forecast validation time, data from the 24 simulations are pooled and both the N96 and N216 models are regridded onto the ISCCP observational grid. The forecast model data are then projected onto the clusters obtained from the HadGAM1 climatology (i.e., the data are categorized according to the HadGAM1 cluster centroid to which they are closest) and the principal cloud regimes are calculated, together with their RFO, SCRF, and LCRF. Since 3-hourly diagnostics are used from T + 0 to T + 24 h, the SCRF is normalized by the local insolation (labeled nSCRF) to enable comparison with the daily mean SCRF at later forecast times.

Each forecast validation time has also been clustered independently to produce its own centroids, and the N96 and N216 output clustered on their respective grids. On average, over several repetitions of the clustering, very similar clusters are produced using the different methods; however, due to the random initial seeding of the cluster centroids, the number of data points at each validation time is not sufficient to ensure reliable reproduction of the clusters when they are generated independently.

3. Evolution of cloud regime errors

WT07 identify five observed principal cloud regimes in the tropics and five in the ice-free extratropics (Figs. 1, 2). Each regime has been given a name based on the principal morphological cloud types that are expected to be present; however, as noted by WT07, a cloud type cannot be uniquely identified by τ, CTP, and total cloud cover alone.

The UM at N96 resolution produces a good simulation of many of the regimes. The main error in the model is the lack of midlevel cloud regimes. Investigation by WT07 and Jakob et al. (2005) of the geographical location and meteorological characteristics of the regime suggests this mainly reflects a lack of tropical and extratropical cumulus congestus cloud in the model, although there may also be contributions from a lack of midlevel cloud in decaying weather systems and a poor simulation of instances where there is thin high cloud overlaying low cloud. In addition, there is little evidence of shallow cumulus in the tropical cloud regime with a low total cloud cover [which has been associated with the observed shallow cumulus regime on the basis of its geographical location and meteorological conditions (not shown)]. The regimes obtained from the N216 model are very similar to the N96 version, but there is some tropical shallow cumulus present. Since the ISCCP dataset is formed from averages over satellite pixels in which there may be several cumulus clouds with clear-sky between, the observations will have a bias to higher cloud cover and lower optical depth. Based on an estimate of this effect by WT07, the N216 shallow cumulus simulation may be considered to compare well with ISCCP.

WT07 show that cloud feedback under climate change is sensitive to the mean present-day RFO and CRF of the regimes. The evolution of these regime characteristics through the forecast, together with the climatological values from observations and HadGAM1, is shown in Fig. 3. The regime mean RFO, nSCRF, and LCRF at T + 108 in the N96 forecasts are very similar to the HadGAM1 climatology, suggesting that only comparatively short simulations are required to analyze cloud regime biases and subsequently test improvements. This is consistent with the work of Strachan (2007) who found that systematic biases in the same GCM across the Indo-Pacific region spin up within the first few days of model integration. The main exception is the stratocumulus regime for which the nSCRF is stronger (more negative) in the N96 forecast than the HadGAM1 climatology (Fig. 3). This is due to the forecast using an aerosol climatology with fixed droplet concentrations, whereas HadGAM1 includes interactive aerosol concentrations and interactive indirect effects of sulfate aerosols on clouds (Jones et al. 1994).

The mean regime characteristics are generally similar in the N216 and N96 models, although there is a slight improvement in the higher-resolution model for several of the regimes. However, the RFO of tropical cirrus is considerably higher at the expense of deep convection in the N216 model, which compares less favorably with the ISCCP climatology. This may be a result of deep convection being triggered in more intense, localized events in the higher-resolution model; hence, less of the tropics is covered by deep convective cloud at a particular time and is replaced by thin cirrus resulting from the detrained moisture. Although the CAPE time scale differs between the two resolutions, a sensitivity experiment has indicated that this alone is not responsible for the difference in the RFO of these regimes.

Over the first few time steps of the model forecast, there is an initial adjustment of the regime characteristics for the N96 model and to a lesser extent in the N216 model, due to the change in resolution from the analyses. (During the period of study, the resolution of the operational global forecast model was changed from N216 to N320; hence, the later analyses had to be regridded for the N216 model used here, accounting for the small initial adjustment in Fig. 3.) Thereafter, there appears to be little trend in the regime characteristics over the duration of the forecast. The exception is for the tropical deep convection and cirrus regimes for which there is a slow evolution of the RFO over the first few days of the forecast. This slow change suggests a dynamical response, which reduces the amount of deep convective cloud during the forecast. Since the meteorological conditions should be well constrained by the data assimilation system, the lack of any notable trend in the cloud characteristics for the other regimes suggests that the errors in these characteristics are a result of weaknesses in the representation of local processes. This implies that, for example, the nSCRF of stratocumulus being too strong is due to problems in the cloud and/or boundary layer schemes or local resolution, rather than being due to a dynamical response to errors in other regimes.

4. Initial tendencies within cloud regimes

RP07 suggest that mean initial tendencies in the model state variables over the first few hours of a forecast are indicative of errors in the model physics, which cause the spread in climate sensitivity amongst GCMs. Since WT07 show that differences in the simulation of cloud regimes contribute to the spread in climate sensitivity, it might be expected that initial tendencies in the state variables are related to particular cloud regimes.

The method of RP07 has been followed to calculate the mean temperature tendency profile over the first 6 h of the N96 forecast for each cloud regime (Fig. 4). Over both the tropics and extratropics, the total temperature tendency profile (i.e., mean tendency for the whole region) is less than 1 K day−1 at all levels. This indicates that the model is in reasonable balance (compared to some models run by RP07). The total tendency profile most closely follows the tendency profile for shallow cumulus as this regime has the largest RFO. However, in the extratropics, there is an initial upper-tropospheric cooling and a warming in the mid- and lower troposphere. These tendencies in the total profile appear to be primarily associated with the cirrus regime. (The cirrus regime is mainly present toward lower latitudes of the extratropics, hence the high tropopause in the profile.) Examination of the regime (Fig. 2) reveals that the simulated cloud is too thick and high (or too thick for the satellite simulator to see cloud below). This is consistent with the simulated LCRF for the regime being stronger than observed (Fig. 3). The data assimilation system should ensure that the temperature profile is initially close to that observed; however, the extratropical cirrus appears to be too thick from the start of the forecast. This results in the tendency to enhance the cooling from the upper part of the cloud and reduce the cooling below (e.g., Liou 1986).

This extratropical upper-tropospheric cool and moist bias (not shown) is present as a systematic error in the HadGAM1 climatology (Martin et al. 2006). However, as the climate simulation evolves, the effect of dynamical processes means it is no longer possible to associate this bias with the cirrus regime. This example highlights the value of analyzing initial tendencies in the context of cloud regimes for identifying which areas of model physics may be responsible for the total initial tendencies and climatological systematic biases.

Over the tropics, the initial temperature tendency in most of the regimes is small; however, there is a warming in the stratocumulus regime with a maximum at 925 hPa. This is consistent with the tropical stratocumulus nSCRF being considerably stronger than observed (Fig. 3). It is likely that the longwave cooling of the cloud is near saturated, so the excessive optical thickness of the cloud, which leads to it being brighter than observed, also results in a warming tendency within the cloud. In the total tendency for the tropics, this warming in the stratocumulus regime partly offsets a small cooling tendency from the shallow cumulus regime.

5. Conclusions

This study illustrates the benefit of having a unified forecast–climate model with its own data assimilation system. It has been demonstrated that climatological errors in cloud regimes can be identified in short-range forecast simulations. This implies that addressing these regime biases will improve both NWP and climate simulations and that future model improvements targeting these errors can be tested in short forecast runs. Generally, the simulated regimes are similar at N96 and N216, although the increased resolution does improve the simulation of tropical shallow cumulus. However, the RFO of tropical deep convection and cirrus compares less well with ISCCP at N216.

Apart from a shift from tropical deep convection to cirrus during the forecast, the cloud regime errors are very similar at the beginning of the model forecast as they are in the model climatology. Since the meteorological conditions are constrained by the data assimilation system, this suggests the cloud regime errors are associated with weaknesses in the representation of local physical processes. However, the UM provides a reasonably good simulation of cloud regimes in comparison with other models (WT07), so it would be interesting to repeat the analysis with other GCMs to see whether this result is applicable generally.

Some of the total initial tendency, which leads to a systematic bias, has been shown to be associated with, and at least in part driven by, errors in the cloud regimes. In the UM, extratropical cirrus is found to be too thick, which leads to a cool, moist bias in the upper troposphere. Analysis of initial tendencies of state variables in cloud regimes is a useful extension to the work of RP07 since it provides information on which regimes are associated with the initial tendencies and hence may assist with identifying areas of the model physics which are in error. Since the variation in cloud response is believed to contribute to much of the range in climate sensitivity, the association between the initial temperature tendency and particular cloud regimes supports the method of RP07. However, WT07 show that errors in the RFO of the regimes provide a large contribution to the range of climate sensitivity and it might be difficult to deduce this particular error from the initial tendency method alone. Therefore analysis of cloud regimes and initial tendencies in state variables may be considered as complementary approaches and when combined, provide a useful evaluation tool for GCMs.

Acknowledgments

This work was funded under the U.K. Government Meteorological Research Programme. We thank Alejandro Bodas-Salcedo, William Ingram, Mark Ringer, and Mark Webb for providing comments on drafts of the paper. ISCCP and ERBE data were obtained from the NASA Langley Research Center Atmospheric Sciences Data Center.

REFERENCES

  • Anderberg, M. R., 1973: Cluster Analysis for Applications. Academic Press, 359 pp.

  • Barkstrom, B., and Coauthors, 1989: Earth Radiation Budget Experiment (ERBE) archival and April 1985 results. Bull. Amer. Meteor. Soc., 70 , 12541262.

    • Search Google Scholar
    • Export Citation
  • Cess, R. D., and Coauthors, 1990: Intercomparison and interpretation of climate feedback processes in 19 atmospheric general circulation models. J. Geophys. Res., 95 , 1660116615.

    • Search Google Scholar
    • Export Citation
  • Cullen, M. J. P., 1993: The unified forecast/climate model. Meteor. Mag., 122 , 8194.

  • Forbes, R. M., and P. A. Clarke, 2003: Sensitivity of extratropical cyclone mesoscale structure to the parametrization of ice microphysical processes. Quart. J. Roy. Meteor. Soc., 129 , 11231148.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., and G. Tselioudis, 2003: Objective identification of cloud regimes in the tropical western Pacific. Geophys. Res. Lett., 30 .2082, doi:10.1029/2003GL018367.

    • Search Google Scholar
    • Export Citation
  • Jakob, C., G. Tselioudis, and T. Hume, 2005: The radiative, cloud, and thermodynamic properties of the major tropical western Pacific cloud regimes. J. Climate, 18 , 12031215.

    • Search Google Scholar
    • Export Citation
  • Jones, A., D. L. Roberts, and A. Slingo, 1994: A climate model study of indirect radiative forcing by anthropogenic sulphate aerosols. Nature, 370 , 450453.

    • Search Google Scholar
    • Export Citation
  • Klein, S. A., and C. Jakob, 1999: Validation and sensitivities of frontal clouds simulated by the ECMWF model. Mon. Wea. Rev., 127 , 25142531.

    • Search Google Scholar
    • Export Citation
  • Klein, S. A., X. Jiang, J. Boyle, S. Malyshev, and S. Xie, 2006: Diagnosis of the summertime warm and dry bias over the U.S. Southern Great Plains in the GFDL climate model using a weather forecasting approach. Geophys. Res. Lett., 33 .L18805, doi:10.1029/2006GL027567.

    • Search Google Scholar
    • Export Citation
  • Klinker, E., and P. D. Sardeshmukh, 1992: The diagnosis of mechanical dissipation in the atmosphere from large-scale balance requirements. J. Atmos. Sci., 49 , 608627.

    • Search Google Scholar
    • Export Citation
  • Liou, K-N., 1986: Influence of cirrus clouds on weather and climate processes: A global perspective. Mon. Wea. Rev., 114 , 11671199.

  • Liou, K-N., and Q. Zheng, 1984: A numerical experiment on the interactions of radiation, clouds, and dynamic processes in a general circulation model. J. Atmos. Sci., 41 , 15131536.

    • Search Google Scholar
    • Export Citation
  • Martin, G. M., M. A. Ringer, V. D. Pope, A. Jones, C. Dearden, and T. J. Hinton, 2006: The physical properties of the atmosphere in the new Hadley Centre Global Environmental Model (HadGEM1). Part I: Model description and global climatology. J. Climate, 19 , 12741301.

    • Search Google Scholar
    • Export Citation
  • Phillips, T. J., and Coauthors, 2004: Evaluating parameterizations in general circulation models: Climate simulation meets weather prediction. Bull. Amer. Meteor. Soc., 85 , 19031915.

    • Search Google Scholar
    • Export Citation
  • Rawlins, F., S. P. Ballard, K. J. Bovis, A. M. Clayton, D. Li, G. W. Inverarity, A. C. Lorenc, and T. J. Payne, 2007: The Met Office global 4-dimensional data assimilation system. Quart. J. Roy. Meteor. Soc., 133 , 347362.

    • Search Google Scholar
    • Export Citation
  • Ringer, M. A., and Coauthors, 2006: Global mean cloud feedbacks in idealized climate change experiments. Geophys. Res. Lett., 33 .L07718, doi:10.1029/2005GL025370.

    • Search Google Scholar
    • Export Citation
  • Rodwell, M. J., and T. N. Palmer, 2007: Using numerical weather prediction to assess climate models. Quart. J. Roy. Meteor. Soc., 133 , 129146.

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

  • Senior, C. A., and J. F. B. Mitchell, 1993: Carbon dioxide and climate: The impact of cloud parameterization. J. Climate, 6 , 393418.

  • Soden, B. J., and I. M. Held, 2006: An assessment of climate feedbacks in coupled ocean–atmosphere models. J. Climate, 19 , 33543360.

    • Search Google Scholar
    • Export Citation
  • Strachan, J., 2007: Understanding and modelling the climate of the Maritime Continent. Ph.D. thesis, University of Reading.

  • WCRP, 2005: The World Climate Research Programme strategic framework 2005–2015: Coordinated Observation and Prediction of the Earth System (COPES). WCRP-123, WMO/TD 1291, 44 pp.

  • Webb, M. J., C. Senior, S. Bony, and J-J. Morcrette, 2001: Combining ERBE and ISCCP data to assess clouds in the Hadley Centre, ECMWF, and LMD atmospheric climate models. Climate Dyn., 17 , 905922.

    • Search Google Scholar
    • Export Citation
  • Webb, M. J., and Coauthors, 2006: On the contribution of local feedback mechanisms to the range of climate sensitivity in two GCM ensembles. Climate Dyn., 27 , 1738.

    • Search Google Scholar
    • Export Citation
  • Williams, K. D., and G. Tselioudis, 2007: GCM intercomparison of global cloud regimes: Present-day evaluation and climate change response. Climate Dyn., 29 , 231250.

    • Search Google Scholar
    • Export Citation
  • Zhang, Y., W. B. Rossow, A. A. Lacis, V. Oinas, and M. I. Mishchenko, 2004: Calculation of radiative fluxes from the surface to top of atmosphere based on ISCCP and other global data sets: Refinements of the radiative transfer model and input data. J. Geophys. Res., 109 .D19105, doi:10.1029/2003JD004457.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Mean CTP–τ regime histograms for the principal cloud regimes over the tropics (20°N–20°S). Shading indicates the cloud amount (%) in each CTP–τ category. (top) Observed climatology by ISCCP (note ISCCP does not observe cloud in the thinnest optical depth category, τ < 0.3); (middle) mean over the 5-day forecast from the N96 model; and (bottom) mean over the 5-day forecast from the N216 model. The model does not simulate the midlevel convection regime.

Citation: Journal of Climate 21, 4; 10.1175/2007JCLI1900.1

Fig. 2.
Fig. 2.

As in Fig. 1 but for the ice-free extratropics. The model does not simulate the midlevel cloud regime.

Citation: Journal of Climate 21, 4; 10.1175/2007JCLI1900.1

Fig. 3.
Fig. 3.

Evolution of regime characteristics through the model forecast. N96 is shown solid; N216 is shown dashed. Observed climatology from ISCCP is shown with an asterisk; from ERBE is shown with a diamond; HadGAM1 climatology is shown with a triangle. Note that the ISCCP tropical stratocumulus and deep convective RFO are identical.

Citation: Journal of Climate 21, 4; 10.1175/2007JCLI1900.1

Fig. 4.
Fig. 4.

Initial temperature tendency profiles for each regime and the total for the whole region over the first 6 h of the N96 forecast.

Citation: Journal of Climate 21, 4; 10.1175/2007JCLI1900.1

Save