SOLA, 2006, Vol. 2, 017 020, doi:10.2151/sola.2006 005
17
A Simple Parameterization Scheme for Subtropical Marine Stratocumulus Hideaki Kawai1 and Toshiro Inoue2 Japan Meteorological Agency, Tokyo, Japan 2 Meteorological Research Institute, Tsukuba, Japan 1
A simple scheme to represent the marine stratocumulus clouds off the west coast of continents is implemented in the Global Spectral Model (GSM) at Japan Meteorological Agency (JMA). The parameterization is based on diagnostic cloud schemes where cloud fraction is diagnosed mainly as a function of inversion strength considering other parameters. The global distribution of marine stratocumulus clouds off the west coast of continents is improved remarkably with this new scheme. Low-level cloud amount shows reasonable agreement with the International Satellite Cloud Climatology (ISCCP). With the improved cloud amount, the radiation fields are also improved in comparison with the Earth Radiation Budget Experiment (ERBE). Seasonal and diurnal variations of marine stratocumulus cloud amount off the west coast of continents also show reasonable agreement with surface-based cloud amount data from Klein and Hartmann (1993) and other observations.
1. Introduction Marine stratocumulus clouds are characterized by high albedo and strong cloud top cooling effect. Therefore, the radiative effect is very large in comparison with other cloud types. For example, Inoue and Ackerman (2002) studied the radiative effect of cloud type classified by the split window data of Advanced Very High Resolution Radiometer (AVHRR) on board the NOAA-9 polar orbiting satellite, using the collocated ERBE data (e.g., Harrison et al. 1990). They showed that the low-level stratocumulus clouds had a large shortwave cloud radiative forcing comparable with cumulonimbus type clouds. The ISCCP (e.g., Rossow and Schiffer 1999) analysis shows that cloud amount of the low-level clouds is very large especially off the west coast of continents. These low-level clouds, that have large radiative effect, cover a large area off the west coast of continents. This suggests that the low-level marine stratocumulus clouds affect the climate system significantly. A comparison of climate sensitivity among climate model predictions for CO2 doubling shows large variability. For instance, two models, AM2.6 from GFDL and the NCAR CAM2.0, have climate sensitivities of 4.5K and 2K (Bretherton et al. 2004a). These are opposite ends of the spectrum of predictions from climate models. The difference of low-level cloud amount distribution in these two models is significant. The AM2.6 shows a strong decrease in subtropical low cloud, leading to albedo decreases and a positive cloud feedback on climate change, while the CAM2.0 shows an increase of lowlevel cloud amount and negative feedback on climate change. At present, these contrasts in sensitivity and behavior are not explained. However, they suggest that the representation of low-level clouds in climate models is important. A realistic representation of marine stratocumulus clouds off the west coast of continents in general circulation models (GCMs) has been a major issue in climate modeling for a long time. Current GCMs still have some deficiency in representing subtropical marine stratocumulus clouds off the west coasts of continents in comparison with ISCCP analysis. About a 5K over estimation of sea surface temperature (SST) off the west coast of continents is probable in the ocean-atmosphere Corresponding author: Hideaki Kawai, Global Environment and Marine Department, Japan Meteorological Agency, Tokyo 100 8122, Japan. E-mail:
[email protected]. ©2006, the Meteorological Society of Japan.
coupled models due to poor representation of marine stratocumulus over the area (Duynkerke and Teixeira 2001). This deficiency is critical for El Niño forecast by GCMs. The reasons behind this deficiency in the representation of subtropical marine stratocumulus in GCMs are due to complicated physical interactions in and around the stratocumulus clouds and a lack of vertical resolution in GCMs. Many efforts to develop prognostic cloud schemes (e.g., Sundqvist 1988; Tiedtke 1993; Del Genio et al. 1996), where the liquid water and cloud fraction can be determined prognostically, have been made. Recently, large eddy simulation (LES) models have been used to evaluate and develop the parameterization for marine stratocumulus in GCMs. However, our knowledge is not yet sufficient to model the life cycle of marine stratocumulus clouds. Here, we introduce a simple and practical diagnostic scheme, and the representation of the marine stratocumulus clouds off the west coast of continents in current GSM at JMA is improved significantly with this new scheme.
2. Parameterization for subtropical marine stratocumulus clouds In the GSM at JMA (detailed description of GSM is available in JMA (2002)), the probability density function (PDF) cloud scheme by Sommeria and Deardorff (1977) is adopted. The so-called top hat type PDF, where the PDF is equal around the mean value of the grid with some fluctuations, is used in the GSM at JMA instead of Gaussian distribution for simplicity. The fluctuation is determined from the critical relative humidity set in advance at each level in the model. However, marine stratocumulus clouds could not be well represented in the GSM at JMA. Even if the width of the fluctuation is set unrealistically broad, little stratocumuli can be formed. Our understanding of this deficiency in the representation of marine stratocumulus, is that the model layer does not become moist enough to form clouds with the current PDF scheme in the GSM at JMA. Presumably, too dry boundary layer in the regions in GSM implies some problems in the physical parameterizations. One of the reasons is that the model layer is often thicker than the geometric thickness of marine stratocumulus, which is sometimes only 50m. The GSM at JMA has only 7 layers below 750 hPa. Therefore, it is very rarely moist enough in this PDF scheme to form the clouds at the model layer of about 300m geometric thickness. The relationship between lower atmospheric stability and persistent stratocumulus was first discussed by Slingo (1980). Klein and Hartmann (1993) studied it in detail and showed the strong relation clearly. Other observational studies also showed that the stability was essential in forming marine stratocumulus clouds (e.g., Bretherton et al. 2004b). Relative humidity (RH) has been used to form cloud in many GCMs. Inoue and Kamahori (2001) studied correspondence between ISCCP cloud types and vertical RH profiles from coincident radiosonde observations. They showed that the vertical profile of RH indicates salient features depending on cloud type. The RH is larger at the level where cloud exists. This study suggests that RH is a good indicator for cloud existence in general. However, their study does not include the marine stratocumulus. Further, Slingo (1980) showed that there was no relationship between vertically averaged RH and cloud amount of marine stratocumulus. The marine stratocumulus is very thin, so it is inherently difficult to use RH to represent the cloud as a thicker layer in the model. Teixeira and Hogan (2002) developed a scheme to form marine stratocumulus with a combination of inversion
18
Kawai and Inoue, A Simple Parameterization Scheme for Subtropical Marine Stratocumulus
strength and RH. They succeeded in increasing the cloud amount. However, the large cloud amount area was slightly away from the west coast of continents. There also have been many studies to simulate the marine stratocumulus using LES which can resolve large eddies explicitly and has a more sophisticated physical process (e.g., Stevens et al. 2005). However, the life cycle of this cloud type cannot be simulated well at present, even by these models. Therefore, in this study, we tried to find a practical way to simulate marine stratocumulus, instead of a detailed consideration of physics in and around stratocumulus. In this study, the simple and classical parameterization by Slingo (1980, 1987) was re-considered with some minor modifications from the inspection of GSM performance and observational studies. We select the following three conditions to form marine stratocumulus clouds in the GSM at JMA. > 0.07[K/hPa] (just above the layer), p < 0.01[K/hPa] (near the surface), p
the height of the layer is below 940[hPa],
(1) (2) (3)
where is potential temperature and p is pressure. The first condition relates to the strength of inversion. The strong inversion at the top of the mixed layer is essential in the existence of marine stratocumulus. The vertical gradient of potential temperature that is greater than 0.07 [K/hPa], is selected for a threshold of marine stratocumulus existence. A strong and thick stable layer near the surface often appears over land or sea ice area during the night. Over the area, the first condition is often satisfied. However, for stratocumulus formation, a supply of moisture from the surface is essential. Therefore, the stability in the very near surface layer should not be large. The second condition is set as a favorable condition for stratocumulus formation, and is set to prevent the pseudo formation of stratocumulus under a strong thick stable layer. The third condition is to prevent the false development of stratocumulus over a shallow convection area, where the inversion is strong, although the inversion height is higher than off the west coast of continents. When we form the stratocumulus as a function of stability only, then stratocumulus is formed all over the subtropical high area where a strong inversion layer exists. The main objective of this study is to represent more realistic marine stratocumulus clouds just off the west coast of continents. Observational studies indicated that the height of marine stratocumulus clouds over the ocean off the California coast was very shallow. Therefore we set the last condition, though this one is arbitrary (see section 3 last paragraph). The stratocumulus is formed in areas and layers where all three conditions above are met. Then cloud amount CL and cloud water content in cloud qcl are determined as below. CL = 12.0[(
0.07)], p
qcl = 0.03qsat
(4) (5)
where qsat represents saturation specific humidity. Cloud water is determined with a similar equation to that used by Teixeira and Hogan (2002). Further, we assume that the cloud thickness is half of the model layer and set the clouds at the upper half of the layer. The cloud thickness of marine stratocumulus is often significantly thin, for example, an observational study by Duynkerke and Teixeira (2001) showed that the cloud thickness off the coast of California was 50 200m, while the thickness of the fourth layer (940 hPa) in the model is about 215m. Therefore, we consider that this assumption is reasonable. This assumption does not change the cloud amount but changes the cloud water content in the grid qc , i.e., qc is 1 obtained by qc = 2 CL qcl . Note that water vapor is reduced by cloud water content to conserve total water. An important process concerning stratocumulus is the cloud top entrainment (CTE) process. The GSM uses a local turbulence closure model by Mellor and Yamada (1982). However, a local turbulence scheme causes a concentration of moisture leading to the excessive production of cloud at the top of the boundary layer, because there is a large difference of vertical diffusivity above and below the top of the boundary layer over
Fig. 1. Low cloud amount (below 850hPa) in units of [%] using T106 for July 2001 calculated without (top) and with (bottom) the new parameterization.
Fig. 2. ISCCP low-level cloud amount (below 680hPa) in units of [%] for July 2001.
almost all oceans. In addition, the fact that neither a shallow convection scheme nor a cloud top entrainment scheme is implemented in the GSM enhances this shortage. To overcome this problem, vertical diffusivity is vertically smoothed to keep some exchange between boundary layer and free atmosphere in the GSM at JMA. However, this treatment also works over the ocean off the west coast of continents. This causes a large reduction of cloud amount over the region. Therefore, we made an exceptional usage of this treatment. We do not apply it over regions where the above three conditions are valid. As a consequence, this treatment on vertical diffusivity works as a simple CTE-like parameterization because the diffusion through the top of the boundary layer is enhanced only in the case of weak inversion. Our parameterization scheme determines cloud amount and cloud water content diagnostically by the respective equations (4) and (5) when the above three conditions are met. This parameterization is included in the GSM as an exceptional treatment in the PDF cloud scheme, where cloud amount and cloud water content should be deduced from the PDF automatically and simultaneously.
3. Results We conducted a one month integration of the GSM for July 2001 with T106 resolution. Figure 1 shows the global distribution of low-level cloud amounts for July 2001 without (top panel) and with (bottom panel) the stratocumulus parameterization scheme described in section 2. Low-level cloud amount has remarkably and selectively increased off the west coast of California, Peru, Mauritania, and Namibia. Figure 2 shows the ISCCP low-level cloud amount from day time observation as a comparative reference, though the definitions and altitude ranges are not completely the same. The retrieved low-level cloud in ISCCP is limited to the areas where low-level
SOLA, 2006, Vol. 2, 017 020, doi:10.2151/sola.2006 005
19
Fig. 3. Difference in upward shortwave radiation at the top of the atmosphere [W/m2] between ERBE observation and model simulation without (top) and with (bottom) the parameterization. Errors are computed from the model minus ERBE. cloud exists in a single layer. The low-level cloud amount in the GSM includes the low-level cloud where upper level cloud coexists. The global distribution of low-level cloud amount, especially off the west coast of continents where low-level cloud is the dominant cloud type, shows reasonable agreement with ISCCP using this parameterization scheme. Figure 3 shows the difference of upward shortwave radiation at the top of the atmosphere between the ERBE observation data and the GSM without (top panel) and with (bottom panel) the parameterization. Serious negative biases caused by a lack of reflection due to stratocumulus in these regions are reduced by the new parameterization, as seen in Fig. 3 (bottom panel). The positive bias in downward shortwave radiation on the surface is also improved (not shown). In addition, the negative bias of downward longwave radiation on the surface is improved (not shown). The height of marine stratocumulus is low, therefore the downward longwave radiation from the cloud is not small. Because of the good representation of stratocumulus off the west coast of continents, the downward longwave radiation increases properly with the parameterization. A similar improvement of the global distribution of cloud amount and radiative field is also seen in the January case (not shown). The liquid water path increased significantly off the west coast of continents with the parameterization (not shown). There used to be almost zero liquid water path before this new parameterization, though cloud water content was automatically deduced from the PDF. With the new scheme, the liquid water paths in these stratocumulus regions become 50 100 [g/ m2 ], which is consistent with observations. For example, the liquid water path observed in July off the west coast of California is about 70 [g/m2] (Blaskovic et al. 1991; Duynkerke and Teixeira 2001). Figure 4 shows the comparisons of the seasonal variation of stratocumulus for 1992 off the coasts of California, Peru, Mauritania, and Namibia. These results are from the one month integration of the GSM for January, June, and October in 1992 with T63 resolution. The seasonal variation of stratocumulus represented by the parameterization shows reasonable agreement with the observed seasonal climatology of stratocumulus studied by Klein and Hartmann (1993). They indicated that the seasonal variation of stratocumulus corresponded to the seasonal variation of stability. The parameterization is based on the inversion strength. This causes good representation of the seasonal variation of stratocumulus off the west coast of continents. Figure 5 shows the diurnal variation of cloud amount off the west coast of California (a), Peru (b), Mauritania (c), and Namibia (d) for January (red), June (green), and October (blue). The cloud amount indicates maximum before sunrise and minimum around noon for all regions. Similar diurnal variation of stratocumulus is reported from observations and satellite
Fig. 4. Seasonal variations of stratocumulus cloud amount [%] using T63 for 1992 off California, Peru, Mauritania, and Namibia. Observation data are climatologies from Klein and Hartmann (1993). The definition of each region is the same as in their study.
Fig. 5. Diurnal variation of cloud amount [%] off the west coast of (a) California, (b) Peru, (c) Mauritania, and (d) Namibia for January (red), June (green), and October (blue). The horizontal axis is local time.
data (e.g., Blaskovic et al. 1991). The mechanisms of the diurnal variation of marine stratocumulus are complex. However, one probable explanation is that solar irradiance heats the cloud layer during daytime, leading to the reduction of cloud amount. During the night, the cloud amount increases, associated with an increase of instability due to the strong radiation cooling at the top of the cloud. Observations show that the amplitude of diurnal variation off California in July is about 25% (Duynkerke and Teixeira 2001) and off Mauritania in June is 50% (Albrecht et al. 1995). Although these values cannot be compared with model grid values or area averaged values directly because these were obtained by point observation, our results on the amplitude of diurnal variation during the summer season are consistent with these observations. In the Southern Hemisphere, the absolute value of cloud amount is larger during austral winter
20
Kawai and Inoue, A Simple Parameterization Scheme for Subtropical Marine Stratocumulus parameterization in climate model’ funded by ME of Japan.
Fig. 6. Vertical cross section of cloud cover [%] along the line between (35°N, 235°E) and (1°S, 187.5°E) without (left) and with (right) the parameterization using T106 for July 2001. The vertical axis is pressure [hPa], and the positions of grid points are shown by latitude in the horizontal axis.
than during summer, while the amplitude of diurnal variation is larger during summer than during winter due to the stronger solar irradiance. These results are consistent with Teixeira and Hogan (2002). Finally, to check the vertical structure of the stratocumulus by the new parameterization, a vertical cross section from the coast of California to the equator is shown in Fig. 6, as done in Siebesma et al. (2004). Without the parameterization (left panel) there is little low cloud cover near the coast of California, but a large amount of cloud cover is produced by the parameterization (right panel). However, the problem is that the height of the cloud layer cannot be lifted southwestward gradually, and it keeps almost the same altitude due to our cloud height limitation lower than 940 hPa. In our earlier experiment, we set the cloud height limitation higher than 940 hPa. However, excessive cloud was generated in a shallow convection area far from the coast, and little cloud was generated adjacent to the coastal line. Therefore, the altitude limitation was set to 940 hPa, considering the observational study off California (e.g., Blaskovic et al. 1991). This treatment is successful in terms of cloud cover near the coast, but has problems in representing a realistic vertical structure of stratocumulus. This indicates the need of a more sophisticated cloud parameterization, including a boundary layer turbulence scheme and a shallow convection scheme.
4. Concluding remarks We implemented a simple stratocumulus parameterization scheme in the GSM at JMA following Slingo (1980, 1987) with some minor modifications (see, Supplement 1). The scheme is based on a diagnostic method using the inversion strength in considering the meteorological conditions over the persistent stratocumulus area off the west coast of continents. The performance of the scheme is reasonably good, especially off the west coast of continents. Diurnal and seasonal variations are consistent with the observations, and the radiation budget is also improved over this region. However, our scheme forms stratocumulus only at the levels lower than 940 hPa. In the real world, this constraint is not reasonable. Therefore, we are now developing a scheme to improve this cloud level issue. The marine stratocumulus is a difficult cloud to simulate in large-scale models. Here, we introduced a basic practical scheme to simulate it. In future, we are planning to develop a scheme to represent the marine stratocumulus, by considering more sophisticated physical processes with a higher vertical resolution GSM.
Acknowledgments This study was partly supported by ‘Kyosei Project 4: Development of super high resolution global and regional climate models’ funded by MEXT, and by ‘Global Environment Research Coordnation System: Improvement of low-level cloud
Albrecht, B. A., C. S. Bretherton, D. Johnson, W. H. Scubert, and A. S. Frisch, 1995: The Atlantic stratocumulus transition experiment - ASTEX. Bull. Amer. Meteor. Soc., 76, 889 904. Blaskovic, M., R. Davies, and J. B. Snider, 1991: Diurnal variation of marine stratocumulus over San Nicolas Island during July 1987. Mon. Wea. Rev., 119, 1469 1478. Bretherton, C. S., R. Ferrari, and S. Legg, 2004a: Climate Process Teams: A new approach to improving climate models. U.S. CLIVAR Variations, 2, 1, 1 6. Bretherton, C. S., T. Uttal, C. W. Fairall, S. E. Yuter, R. A. Weller, D. Baumgardner, K. Comstock, R. Wood, and G. B. Raga, 2004b: The EPIC 2001 stratocumulus study. Bull. Amer. Meteor. Soc., 85, 967 977. Del Genio, A. D., M.-S. Yao, W. Kovari, and K. K.-W. Lo, 1996: A prognostic cloud water parameterization for global climate models. J. Climate, 9, 270 304. Duynkerke, P. G., and J. Teixeira, 2001: A comparison of the ECMWF reanalysis with FIRE I observations: Diurnal variation of marine stratocumulus. J. Climate, 14, 1466 1478. Harrison, E. F., P. Minnis, B. R. Barkstrom, V. Ramanathan, R. D. Cess, and G. G. Gibson, 1990: Seasonal variation of cloud radiative forcing derived from the Earth Radiation Budget Experiment. J. Geophys. Res., 95, 18687 18703. Inoue, T., and H. Kamahori, 2001: Statistical relationship between ISCCP cloud type and vertical relative humidity profile. J. Meteor. Soc. Japan, 79, 1243 1256. Inoue, T., and S. Ackerman, 2002: Radiative effect of various cloud types as classified by the split window technique over the eastern sub-tropical Pacific derived from collocated ERBE and AVHRR data. J. Meteor. Soc. Japan, 80, 1383 1394. JMA, 2002: Outline of the operational numerical weather prediction at the Japan Meteorological Agency (Appendix to WMO numerical weather prediction progress report). Japan Meteorological Agency, 157pp. (available online at http://www.jma. go.jp/jma/jma-eng/jma-center/nwp/outline-nwp/index.htm) Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6, 1587 1606. Mellor, G. L., and T. Yamada, 1982: Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys. Space Phys., 20, 851 875. Rossow, W. B., and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80, 2261 2287. Siebesma, A. P., C. Jakob, G. Lenderink, R. Neggers, J. Teixera, J. Calvo, A. Chlond, H. Grenier, C. Jones, M. Kohler, H. Kitagawa, P. Marquet, A. P. Lock, F. Muller, D. Olmeda, and C. Serverijns, 2004: Cloud representation in general-circulation models over the northern Pacific Ocean: A EUROCS intercomparison study. Quart. J. Roy. Meteor. Soc., 130, 3245 3267. Slingo, J. M., 1980: A cloud parameterization scheme derived from GATE data for use with a numerical model. Quart. J. Roy. Meteor. Soc., 106, 747 770. Slingo, J. M., 1987: The development and verification of a cloud prediction scheme in the ECMWF model. Quart. J. Roy. Meteor. Soc., 113, 899 927. Sommeria, G., and J. W. Deardorff, 1977: Subgrid-scale condensation in models of non-precipitating clouds. J. Atmos. Sci., 34, 345 355. Stevens, B., C.-H. Moeng, A. S. Ackerman, C. S. Bretherton, A. Chlond, S. Roode, J. Edwards, J.-C. Golaz, H. Jiang, M. Khairoutdinov, M. P. Kirkpatrick, D. C. Lewellen, A. Lock, F. Muller, D. E. Stevens, E. Whelan, and Ping Zhu, 2005: Evaluation of Large-Eddy Simulations via Observations of Nocturnal Marine Stratocumulus. Mon. Wea. Rev., 133, 1443 1462. Sundqvist, H., 1988: Parameterization of condensation and associated clouds in models for weather prediction and general circulation simulation. Physically-Based Modeling and Simulation of Climate and Climatic Change, M. E. Schlesinger, Ed., Reidel, 433 461. Teixeira, J., and T. F. Hogan, 2002: Boundary layer clouds in a global atmospheric model: Simple cloud cover parameterizations. J. Climate, 15, 1261 1276. Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea. Rev., 121, 3040 3061. Manuscript received 24 October 2005, accepted 27 December 2005 SOLA: http://www.jstage.jst.go.jp/browse/sola/
