Microphysics of Clouds with the Relaxed Arakawa–Schubert Scheme (McRAS). Part II: Implementation and Performance in GEOS II GCM

Y. C. Sud Climate and Radiation Branch, Laboratory for Atmospheres, NASA/Goddard Space Flight Center, Greenbelt, Maryland

Search for other papers by Y. C. Sud in
Current site
Google Scholar
PubMed
Close
and
G. K. Walker Climate and Radiation Branch, Laboratory for Atmospheres, NASA/Goddard Space Flight Center, Greenbelt, Maryland

Search for other papers by G. K. Walker in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

A prognostic cloud scheme named the Microphysics of Clouds with the Relaxed Arakawa–Schubert Scheme (McRAS) and the Simple Biosphere Model have been implemented in a version of the Goddard Earth Observing System (GEOS) II GCM at a 4° latitude × 5° longitude × 20 sigma-layer resolution. The McRAS GCM was integrated for 50 months. The integration was initialized with the European Centre for Medium-Range Weather Forecasts analysis of observations for 1 January 1987 and was forced with the observed sea surface temperatures and sea-ice distribution; on land, the permanent ice and vegetation properties (biomes and soils) were climatological, while the soil moisture and snow cover were prognostic. The simulation shows that the McRAS GCM yields realistic structures of in-cloud water and ice, and cloud-radiative forcing (CRF) even though the cloudiness has some discernible systematic errors. The simulated intertropical convergence zone (ITCZ) has a realistic time mean structure and seasonal cycle. The simulated CRF is sensitive to vertical distribution of cloud water, which can be affected hugely with the choice of minimum in-cloud water for the onset of autoconversion or critical cloud water amount that regulates the autoconversion itself. The generation of prognostic cloud water is accompanied by reduced global precipitation and interactive CRF. These feedbacks have a profound effect on the ITCZ. Even though somewhat weaker than observed, the McRAS GCM simulation produces robust 30–60-day oscillations in the 200-hPa velocity potential. Comparisons of CRFs and precipitation produced in a parallel simulation with the GEOS II GCM are included.

Several seasonal simulations were performed with the McRAS–GEOS II GCM for the summer (June–July–August) and winter (December–January–February) periods to determine how the simulated clouds and CRFs would be affected by (i) advection of clouds, (ii) cloud-top entrainment instability, (iii) cloud water inhomogeneity correction, and (iv) cloud production and dissipation in different cloud processes. The results show that each of these processes contributes to the simulated cloud fraction and CRF. Because inclusion of these processes helps to improve the simulated CRF, it is inferred that they would be useful to include in other cloud microphysics schemes as well.

Two ensembles of four summer (July–August–September) simulations, one each for 1987 and 1988, were produced with the earlier 17-layer GEOS I GCM with McRAS. The differences show that the model simulates realistic and statistically significant precipitation differences over India, Central America, and tropical Africa. These findings were also confirmed in the new 20-layer GEOS II GCM with McRAS in the 1987 minus 1988 differences.

* Current affiliation: General Sciences Corporation, Laurel, Maryland.

Corresponding author address: Dr. Yogesh C. Sud, Climate and Radiation Branch, Laboratory for Atmospheres, NASA/Goddard Space Flight Center, Greenbelt, MD 20771.

Email: sud@climate.gsfc.nasa.gov

Abstract

A prognostic cloud scheme named the Microphysics of Clouds with the Relaxed Arakawa–Schubert Scheme (McRAS) and the Simple Biosphere Model have been implemented in a version of the Goddard Earth Observing System (GEOS) II GCM at a 4° latitude × 5° longitude × 20 sigma-layer resolution. The McRAS GCM was integrated for 50 months. The integration was initialized with the European Centre for Medium-Range Weather Forecasts analysis of observations for 1 January 1987 and was forced with the observed sea surface temperatures and sea-ice distribution; on land, the permanent ice and vegetation properties (biomes and soils) were climatological, while the soil moisture and snow cover were prognostic. The simulation shows that the McRAS GCM yields realistic structures of in-cloud water and ice, and cloud-radiative forcing (CRF) even though the cloudiness has some discernible systematic errors. The simulated intertropical convergence zone (ITCZ) has a realistic time mean structure and seasonal cycle. The simulated CRF is sensitive to vertical distribution of cloud water, which can be affected hugely with the choice of minimum in-cloud water for the onset of autoconversion or critical cloud water amount that regulates the autoconversion itself. The generation of prognostic cloud water is accompanied by reduced global precipitation and interactive CRF. These feedbacks have a profound effect on the ITCZ. Even though somewhat weaker than observed, the McRAS GCM simulation produces robust 30–60-day oscillations in the 200-hPa velocity potential. Comparisons of CRFs and precipitation produced in a parallel simulation with the GEOS II GCM are included.

Several seasonal simulations were performed with the McRAS–GEOS II GCM for the summer (June–July–August) and winter (December–January–February) periods to determine how the simulated clouds and CRFs would be affected by (i) advection of clouds, (ii) cloud-top entrainment instability, (iii) cloud water inhomogeneity correction, and (iv) cloud production and dissipation in different cloud processes. The results show that each of these processes contributes to the simulated cloud fraction and CRF. Because inclusion of these processes helps to improve the simulated CRF, it is inferred that they would be useful to include in other cloud microphysics schemes as well.

Two ensembles of four summer (July–August–September) simulations, one each for 1987 and 1988, were produced with the earlier 17-layer GEOS I GCM with McRAS. The differences show that the model simulates realistic and statistically significant precipitation differences over India, Central America, and tropical Africa. These findings were also confirmed in the new 20-layer GEOS II GCM with McRAS in the 1987 minus 1988 differences.

* Current affiliation: General Sciences Corporation, Laurel, Maryland.

Corresponding author address: Dr. Yogesh C. Sud, Climate and Radiation Branch, Laboratory for Atmospheres, NASA/Goddard Space Flight Center, Greenbelt, MD 20771.

Email: sud@climate.gsfc.nasa.gov

1. Introduction

GCMs are used extensively for understanding the influence of clouds and cloud-radiative forcings on global climate, that is, seasonal circulation, hydrologic cycle, and precipitation and its interannual variability (e.g., Slingo and Slingo 1988; Randall et al. 1989; Miller and Del Genio 1994). The prognostic cloud-radiative forcing (CRF) in GCMs is particularly important for assessing the influence of anthropogenic activity on climate. The recent International Panel on Climate Change documents (Houghton et al. 1995, p. 201) emphasize the crucial importance of cloud processes in assessing CO2-related climate change issues. GCM modelers traditionally parameterize clouds empirically and prescribe their optical properties from observations (Del Genio et al. 1996), which has been the case for Goddard Earth Observing System (GEOS) I and GEOS II GCMs until now. Such assumptions can produce forced agreement with observations, which in turn can easily impair interpretation and understanding of the physical processes involved in assessing the influence of global change issues. For example, Del Genio et al. (1996) show how prescribed optical thickness of clouds has led to a biased understanding of the physical effects of clouds for the greenhouse gas–induced warming in GCMs. A systematically evaluated and calibrated prognostic cloud scheme is expected to dispel such problems as well as generate physically robust cloud–radiation interactions.

In Part I, Sud and Walker (1999) designed and evaluated a new prognostic scheme to represent Microphysics of clouds with Relaxed Arakawa–Schubert Scheme (McRAS) for the GEOS GCM. To put the current work into proper perspective, we review the development of fractional cloudiness scheme(s) employed in GCMs at the Goddard Laboratory for Atmospheres (GLA). Let us begin with a version of the GLA GCM that has been used in the Climate and Radiation Branch until recently (Sud and Walker 1993). The GCM employed the original Arakawa–Schubert (1974) cumulus scheme in which cumulus clouds grow exponentially while quasi equilibrium is restored at every time step. Its cumulus anvils were assumed to be conical in shape, yielding fractional cloud cover of 3 times the detrained cumulus mass fraction (Sud and Molod 1988). In the Data Assimilation Office (DAO) of the GLA, the name GEOS was coined and the first operational version of the GCM was called the GEOS I GCM (Takacs et al. 1994). It employed the Relaxed Arakawa–Schubert Scheme (RAS) of Moorthi and Suarez (1992), which used a 10-min adjustment time step with a typical relaxation timescale of 1 h. Since multiplication of 3 by 6 equals roughly 10, its cumulus cloud fraction was set to 10 times the cumulus mass fraction detrained. Supersaturation, condensation, and evaporation of falling precipitation followed Sud and Molod (1988) in both models. For cloud radiative forcing in the stratiform regime, the Slingo (1987) scheme was invoked; the scheme diagnoses cloudiness as a function of relative humidity (RH) and pressure, while the cloud optical thickness is a function of temperature, following Harshvardhan et al. (1987).

Since the GLA GCM was used exclusively for climate studies, its climate simulation was extensively analyzed. Its salient features were identified to be (i) robust 30–60 oscillations (Sperber et al. 1996; Waliser et al. 1999), (ii) successful predictions of the interannual variability of the Indian monsoon (Lau and Yang 1996), (iii) several realistic and distinct features of the simulated hydrologic cycle (Lau et al. 1996) in the Atmospheric Model Intercomparison Project (AMIP) exercise (Gates 1992), and (iv) a reasonable conservation of angular momentum (Hide et al. 1997). Regardless, the GCM had discernible deficiencies in cloud-radiative forcing (Santer et al. 1995).

With the available increase in computer power, several modeling groups have designed prognostic cloud schemes for GCMs; among the more recent ones are Tiedtke (1993), Schlesinger and Oh (1993), Fowler et al. (1996), Del Genio et al. (1996), Zhao and Carr (1997), Cheng and Arakawa (1997), and Rasch and Kristjansson (1998). At Goddard, we designed and developed McRAS with the aim of improving the moist processes, the microphysics of clouds, and the cloud–radiation interactions in the GEOS GCMs. The design and evaluation of McRAS with the physics packages of the GEOS GCMs and GATE Phase III data was discussed in Part I. In Part II, we assess the performance of McRAS in two recent versions of GEOS GCM.

Implementation of McRAS in GEOS I and GEOS II GCMs replaced all the moist processes, cloud physics, and cloud–radiation interactions in both models. The hydrologic and cloud processes of the atmosphere simulated with McRAS–GEOS II GCM (hereafter McRAS II) are evaluated vis-à-vis observations and a parallel simulation with GEOS II GCM (hereafter “control”). We also evaluate the ability of the McRAS GCMs to simulate the circulation and precipitation differences for the summers of 1987 and 1988, which were El Niño and La Niña years. The rest of the paper is organized as follows. Section 2 describes the design features of McRAS and GEOS II GCM. Section 3 describes the simulation experiments. Section 4 describes the results, delineating strengths and weaknesses of McRAS II simulation, and identifies areas of cloud microphysics that need further attention; several sensitivity studies to evaluate submodels of McRAS are included. Section 5 concludes with a summary of results and future research necessary to improve and calibrate McRAS.

2. McRAS and GEOS GCM

a. McRAS

In Part I of this paper, McRAS was comprehensively described. In its current form, McRAS can be easily implemented in any GCM. McRAS explicitly provides for three types of clouds: convective, stratiform, and boundary layer. These cloud processes occur simultaneously; therefore, the prognostic cloud water tendency equations are solved in Tiedtke [1993, Eqs. (37), (39)] style, invoking condensation, cloud generation and dissipation, and precipitation production. We briefly summarize different modules of McRAS.

The RAS of Moorthi and Suarez (1992) is the convective scheme of the GEOS II GCM and is retained in McRAS. In our implementation, however, the convective cloud base is diagnosed to be the top of the nearest layer from the bottom (the search is limited to the four nearest levels above the surface layer) where the RH exceeds 0.9 times the critical value of relative humidity, RHcrit, for stratiform condensation. The energy necessary to carry the convective mass flux (with precipitation loading and momentum dissipation) to its detraining level must be provided by the thermal buoyancy generated by moist convection. Critical cloud work function, a concept introduced into moist convection by Lord et al. (1982), is invoked to enforce the limit implicitly. New stratiform clouds form when the average RH in a layer exceeds the threshold RH necessary to maintain the existing cloud fraction. This condition causes cloud fraction and in-cloud condensation to increase. However, if the in-cloud RH becomes subsaturated, the cloud water evaporates to maintain a 100% in-cloud RH. When not enough cloud water is present, all the cloud water evaporates and the cloud vanishes instantly. The boundary layer (BL) clouds are produced when the BL convection (which almost always commences as dry convection) enables BL eddies to become supersaturated at or before the detrainment level. The cloudy air is deposited at detrainment, which may produce higher water vapor content than the air above or below it—a typical configuration of countergradient fluxes.

Conversion of condensate into precipitation follows the Sundqvist (1988) algorithm. There is no special treatment for ice-phase hydrometeors as in Sundqvist (1993) and Fowler et al. (1996). Full cloud microphysics remains active at all times, affecting all cloud condensate, including that in the cumulus towers and anvils. Clouds in McRAS convect, diffuse, and advect both horizontally and vertically. However, cumulus-tower debris also leads to grid-scale humidification and cloudiness. Both of these are new features of our convective microphysics. The cloud destruction mechanisms are the same for all clouds. They include (i) diffusion of dry air into the cloud at subgrid scale; (ii) evaporation of in-cloud water through convective-scale subsidence; (iii) cloud-top entrainment instability (CTEI) among adjacent cloudy and clear layers (Del Genio et al. 1996);and (iv) cloud mergers, including entrainment of ambient clouds into convective towers and downdrafts.

The convective downdrafts follow Sud and Walker (1993). The precipitation falling inside cumulus towers (assumed saturated) does not evaporate. However, the anvil precipitation, which falls through the unsaturated environment, undergoes evaporation and could produce downdrafts if the free-fall criteria were satisfied. Downdrafts entrain cloud water, which evaporates first because of drying by subsidence in the downdraft. In addition, the tower precipitation emerging beneath the cloud base will often satisfy the free-fall criteria and induce downdrafts. The statistical distribution of hydrometeors in an idealized cloud geometry follows Del Genio et al. (1996) for water clouds and Ou and Liou (1995) for ice clouds. For addressing the cloud-radiative influence of cloud water inhomogeneity, we invoke the Cahalan (1994) correction. For more details, refer to Sud and Walker (1999) and Part I.

b. GEOS GCM

We first used a 17-layer version of GEOS I GCM (Takacs et al. 1994) and then switched to the new GEOS II GCM with 20-layer in the vertical. The additional three layers appear in the stratosphere and lead to some reorganization of the layers above the 400-hPa level. Both versions of the GEOS GCM were used at a 4° latitude × 5° longitude horizontal resolution. The current GEOS II GCM replaces both the GEOS I and the GLA GCMs. It is described in algorithm theoretical basis document (Rood 1999). The main differences between GEOS I and GEOS II GCMs are in the new features, such as (i) ability to perform coordinate translation and rotation with a proviso for relocating the mathematical poles to any arbitrary location (not used in this investigation); (ii) inclusion of a gravity wave drag parameterization due to Zhou et al. (1996); and (iii) modification in the diagnosed fractional cloudiness and cloud optical properties. The GEOS II GCM diagnoses clouds and their optical properties as follows.

The convective cloud fraction (FRAS) is diagnosed as a ratio of the convective condensation (lRAS) produced within the time step divided by its prescribed critical value (lc = 1.25 g kg−1), giving
i1520-0469-56-18-3221-e1
The newly created FRAS is assimilated into the already existing Fn−1RAS, with a convective dissipation timescale, τ ≡ 1 h (assumed). This leads to the following equation:
i1520-0469-56-18-3221-e2
An “iffy” (implemented with FORTRAN “IF”) feature of the scheme is to eliminate RAS clouds as soon as they fall below the 5% limit. The stratiform clouds follow Slingo and Ritter’s (1985) empirical formulas. They are diagnosed from the average RH of the grid cell in the following way
i1520-0469-56-18-3221-e3
In (3), RHcrit is determined by
critss33sr,
where
i1520-0469-56-18-3221-e4b
and
sppsurfp
In the current setting, RHmin = 0.75 and α = 0.573 285. The second iffy feature of the scheme is to eliminate large-scale clouds in regions where the cloud sublayers are convectively unstable. Finally, the cloud fraction is taken to be the larger of the two, FRAS and FLS. For estimating the optical thickness, the following empirical relations are used for ice and water in the clouds:
−4τice−1−3l−1
and
−2τwater−1−1l−1
The ice fraction is diagnosed as a linear function of temperature: it is zero at 253.15 K and grows to unity at 233.15 K. When both ice and water clouds coexist, the optical thickness of the mixture is the sum of the mass fraction–weighted optical thickness of both cloud species. The optical thickness of the convective clouds is assumed to be 0.16 hPa−1 irrespective of the cloud height, cloud water amount, or the ice–water ratio. When both convective and large-scale clouds are present in a grid cell, the optical thickness of each cloud type is linearly weighted by the respective cloud fractions.

Our additions to the GEOS I and GEOS II GCMs, besides McRAS, include implementation of the simplified Simple Biosphere Model (SS:B) of Xue et al. (1991) for land–atmosphere interaction both with and without the new snow model of Sud and Mocko (1999, manuscript submitted to J. Meteor. Soc. Japan). The boundary layer scheme for turbulent transport is by Helfand and Lebraga (1988). The radiation package of McRAS is due to Chou and Suarez (1994) with a provision for handling prognostic clouds and in-cloud water and ice fractions (Chou et al. 1998, 1999). It is not too different from that of the GEOS II GCM, except for a revised calculation for the optical thickness of clouds for shortwave and longwave radiation.

3. Simulation experiments

We performed two 4-yr integrations with the GEOS II GCM, one with McRAS in GEOS II GCM (McRAS II) and one without McRAS (control). Each integration was started from analysis of observations for 1 January 1987 and was forced with the prescribed sea surface temperatures available in the form of weekly averages from National Centers for Environmental Prediction analysis, observed sea ice, and climatological permanent ice available in the AMIP datasets. In addition, McRAS–GEOS I GCM (hereafter, McRAS I) simulations were used previously to study the influence of SST anomalies on the circulation and rainfall simulated for the summers of 1987 and 1988. A comparison of new McRAS II results for 1987 minus 1988 with the old McRAS I was also performed.

Subsequently, we performed several pairs of additional December–January–February (DJF) and June–July–August (JJA) integrations to determine the sensitivity of McRAS II simulations to different cloud dissipation processes of McRAS. All such integrations were initialized from the data generated in McRAS II and were forced by the prescribed boundary forcings of McRAS II for the chosen period. The simulation experiments are summarized in Table 1. We only emphasize those differences that have discernible impact on the simulated climatology; surprisingly, almost all the new results can be explained on rational physical basis. We would eventually like to calibrate the scheme further in single-column model (SCM) evaluations with the goal of eliminating some of the remaining deficiencies identified in the current investigations.

4. Results

a. Cloud–radiation interaction

The McRAS II–simulated planetary albedo and outgoing longwave radiation (OLR) for the winter (DJF) and summer (JJA) seasons are in reasonable agreement with the Earth Radiation Budget Experiment (ERBE) (Barkstrom 1984) data (Fig. 1a). The most notable systematic error is larger than observed tropical planetary albedo. In comparison, the control with the original GEOS II GCM is also quite reasonable. Only for midlatitude summers is the simulated albedo less than that of the ERBE data. To better delineate the behavior of the variation of annual cycle of planetary albedo, we examine the progression of monthly zonal average albedo errors: McRAS II simulation minus ERBE data (Fig. 1b). We note bands of higher planetary albedo in most places (including the Tropics) except for high-latitude summers. As we shall see later, these errors can be reduced by tuning lcrit (the critical cloud water value that influences the rate of precipitation generation) in the parameterization of autoconversion. We will see in section 4d that the simulated in-cloud water in stratiform clouds is too high, which is consistent with excessive planetary albedo at those latitudes. We could easily play a number of games with optical properties of clouds or other parameters of McRAS, but for now we will postpone the calibration of the prognostic cloud scheme until the availability of better observational data and future SCM evaluations.

Cloud radiative forcing (CRF) is a key modulator of the global circulation (Kiehl and Ramanathan 1990). The simulated OLR errors vis-à-vis ERBE data, though much smaller, are consistent with cloudiness errors (Fig. 1c). The simulated OLR is lower everywhere, except for midlatitude summers when cloudiness is predominantly convective. This behavior corresponds well with the planetary albedo errors. Indeed, errors in cloud height will introduce another dimension to the response of OLR and planetary albedo to cloudiness because high (low) clouds affect OLR strongly (weakly), while the planetary albedo is largely determined by the optical path irrespective of the cloud height, except for the differences in the optical effects of cloud water and ice. From these considerations alone, one could argue that the OLR errors are sensitive to the high-level cloudiness and cloud water errors. All it takes is a few percent error in high clouds or the in-cloud water content to produce large systematic longwave (LW) CRF errors. We also show the longwave and shortwave CRFs for the winter and summer periods and compare them with the ERBE data (Fig. 1d). Again, the errors are consistent with the diagnostics shown earlier. The simulation yields global cloud-radiative forcing of the longwave (shortwave) radiation of 33 (−59) W m−2, while the corresponding ERBE values are 28 (−47) W m−2. This clearly suggests too much cloudiness and CRF in McRAS II. Overall, the results compare well with Fowler and Randall (1996a).

b. Precipitation simulation

We examine the structure of the zonal averaged precipitation simulated by McRAS II vis-à-vis Huffman et al. (1997) data. The Huffman et al. dataset begins in July 1987 (with the launch of the DMSP-F8 satellite) and is continuous for the entire simulation period except for December 1987 (missing data). In comparison, the model simulates reasonable zonal mean precipitation (Fig. 2a). Its intertropical convergence zone (ITCZ) migrates into the Northern and Southern Hemispheres in the boreal summer and winter seasons in agreement with the Huffman et al. (1997) data. The mean of the entire time series gives the zonal structure of the precipitation. Since these patterns were similar year after year (not shown), we can safely infer that the annual average differences between the simulations and observations are systematic and contain very little variability. However, the errors in precipitation retrievals over oceans may be nonnegligible, as evidenced in a comparison of the Huffman et al. (1997) and Spencer (1993) precipitation datasets. Consequently, the entire discrepancy between the simulation- and satellite-based rainfall data may not be interpreted as simulation errors. The 70°S–70°N precipitation in McRAS II (control) simulations for the 4-yr period is 2.90 (3.08) mm day−1. These differences are systematically maintained in both the evaporation as well as precipitation simulations. The corresponding value in the Huffman et al. (1997) data is 2.74 mm day−1.

We show seasonal plots for January–February–March (JFM) and July–August–September (JAS) in Figs. 2b and 2c. The odd periods are chosen to avoid missing/unavailable data for the months of June and December of 1987. The distribution of simulated precipitation has the characteristic spatial structure with an ITCZ, a South Pacific convergence zone (SPCZ), and clearly discernible convective precipitation over tropical land regions that receive most of the sunshine and yield copious rainfall over the tropical warm pool regions. The model also simulates winter storm tracks over the Gulf Stream, the Kuroshio, and South Equatorial Currents along the eastern boundaries of North America, Asia, and South America (Brazil), respectively. However, the precipitation over the tropical Pacific (the ITCZ region) is somewhat less than the observed (Huffman et al. 1997), while it is too high over the rest of the regions. We compared the rainfall time series of Fig. 2a at every longitude and noted that it is larger and more diffused in the midlatitudes. These are puzzling systematic errors that were present even in the control (see Fig. 2a for correspondence of McRAS II and control precipitation). Since McRAS was specifically aimed at introducing prognostic clouds and cloud water into the model, we primarily examined simulated clouds and CRF; some of the deficiencies are a by-product of other problems in the model and cannot be resolved in the current paper. Among the desirable features of McRAS II are a single ITCZ as well as reasonable precipitation over the continents. However, over Central America, McRAS II precipitation is too high and worse than the control (comparison not shown). The simulations with no cloud advection revealed that its inception is responsible for the precipitation error noted above (not shown). Overall, McRAS has reduced both the evaporation and precipitation by roughly 0.2 mm day−1; this finding is in agreement with Fowler et al. (1996) who also show reduced precipitation in prognostic cloud water simulations.

c. Cloudiness

One of the primary goals of a prognostic cloud scheme is to simulate more accurate CRF. Our global cloudiness is 60.1% as compared to 57.4% (sum of high-, low-, and middle-level cloudiness) and 62.0% as the overall cloudiness (produced with nighttime corrections) in the International Satellite Cloud Climatology Project (ISCCP) C2 data (Rossow and Schiffer 1991). Figure 3a shows the zonal picture of 4-yr average monthly climatology of cloudiness. McRAS produces more (less) than observed high-level (middle-level) clouds. Comparatively, the overall cloudiness is in better agreement with observations. Recognizing that thin cirrus clouds are not only difficult to detect but may have been assigned lower heights particularly when they appear above the midlevel clouds, which generally have higher optical thickness (Rossow 1998, personal communication), the differences may not be due to model deficiencies alone. The zonal average total cloudiness shows several conspicuous deficiencies. First, the simulation has less meridional as well as annual cycle structure. The simulated cloudiness for the DJF and JJA periods is shown in Figs. 3b and 3c and is compared with two observational datasets [ISCCP and Special Sensor Microwave Imager (SSM/I)]. The simulated cloudiness is less than observed in the Tropics and higher than observed in the high latitudes. The model shows better spatial structures over land and poorer spatial structures over the oceans as compared to the ISCCP data. It also shows poor definition of the ITCZ, subtropical highs, SPCZ, and other regions of large cloudiness variations over the oceans. In the convective regions, the model simulates less cloudiness (e.g., in the ITCZ over the tropical Pacific), while the subsidence regions surrounding the ITCZ are not as clearly distinguished as in observations. The model also simulates about 10% higher cloudiness over the warm pool region. However, large differences of very similar magnitudes are noted between the two cloud retrieval algorithms (ISCCP versus SSM/I). Hence, it is not clear how much of the error is due to deficiencies in the McRAS II simulation.

d. Cloud water

The separation of cloud condensate into cloud water and cloud ice is important for the cloud–radiation interaction. The cloud water–ice distinction in McRAS is diagnosed as a function of temperature [cf. Del Genio et al. 1996, Eq. (12)]. The heat of melting for conversion of cloud ice into cloud water is ignored in the current version. We produced the zonal distribution of cloud water and ice (not shown). McRAS cloud ice was smaller than that of Fowler et al. (1996), but it is in better agreement with the Rasch and Kristjansson (1998) validation data, which examine cloud water as a function of temperature (also see section 4h). There were two available datasets for assessing cloud water distribution for the cloud water path using SSM/I observations. One is by Weng and Grody (1994) and the other is by Greenwald et al. (1993). Nevertheless, there are huge differences between these datasets: for example, Greenwald et al.’s estimates are larger than those of Weng and Grody by about a factor of 2 even though they both use the same SSM/I data. The climatology of these cloud water fields is available on the Web pages of the authors. For the current evaluation, we again use the more recent SSM/I dataset of Ferraro et al. (1996) and ISCCP data (Rossow and Schiffer 1991). The distribution of zonal average cloud water in all clouds is shown in Fig. 4a. The simulated and satellite-derived total in-cloud water fields for DJF and JJA are shown in Figs. 4b and 4c. The simulation yields a reasonable structure of cloud water in the Tropics and midlatitude, but its magnitude is somewhat high, which is consistent with higher planetary albedo, and stronger than observed cloud-radiative forcing discussed in section 4a. Generally, the spatial structure of in-cloud water is analogous to the precipitation (Figs. 2b, c) and that is to be expected. However, this also implies that the precipitation errors reflect into cloud water errors, which is indeed so. For example, weaker-than-observed ITCZ simulated for the JJA period also produces low cloud water paths. Large-scale clouds are generally associated with cloud water (ice) at low (high) latitudes and those fields correspond well with large-scale precipitation.

e. Madden–Julian oscillations (MJOs)

In the AMIP evaluations, it was found that a large number of models were unable to show signatures of MJOs in the 200-hPa tropical velocity potential fields even after the data were filtered to show 20–100-day oscillations. The GLA GCM produced fairly decent MJOs (Sperber et al. 1996; Waliser et al. 1999), but the primary cause of the success of some (and not other) GCMs has been a mystery. The GEOS I and now GEOS II GCMs with the current implementation of McRAS using the same philosophy as in Sud and Walker (1993) both simulate discernible but weaker-than-observed [based on European Centre for Medium-Range Weather Forecasts (ECMWF) analysis] MJOs (Fig. 5a). The control simulation with GEOS I GCM had a much poorer simulation of MJOs (not shown) but the new model, GEOS II, yields equally reasonable MJOs (Fig. 5b). DAO analysis (not shown) also produces MJOs that are very similar to those of ECMWF analysis (Fig. 5c). In the current implementation of moist convection, we continue to invoke the minimum RH criteria for the onset of moist convection and use both the critical cloud work function and the “λ-bounds criteria” for convective entrainment to eliminate spurious and/or unrealistic clouds (Sud et al. 1991). Presumably the answer for the success of MJOs lies in one of those effects. Our sensitivity simulations with McRAS I simulations had shown that disabling the 90% RH criteria for the onset of moist convection substantially weakened the MJO, which concurs with the findings of Wang and Schlesinger (1999). Nevertheless, these are only preliminary results; a full investigation of the influence of different ways of implementing convection on MJOs will be the subject of a separate paper. We submit that, if the relaxation time of a convective scheme is adjusted to maintain strong convective available potential energy (CAPE) in the Tropics, it can mimic the 90% RH trigger without invoking it explicitly. Alternatively, the improved CRF of the GEOS II GCM, introduced by a new relaxation timescale for convective clouds, may be an important element of physical forcings that capture MJOs. Nevertheless, since both models have weaker-than-observed MJOs, we believe that improving CRF and designing a reasonable trigger, based on CAPE, for the onset of moist convection (as discussed in Emanuel et al. 1994) is likely to improve the vigor of MJOs.

f. Cloud production and dissipation

The mechanisms for cloud production and dissipation are quite different in McRAS as compared to the earlier and new diagnostic cloud schemes of GEOS I and GEOS II GCMs. Figure 6a shows the simulated cloud water production and dissipation tendencies, and Fig. 6b shows the corresponding cloudiness production and dissipation tendencies. Both datasets had to be regenerated by including the cloud production and destruction diagnostics into McRAS II. Therefore, the data have only one simulated summer: June–July–August. The lhs panels in these figures shows tendencies of cloud water and cloud fraction production by (a) stratiform processes, (b) moist convection, (c) boundary layer clouds, and (d) sum of all the cloud processes. The corresponding rhs panels show the cloud water and cloud fraction dissipation by (e) cloud munching (subgrid-scale diffusion);(f) cloud top entrainment instability, which is quite vigorous; (g) the influence of advection, subsidence, and downdrafts shown as a sum; and (h) sum of all cloud dissipation processes. The negative production of large-scale cloud water, which is evidenced in the stratiform cloud water production, is due to evaporation of the cloud water that advects into the region, as can be seen in Fig. 6a(g) as a positive dissipation. A part of the advected cloud water evaporates even when there may be no change in cloudiness because that occurs only if all of the cloud water evaporates. Therefore, we do not see the signatures of cloud production and removal in the corresponding cloudiness fields shown in Fig. 6b. Another interesting feature is the negative cloud dissipation near the surface [Fig. 6b(g)]. This is caused by convective-scale subsidence of cloudy air mass aloft. It appears in both figures (Figs. 6a, b), suggesting that clouds and cloud water are consistently eliminated by subsidence (further details of the plots are not shown). The diagnosed convective cloud generation is quite typical of convective condensation in the Arakawa–Schubert (1974) scheme [Figs. 6a(b), 6b(b)]. Evidently, the production rate of boundary layer cloud water is nonnegligible [Fig. 6a(c)], but the corresponding cloud production rate is substantially lower [Fig. 6b(c)]. Cloud microphysics in the convective tower has been noted for scavenging excessive amounts of cloud water and making the detraining cumulus anvils relatively dry (based on examination of in-cloud ice data); this may be the result of an assumption of vertically oriented convective towers through which most of the convectively generated precipitation recycles before falling through. On the other hand, if we invoke a parameterization that allowed some precipitation to fall outside of towers, as in Cheng and Arakawa (1997), this deficiency is likely to be reduced. Figure 6b(a) shows negative production of large-scale clouds in the midlatitudes. This happens by evaporation of cloud water in cloud advection associated with large-scale subsidence. There is some evidence of cloud removal by “munching.” Correspondingly, Fig. 6b(g) shows cloud dissipation except for small cloud production by subsidence near the surface; this is consistent with cloud water production in Fig. 6a(g). We find cloud production by large-scale processes up to the 100-hPa level in the winter hemisphere. It is caused by the deep rising motions through theexcessively cold upper levels (a model deficiency). The deficiency was present in the GEOS II GCM and belongs to a typical problem of cold-poles in GCMs. Indeed, it needs resolution, but we can safely argue that it is not related to the parameterization of cloud microphysics or CRF. (Note: we recently discovered that use of 65% prescribed surface albedo of ice over Antarctica in SSiB vs a typical value of 80% in some other datasets contributes to the warm pole problem in Austral summers.) The rest of the panels are self-evident and show the expected behaviors.

It is not obvious whether or not our scheme produces a satisfactory moistening and CRF at various levels of the atmosphere, but it can be easily improved by modifying the functional lcrit in relation to the atmospheric temperature. The good features of the scheme are realistic cloud water and cloud fraction tendencies, significant influence of CTEI [adapted from Del Genio et al. (1996)], and cloud munching [adapted from Tiedtke (1993)]. The system conserves cloud water because cloud water production and dissipation tendencies balance [Figs. 6a(d), 6b(d) vis-à-vis Figs. 6a(h), 6b(h)]; however, cloudiness is not conserved, particularly near the surface, due to the enforcement of upper limits on cloudiness (e.g., cloud fraction cannot exceed unity). To get this to balance, the cloudiness diagnostics need to be reworked, but since cloudiness is not a conservative field, the simulation has no errors. McRAS II–simulated cloud tendencies cannot be compared with the empirical scheme used in the GEOS II GCM that was used for the control case because there is no cloud munching or CTEI (except by suppressing stratiform clouds emerging in the path of convection). Cloud dissipation by subsidence, downdraft, and entrainment of cloud water is folded into cloudiness as a function of RH.

g. Sensitivity to CTEI, cloud advection, and Cahalan correction

Following Fowler and Randall (1996b), who comprehensively investigated the influence of modeling assumptions on the performance of the Colorado State University GCM, we examine the influence of the key modules of McRAS on its performance in the GEOS II GCM. The CTEI and cloud advection (CA) are well known (but not well understood) features of the real atmosphere; however, when invoked in a cloud scheme of a GCM, they become controversial entities. Del Genio et al. (1996) comprehensively justified the need for parameterizing CTEI; they used observations to isolate the criteria for the inception and vigor of CTEI; the timescales used for CTEI adjustment are also based on observations of the dissipation of marine stratocumulus clouds. On the other hand, CA has been ignored in almost all the GCMs. Even a few modelers who included CA (e.g., Fowler et al. 1996) conclude that its influence is not too significant. Such a finding is well justified because the timescale of CA (on the order of a few days) is much larger than the timescale for cloud dissipation (ranging from several minutes to an hour). Let us first show our results and then discuss how and why we may get a different result.

In concurrence with the results in section 4f, Fig. 7a shows an increase in cloudiness caused by removal of CTEI (i.e., no-CTEI simulation minus McRAS II); the differences naturally emerge in regions where CTEI is most vigorous, that is, in the Tropics. Some middle- and high-latitude differences for no-CTEI simulations are also produced, but these are related to circulation changes because there are no CTEI tendencies at those locations (Fig. 6b). Thus, the CTEI parameterization makes a difference in the simulated marine stratus clouds, which in the control simulation are eliminated by an empirical criteria (see section 2). Figure 7a also shows the influence of no CA (thin contours); it shows no-CA minus McRAS II differences. The no-CA simulation has much higher cloudiness upper levels of the middle and high latitudes. Within the rising branch of the Ferrel cell, particularly in the winter hemisphere, clouds with very small cloud water do form, but CA provides an escape path for them; in addition, they will naturally evaporate during subsidence episodes associated with surface high pressure and clear days, which happen quite frequently in winter. In this way, CA reduces middle and high-latitude clouds. Since the dissipation timescale of convective clouds is small, CA has a much smaller, nonetheless quite similar, influence on the convective clouds (reduced by about 5%, which is within the observational error margin). Quite surprisingly there are no regions of cloudiness increase due to CA. This suggests that clouds advect much more slowly, as expected, but dissipate rather quickly after advection. This is particularly true of sinking motion episodes with embedded clouds. Indeed, neglect of CA is better justified in models in which clouds are inferred diagnostically from RH or an implicit influence thereof. In such models, advection of specific humidity and temperatures fields implicitly advects clouds. However, when the premise of a parameterization of stratiform clouds is based on the rate of production of clouds and cloud water from the rate of change of RH, the cloud advection process naturally assumes a central role, as in McRAS. Without such a design, we will need to make the deep convective clouds follow the RH like their large-scale counterparts, while in the dry environment they must dissipate quickly, thereby loosing the benefit of a physically based life cycle of clouds that relies on the cloud dissipation by subgrid-scale diffusion and incipient evaporation in subsidence.

Figure 7b clearly reveals that without CTEI or CA we will generate much larger errors in the cloud-radiative forcing of both shortwave and longwave radiation(s). As can be expected, tropical CRF errors increase in the absence of CTEI, while middle- and high-latitude errors increase in the absence of CA. Since ERBE climatologies of CRF are not too reliable in the polar regions, we largely focus on the region from 70°S to 70°N even though the GCM simulates everything from pole to pole. We specifically note that the absence of CA increases both shortwave and longwave CRFs in both seasons. Therefore, in the context of McRAS, both CTEI and CA are useful because they enable the model to perform more realistically. Alternatively, without CA, the high cloudiness over midlatitudes and Tropics becomes more unrealistic. This was also seen in the GATE evaluation (Fig. 3c) (Sud and Walker 1999; Part I). The Cahalan (1994) correction (CC) for cloud water inhomogeneity is a part of our design. In Fig. 7b we show that lack of the CC would make shortwave (SW) CRF worse, yet it does not degrade LW CRF. A simple rationale for this is as follows. The longwave is strongly affected by high clouds that, in our moist-convection scheme, excessively scavenges the convective towers, thereby removing a lot of cloud water. Removal of inhomogeneity correction cancels for that error; thereby it produces the proverbial right answer for the wrong reason. However, such an answer cannot survive the scrutiny of SW CRF; therefore, we still infer that CC is needed, but together with other parts of McRAS, it also requires some attention.

h. Influence of limiting in-cloud water for autoconversion

As we shall see, the choice of lmin makes a significant difference to the cloud water distribution in the atmosphere. In the current model, lmin is set by the following relation:
i1520-0469-56-18-3221-e6
However, Sundqvist (1988) did not invoke such a minimum; without any limit, there will be a continuous precipitation in the form of a drizzle from all clouds, which indeed may or may not evaporate in fall (depends upon the numerical scheme of the precipitation–evaporation package, which may have an exponential extinction). We felt the need for such a limit, which was initially based on an estimate of the average physical radius of the cloud water particles. In the implementation of McRAS in GEOS I GCM, we used a fixed value of lmin = 100 × 10−6, a value derived from 5–6-μm radius of cloud particles and Del Genio et al.’s (1996) estimates for number density of cloud particles. In a separate simulation, we used lmin = 0.5 × lcrit as a sensitivity test and found a large sensitivity to the choice of lmin. We also ran another parallel simulation without lmin, following Sundqvist (1988). These simulations produced vastly different in-cloud water amounts and CRFs; we first distinguish them by cloud water distribution binned as a function of temperature (Fig. 8a), following Rasch and Kristjansson (1998). The functional form of lmin in Eq. (6) was adopted in McRAS; its cloud water distribution is very similar to the original Sundqvist (1988) result simulated with no lmin; however, the small limit leaves a little more cloud water for convective anvils, which help to reproduce the observed OLR/LW CRF structure in the tropical regions (Fig. 1a). The value of lcrit has roughly the same shape as that of in-cloud water from Sundqvist (1988), which is to be expected. Hence, we conclude that if we have a good estimate of the observed in-cloud water as a function of temperature, it will be easy to readjust the magnitude of lcrit function to reproduce the form of the observed cloud water behavior in clouds. However, because there is such a huge variability in the cloud water observations, while the simulated data is well within the range of such a variability, we continue with the Sundqvist’s (1988) formulation. Since lmin at cold temperatures influences the cloud water content of the deep convective anvils, a small difference in the choice of lmin yields a systematic biases in temperature, the cloud water content of high-clouds as well as OLR in the Tropics.

Figure 8b shows all sky zonal cloud water in the column atmosphere. Evidently, it can be affected by the choice of lmin. However, even lmin = 0 produces significantly more clouds as compared to observations. Therefore, the problem may be related to lcrit or choice of RHcrit for stratiform clouds. On the other hand, McRAS II is cooler in the lower troposphere as compared to ECMWF analysis (Fig. 8c). A higher limit on minimum cloud water does not ameliorate these errors because there is enough water in the warm tropical clouds, but at the higher levels, a high cloud water content can produce warmer temperatures beneath the clouds. Since the control simulation also produces systematic errors that are similar to those of McRAS II, we can safely conclude that the errors are real and are emanating from other aspects of model physics. Indeed, downdrafts cool the tropical atmosphere, but the control simulation, which is similarly cooler, does not have downdrafts; therefore, the temperature error may be related to a convective–radiative equilibrium that constrains the vertical profile of temperature to have a certain lapse rate. Since a number of factors can affect the outcome, this problem needs a thorough systematic investigation. It must be pointed out, however, that a cold bias of the tropical troposphere has been a persistent problem with several models.

i. Influence of C0 and lcrit

We reasoned about the logic of the distinctions introduced into C0 and lcrit values for convective and large-scale clouds by Sundqvist (1988). If one asks the question, why should lcrit be different for convective and large-scale clouds, there is no obvious answer. In our view, the only physical difference between the convective and large-scale clouds is due to cloud water inhomogeneity; nevertheless, this should make the lcrit for convective clouds become smaller (and not larger) because they are more inhomogeneous. In an earlier implementation, we included McRAS in the GEOS I GCM with the same lcrit and autoconversion time constant C0 for all clouds. In that implementation, we allowed the convective clouds to become stratiform in one time step;that is, they had a life cycle of only 10 min. However, subsequently, when we discovered that the simulated condensation heating for boreal winter stayed much too close to the equator, we thought it to be the consequence of erroneous CRF. Indeed, we also noted that higher values of lcrit for stratiform clouds (lcrit = 1000.0 × 10−3 g kg−1 as opposed to lcrit = 333.33 × 10−3 g kg−1; see Part 1, Table I) increased the in-cloud water in stratiform regions (even beyond what we have currently), producing a worse CRF. In our attempt to resolve this problem (which we now know was only limited to the winter of 1987, i.e., the first year of integration as opposed to all the years; Fig. 2a) we went back to using the original constants of Sundqvist (1988) because we thought it may have something to do with changing the Sundqvist constants. However, first we set C0 to be 10−3 s for all clouds, which was also done by Tiedtke (1993). Setting lcrit for the convective clouds to a smaller value than the one for stratiform clouds reduced the cloud water in highly convective regions, thereby flattening out the OLR structure. Overall, the CRF fields were not discernibly better. We therefore continue to use Sundqvist’s (1988) original constants. However, the potential for significant adjustments in the cloud water climatology is there, but we are deferring them at the present time.

Figures 9a and 9b show the influence of lcrit on in-cloud water and CRF. It shows a minuscule effect of making lcrit = 0.333 g kg−1 for all clouds, but there was a significant error in the tropical CRF when lcrit was changed to 1.0 g kg−1 for all clouds. This shows that using the smaller value of lcrit for all clouds would not hurt provided we have a good estimate for lmin, which agrees with our premise that all clouds should be treated the same way because they must follow the same laws of physics.

j. Tropical response of the new model

The summer of 1987 produced the worst recorded drought over India, as evidenced in the precipitation data of the last 100 years, while the summer of 1988 had greater-than-normal precipitation. These differences have been related to El Niño and La Niña episodes of the respective years. Similar precipitation differences were noted over the Sahel and tropical Central America, even though the precipitation reduction was the highest over India. These were well simulated in both GCMs with McRAS and SiB—the GEOS I and the GEOS II—in a 2-yr 1 January 1987–1 January 1989) integration. However, following the above success, two sets of three additional cases, one each for 1987 and 1988, were produced. This was accomplished much before we started using the GEOS II GCM. These integrations were started from ECMWF analysis for 1, 2, and 3 April (three consecutive days) of each year and ended on 30 September at the same time of the same year. We analyzed the ensemble of four simulations for statistical significance of the precipitation anomalies. The model successfully reproduced the Indian and Central American droughts (Fig. 10). The statistically significant regions of precipitation differences are shaded (90% and 95% significance on a two-tailed t test). The Indian drought is well simulated both in magnitude as well as in intensity as validated vis-à-vis the Huffman et al. (1997) data. We note that our model’s success is substantially better than that of Krishnan and Fennessy (1997) who used the Center for Ocean–Land–Atmosphere GCM. In other regions consisting of Sahelian Africa, we have limited success, while for Central America, we have a substantial success. This is not a direct evaluation of the performance of McRAS in the GEOS GCM, but it suggests that the GCM with McRAS appears to be performing at a satisfactory level.

5. Summary and conclusions

The current version of McRAS was first implemented in the GEOS I GCM and later in the new GEOS II GCM. McRAS significantly improved the climatology of hydrologic processes simulated in GEOS I GCM, but it had only a marginal impact on the cloud-radiative forcing of the GEOS II GCM. The precipitation climatology has several systematic deficiencies, such as too little in the Tropics and too much at high latitudes, as compared to observations. The planetary albedos and OLR also yield internally consistent systematic errors for the Tropics and high latitudes. The simulated high cloudiness is much too large and the middle-level cloudiness is too small as compared to ISCCP C2 data. As explained in section 4c, it may partly be a consequence of interpretation of high clouds as middle-level clouds in the ISCCP data analysis and partly a model deficiency. One must also remember, however, that thin cirrus clouds with optical thicknesses below 0.3 are not picked up by the ISCCP algorithm (W. B. Rossow 1998, personal communication). Since the high-level clouds often have very low optical thickness and may appear above the middle-level clouds particularly in the Tropics, these considerations would address some of the disagreement between the simulations and validation data. On the other hand, our cloud water amount has a severe systematic bias in its structure. It is a little too low in the Tropics over the oceanic rainbelt and a little too high over the winter storm track regions. Since these errors are accompanied by analogous precipitation errors, their resolution may be more complicated, that is, beyond simple resolution through logical adjustment of parameters.

The in-cloud water content can be easily modified by adjusting the constants in Sundqvist’s (1988) autoconversion equation. In addition, the importance of minimum cloud water for the onset of autoconversion shows a large sensitivity. In an earlier simulation with GEOS I GCM we had a limit of 0.1 gw kg−1a instead of the current limit in which lmin is a function of density of air and lcrit. We have not systematically investigated the influence of different algorithms of autoconversion that have been published recently (see, e.g., Fowler et al. 1996; Smith 1990; Schlesinger and Oh 1993; Cheng and Arakawa 1997). Following such a study, we plan to include more detailed features of cloud microphysics in a future version of McRAS. Nevertheless, the implementation of CTEI for cloud dissipation, cloud advection and diffusion, and Cahalan (1994) correction to allow for the cloud water substance inhomogeneity have all helped to improve the cloud-radiative forcing.

The summers of 1987 and 1988 were highly anomalous because a major El Niño event in 1987 was followed by a strong La Niña event in 1988. India suffered the worst drought of the century in 1987, along with huge precipitation deficits or excess in various other parts of the world. This was followed by abundant precipitation in the summer of 1988. Our model with McRAS has done a decent job of simulating the key features of the JAS simulation of 1987 minus 1988 precipitation differences. For Sahel, the GCM with McRAS got the sign of the large-scale precipitation anomaly right but did quite poorly on the magnitude as well as location. Over tropical Central America, however, the model performed much more skillfully. Each of these features is better simulated in the current study as compared to our earlier AMIP integration with both GLA and GEOS I GCMs. This gives us some confidence in the relative success of McRAS. Since several degrees of freedom were added into McRAS, it is pleasantly satisfying to learn that there is an overall improvement. However, this is only a start; we believe that McRAS needs additional design improvements and will get much better when new evaluation and calibration with atmospheric radiation measurement CART datasets have been conducted.

We were also successful in simulating the MJOs (30–60-day oscillations) in the Tropics. The earlier GLA GCM did a fairly decent job of simulating MJOs in the AMIP environment and two subsequent 10-yr integrations. It was duly recognized for this success by the outside community (e.g., Sperber et al. 1996). We now know that the GEOS II GCM can also produce MJOs even though GEOS I did not. Indeed, the new DAO GDAS with GEOS II has fairly strong and robust signatures for MJOs that are very similar to those of the ECMWF analysis. Consequently, we argue that the incidence of MJOs is intimately linked to the model physics. In our future research, we will be looking into the modus operandi of MJOs in the GEOS II GCM simulation so that we can capture, explain, and understand its vagaries. It must be emphasized that understanding MJOs is not only important for tropical dynamics, but MJOs have a strong influence on global dynamics and weather.

Acknowledgments

We thank K. Bergman of NASA Headquarters for supporting our research. This work was undertaken on advisement from Dr. W. K.-M. Lau, who has been a constant source of encouragement. We thank Dr. A. Del Genio of GISS for several useful interactions on the subject and Dr. M.-D. Chou for help with the implementation of his radiation package. Useful discussions with Dr. S. Moorthi of NCEP for understanding the nitty-gritty details in RAS and Dr. M. J. Suarez for suggestions in designing the cloud advection scheme are also gratefully acknowledged. David M. Mocko made thoughtful comments and provided useful English editing.

REFERENCES

  • Arakawa, A., and W. H. Schubert, 1974: Interaction of cumulus cloud ensemble with the large-scale environment, Part I. J. Atmos. Sci.,31, 674–701.

  • Barkstrom, B. R., 1984: The Earth Radiation Budget Experiment (ERBE). Bull. Amer. Meteor. Soc.,65, 1170–1185.

  • Cahalan, R., 1994: Bounded cascade clouds: Albedo and effective thickness. Non-linear Process. Geophys.,1, 156–167.

  • Cheng, M.-D., and A. Arakawa, 1997: Inclusion of rainwater budget and convective downdrafts in the Arakawa–Schubert cumulus parameterization. J. Atmos. Sci.,54, 1359–1378.

  • Chou, M.-D., and M. J. Suarez, 1994: An efficient thermal infrared radiation parameterization for use in general circulation models. NASA Tech. Memo. 104606, Vol. 3, 85 pp. [Available from NASA Center for Aerospace Information, 800 Elkridge Landing Rd., Linthieum Heights, MD 21090-2934.].

  • ——, ——, C.-H. Ho, M.-H. Yan, and K.-T. Lee, 1998. Parameterization for cloud overlapping and shortwave single-scattering properties for use in general circulation and cloud ensemble models. J. Climate,11, 202–214.

  • ——, K.-T. Lee, and S.-C. Tsay, 1999: Parameterization of cloud longwave scattering for use in atmospheric models. J. Climate,12, 159–169.

  • Del Genio, A. D., N.-S. Yao, W. Kovari, and K. K.-W. Lo, 1996: A prognostic cloud water parameterization for general circulation models. J. Climate,9, 270–304.

  • Emanuel, K. A., J. D. Neelin, and C. S. Bretherton, 1994: On large scale circulation in convective atmospheres. Quart J. Roy. Meteor. Soc.,120, 1111–1143.

  • Ferraro, R., F. Weng, N. Grody, and A. Basist, 1996: An eight-year (1987–1994) time series of rainfall, clouds, water vapor, snow cover, and sea ice derived from SSM/I measurements. Bull. Amer. Meteor. Soc.,77, 891–905.

  • Fowler, L. D., and D. A. Randall, 1996a: Liquid and cloud ice microphysics in the CSU GCM. Part II: Impact of cloudiness, the earth’s radiation budget, and general circulation of the atmosphere. J. Climate,9, 530–560.

  • ——, and ——, 1996b: Liquid and cloud ice microphysics in the CSU GCM. Part III: Sensitivity to modeling assumptions. J. Climate,9, 561–586.

  • ——, ——, and S. A. Rutledge, 1996: Liquid and cloud ice microphysics in the CSU GCM. Part I: Model description and simulated microphysical processes. J. Climate,9, 489–529.

  • Gates, W. L., 1992: AMIP: The Atmospheric Model Intercomparison Project. Bull. Amer. Meteor. Soc.,73, 1962–1970.

  • Greenwald, T. J., G. L. Stephens, and D. L. Jackson, 1993: A physical retrieval of cloud liquid water over global oceans using Special Sensor Microwave/Imager (SSM/I) observations. J. Geophys. Res.,98, 18 471–18 488.

  • Harshvardhan, D. A. Randall, and T. G. Corsetti, 1987: A fast radiation parameterization for general circulation models. J. Geophys. Res.,92, 1009–1016.

  • Helfand, H. M., and J. C. Lebraga, 1988: Design of a nonsingular level 2.5 second-order closure model for prediction of atmospheric turbulence. J. Atmos. Sci.,45, 113–132.

  • Hide, R., J. O. Dickey, S. L. Marcus, R. D. Rosen, and D. A. Salstein, 1997: Atmospheric angular momentum fluctuations during 1979–1988 simulated by global circulation models. J. Geophys. Res., (D), 102, 16 423–16 438.

  • Houghton, J. T., L. G. Meria Filho, B. A. Callander, N. Harris, A. Kattenberg, and K. Maskell, 1995: The Science of Climate Change: Intergovernmental Panel on Climate Change. Cambridge University Press, 572 pp.

  • Huffman, G. J., and Coauthors, 1997: The global precipitation climatology project (GPCP) combined precipitation datasets. Bull. Amer. Meteor. Soc.,78, 5–20.

  • Kiehl, J. T., and V. Ramanathan, 1990: Comparison of cloud forcing derived from Earth Radiation Budget Experiment with that simulated by NCAR Community Climate Model. J. Geophys. Res.,95, 11 679–11 698.

  • Krishnan, R., and M. J. Fennessy, 1997: GCM simulation of the intraseasonal variability in the Indian summer monsoon. Center for Ocean–Land–Atmosphere Studies, Rep. 40, 54 pp. [Available from COLAS, 4041 Powder Mill Road, Suite 302, Calverton, MD 20705-3106.].

  • Lau, K.-M., and S. Yang, 1996: Seasonal variation, abrupt transition, and intraseasonal variability associated with the Asian summer monsoon in the GLA GCM. J. Climate,9, 965–985.

  • ——, Y. C. Sud, and J.-H. Kim, 1996: Intercomparison of hydrologic processes in GCMs. Bull. Amer. Meteor. Soc.,77, 2209–2227.

  • Lord, S. J., W. C. Chao, and A. Arakawa, 1982: Interaction of a cumulus cloud ensemble with the large-scale environment. Part IV: The discrete model. J. Atmos Sci.,39, 104–113.

  • Miller, R. L., and A. D. Del Genio, 1994: Tropical cloud feedback and natural variability of climate. J. Climate,7, 1388–1402.

  • Moorthi, S., and M. J. Suarez, 1992: Relaxed Arakawa–Schubert: A parameterization of moist convection for general circulation models. Mon. Wea. Rev.,120, 978–1002.

  • Ou, S. C., and K. N. Liou, 1995: Ice microphysics and climate temperature feedback. Atmos. Res.35, 127–138.

  • Randall, D. A., Harshvardhan, D. A. Dazlich, and T. G. Corsetti, 1989: Interaction among radiation, convection, and large-scale dynamics in a general circulation model. J. Atmos. Sci.,46, 1943–1970.

  • Rasch, P., and J. E. Kristajansson, 1998: A comparison of the CCM3 model climate using diagnosed and predicted condensate parameterizations. J. Climate,11, 1587–1614.

  • Rood, R., cited 1999: Algorithm theoretical basis document for Goddard Earth Observing System Data Assimilation System (GEOS DAS) with a focus on version 2. [Available online at http://dao.gsfc.nasa.gov/subpages/atbd.html.].

  • Rossow, W. B., and R. A. Schiffer, 1991: ISCCP cloud data products. Bull. Amer. Meteor. Soc.,72, 2–20.

  • ——, and L. C. Gardner, 1993: Cloud detection using satellite measurements of infrared and visible radiances for ISSCP. J. Climate,6, 1403–1420.

  • Santer, B. D., K. E. Taylor, and L. C. Corsetti, 1995: Statistical evaluation of AMIP model performance. Proc. First International AMIP Scientific Conf., Monterey, CA, World Climate Research, 13–18. [WMO/TD 732.].

  • Schlesinger, M. E., and J.-H. Oh, 1993: A cloud-evaporation parameterization for general circulation models. Mon. Wea. Rev.,121, 1239–1248.

  • Slingo, A., and J. M. Slingo, 1988: The response of a GCM to cloud long wave radiative forcing. I: Introduction and initial experiments. Quart. J. Roy. Meteor. Soc.,114, 1027–1062.

  • Slingo, J., 1987: The development and verification of a cloud prediction scheme for the ECMWF model. Quart. J. Roy. Meteor. Soc.,113, 899–927.

  • Smith, R. N. B., 1990: A scheme for predicting layer clouds and their water content in a general circulation model. Quart J. Roy. Meteor. Soc.,116, 435–460.

  • Spencer, R. W., 1993: Global oceanic precipitation from MSU during 1979–91 and comparison to other climatologies. J. Climate,6, 1301–1326.

  • Sperber, K. R., J. M. Slingo, P. M. Inness, and W. K.-M. Lau, 1996:On the maintenance and initiation of the intraseasonal oscillation in the NCEP/NCAR reanalysis and the GLA and UKMO AMIP simulations. Rep. 36, 60 pp. [Available from PCMDI, Lawrence Livermore National Laboratory, P. O. Box 808, L-264, Livermore, CA 94550.].

  • Sud, Y. C., and A. Molod, 1988: The roles of dry convection, cloud-radiation feedback processes and the influence of recent improvements in the parameterization of convection in the GLA GCM. Mon. Wea. Rev.,116, 2366–2387.

  • ——, and G. K. Walker, 1993: A rain evaporation and downdraft parameterization to complement a cumulus updraft scheme and its evaluation using GATE data. Mon. Wea. Rev.,121, 3019–3039.

  • ——, and G. K. Walker, 1999: Microphysics of clouds with the Relaxed Arakawa–Schubert Scheme (McRAS). Part I: Design and evaluation with GATE Phase III data. J. Atmos. Sci.,56, 3196–3220.

  • ——, W. C. Chao, and G. K. Walker, 1991: Contributions to the implementation of Arakawa–Schubert cumulus parameterization in the GLA GCM. J. Atmos. Sci.,48, 1573–1586.

  • Sundqvist, H., 1988: Parameterization of condensation and associated clouds in models for weather prediction and general circulation simulation. Physically Based Modelling and Simulation of Climate and Climatic Change, M. E. Schlesinger, Ed., D. Riedel, 433–461.

  • ——, 1993: Inclusion of ice-phase of hydrometeors in cloud parameterization of large-scale and meso-scale models. Contrib. Atmos. Phys.,66, 137–147.

  • Takacs, L. L., A. Molod, and T. Weng, 1994: Goddard Earth Observing System (GEOS) general circulation model (GCM) version 1. NASA Tech. Memo. 104606, Vol. 1, 97 pp. [Available from NASA/Goddard Space Flight Center, Greenbelt, MD 20771.].

  • Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea Rev.,121, 3040–3061.

  • Waliser, D. E., K.-M. Lau, and J.-H. Kim, 1999: The influence of coupled sea surface temperatures on the Madden–Julian Oscillation: A model perturbation experiment. J. Atmos. Sci.,56, 333–358.

  • Wang, W., and M. E. Schlesinger, 1999: The dependence on convection parameterization of the tropical intraseasonal oscillation simulated by the 11-layer UIUC atmospheric GCM. J. Climate,12, 1423–1457.

  • Weng, F., and N. Grody, 1994: Retrievals of cloud liquid water using Special Sensor Microwave/Imager (SSM/I). J. Geophys. Res.,99, 25 535–25 551.

  • Xue, Y. K., P. J. Sellers, J. L. Kinter III, and J. Shukla, 1991: A simplified biosphere model for global climate studies. J. Climate,4, 345–364.

  • Zhao, Q., and F. H. Carr, 1997: A prognostic cloud scheme for operational NWP models. Mon. Wea Rev.,125, 1931–1953.

  • Zhou, J., Y. C. Sud, and K.-M. Lau, 1996: Impact of orographically induced gravity wavedrag in the GLA GCM. Quart. J. Roy. Meteor. Soc.,122, 903–927.

i1520-0469-56-18-3221-f01a

Fig. 1a. Zonal average planetary albedo (%, circles) and OLR (W m−2, squares) emulated in McRAS II vs ERBE data for the 2-yr (1987–88) average during the simulated periods: DJF (top) and JJA (bottom). Filled symbols are for McRAS II, plain line is for control, and hollow symbols are for ERBE data.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f01b

Fig. 1b. Monthly time series from 1 Jan 1987 through 1 Jan 1991, for planetary albedo (%) and simulated in McRAS II and simulated minus ERBE data. Verification data were available for two years only.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f01c

Fig. 1c. Same as Fig. 1b except for OLR (W m−2). Verification data were available for three years only.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f01d

Fig. 1d. Same as Fig. 1a except for shortwave and longwave CRF (W m−2).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f02a

Fig. 2a. Time series of zonal average monthly precipitation (mm day−1) and its 4-month mean: (top) McRAS II simulation; (bottom) Huffman et al. (1997) data. All 4-yr averages are thick, winter (JFM) average is dotted, and summer (JAS) average is dashed. An additional thin line (top) is for 4-yr mean of the control simulation.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f02b

Fig. 2b. Simulated vis-à-vis observed precipitation (mm day−1) for all four winters (JFM).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f02c

Fig. 2c. Simulated vis-à-vis observed precipitation (mm day−1) for all four summers (JAS).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f03a

Fig. 3a. Zonal average total cloudiness for all four years (top) High-level clouds; (middle) middle-level clouds; and (bottom) total clouds. McRAS II (SSM/I) fields appear in the left (right) panels.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f03b

Fig. 3b. Total cloudiness for one DJF for McRAS II (top), ISSCP data (middle), and SSM/I data of Ferraro et al. (1996) (bottom).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f03c

Fig. 3c. Total cloudiness for one JJA for McRAS II (top), ISSCP data (middle), and SSM/I data of Ferraro et al. (1996) (bottom).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f04a

Fig. 4a. Annual cycle of zonally averaged total-column cloud water in 10−2 g kg−1, averaged over the oceans for McRAS II (top), ISCCP data (middle), and SSM/I data (bottom).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f04b

Fig. 4b. Total in-cloud water in 10−2 g kg−1 for one DJF for McRAS II (top), and SSM/I data of Ferraro et al. (1996) (bottom).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f04c

Fig. 4c. Total in-cloud water in 10−2 g kg−1 for one JJA for McRAS II (top), and SSM/I data of Ferraro et al. (1996) (bottom).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

Fig. 5.
Fig. 5.

Madden–Julian (30–60 day) oscillation as seen in the 200-hPa velocity potentials (106 kg s−1) filtered with high-and low-pass filters for (a) McRAS II, (b) the control case,

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

Fig. 5.
Fig. 5.

(Continued) and (c) ECMWF analysis. The data are averaged for 10°S–10°N and filtered to retain 20–100-day periods.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f06a

Fig. 6a. Cloud water production and dissipation tendencies in 10−3 g kg−1 day−1 in McRAS II for a single JJA simulation. (left) Cloud water generation tendency by (a) stratiform cloud processes, (b) moist-convective processes, (c) boundary layer eddies, and (d) all cloud water generation processes. (right) Cloud water dissipation tendencies by (e) cloud munching; (f) CTEI; (g) cloud advection, subsidence, and entrainment by downdrafts; and (h) all cloud water dissipation processes.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f06b

Fig. 6b. Same as (a) except for cloud fraction (% day−1).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f07a

Fig. 7a. Zonal and seasonal mean cloud fraction differences for one DJF (top) and one JJA (bottom) due to no CTEI (thick lines) and no cloud advection (thin lines). Differences are generated as a sensitivity simulation minus McRAS II.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f07b

Fig. 7b. Influence of CTEI and CA on CRF error (McRAS II minus ERBE data) for DJF (top) and JJA (bottom). Solid (dashed) lines for SW (LW) CRF. The simulations are distinguished by filled circles (McRAS II), hollow squares (no CTEI), hollow circles (no CA), and plain line (without Cahalan correction)

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f08a

Fig. 8a. Simulated in-cloud water (10−3 g kg−1) binned from 60°S to 60°N as a function of temperature for a single JJA simulation. Line distinctions are McRAS II control (thick), lmin = zero (thin), lmin = 100 (dash), lmin = 0.5 × lcrit (dotted), and lcrit = 333.33 (thick chain).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f08b

Fig. 8b. Zonal average all-sky cloud water (10−3 g kg−1) for a single DJF McRAS II simulation. Line distinctions are McRAS II control (thick), lmin = zero (thin), lmin = 100 (dash), lmin = 0.5 × lcrit (dotted), and ISCCP data (thick dash).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

i1520-0469-56-18-3221-f08c

Fig. 8c. Zonal average temperatures (K) for DJF (top) and JJA (bottom). ECMWF analysis (thick), McRAS (thin), lmin = zero (dotted), and lmin = 100 × 10−3 g kg−1 (dash).

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

Fig. 9.
Fig. 9.

(a) Influence of lcrit on the cloud water substance (10−3 g kg−1) as a function of temperature. McRAS (thick), lcrit = 333.33 (thin), and lcrit = 1000 (dash). (b) Same as Fig. 1d except for the influence of the choice of lcrit (10−3 g kg−1). Line types correspond to those in Fig. 9a.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

Fig. 10.
Fig. 10.

Precipitation differences (mm day−1), 1987–88, for JAS period as (a) simulated by the GEOS GCM with McRAS and (b) in GPCP data analysis of Huffman et al. (1997). Shaded regions show statistical significance at 90% and 95% levels in the ensemble set of four simulations.

Citation: Journal of the Atmospheric Sciences 56, 18; 10.1175/1520-0469(1999)056<3221:MOCWTR>2.0.CO;2

Table 1.

Simulation experiments.

Table 1.
Save
  • Arakawa, A., and W. H. Schubert, 1974: Interaction of cumulus cloud ensemble with the large-scale environment, Part I. J. Atmos. Sci.,31, 674–701.

  • Barkstrom, B. R., 1984: The Earth Radiation Budget Experiment (ERBE). Bull. Amer. Meteor. Soc.,65, 1170–1185.

  • Cahalan, R., 1994: Bounded cascade clouds: Albedo and effective thickness. Non-linear Process. Geophys.,1, 156–167.

  • Cheng, M.-D., and A. Arakawa, 1997: Inclusion of rainwater budget and convective downdrafts in the Arakawa–Schubert cumulus parameterization. J. Atmos. Sci.,54, 1359–1378.

  • Chou, M.-D., and M. J. Suarez, 1994: An efficient thermal infrared radiation parameterization for use in general circulation models. NASA Tech. Memo. 104606, Vol. 3, 85 pp. [Available from NASA Center for Aerospace Information, 800 Elkridge Landing Rd., Linthieum Heights, MD 21090-2934.].

  • ——, ——, C.-H. Ho, M.-H. Yan, and K.-T. Lee, 1998. Parameterization for cloud overlapping and shortwave single-scattering properties for use in general circulation and cloud ensemble models. J. Climate,11, 202–214.

  • ——, K.-T. Lee, and S.-C. Tsay, 1999: Parameterization of cloud longwave scattering for use in atmospheric models. J. Climate,12, 159–169.

  • Del Genio, A. D., N.-S. Yao, W. Kovari, and K. K.-W. Lo, 1996: A prognostic cloud water parameterization for general circulation models. J. Climate,9, 270–304.

  • Emanuel, K. A., J. D. Neelin, and C. S. Bretherton, 1994: On large scale circulation in convective atmospheres. Quart J. Roy. Meteor. Soc.,120, 1111–1143.

  • Ferraro, R., F. Weng, N. Grody, and A. Basist, 1996: An eight-year (1987–1994) time series of rainfall, clouds, water vapor, snow cover, and sea ice derived from SSM/I measurements. Bull. Amer. Meteor. Soc.,77, 891–905.

  • Fowler, L. D., and D. A. Randall, 1996a: Liquid and cloud ice microphysics in the CSU GCM. Part II: Impact of cloudiness, the earth’s radiation budget, and general circulation of the atmosphere. J. Climate,9, 530–560.

  • ——, and ——, 1996b: Liquid and cloud ice microphysics in the CSU GCM. Part III: Sensitivity to modeling assumptions. J. Climate,9, 561–586.

  • ——, ——, and S. A. Rutledge, 1996: Liquid and cloud ice microphysics in the CSU GCM. Part I: Model description and simulated microphysical processes. J. Climate,9, 489–529.

  • Gates, W. L., 1992: AMIP: The Atmospheric Model Intercomparison Project. Bull. Amer. Meteor. Soc.,73, 1962–1970.

  • Greenwald, T. J., G. L. Stephens, and D. L. Jackson, 1993: A physical retrieval of cloud liquid water over global oceans using Special Sensor Microwave/Imager (SSM/I) observations. J. Geophys. Res.,98, 18 471–18 488.

  • Harshvardhan, D. A. Randall, and T. G. Corsetti, 1987: A fast radiation parameterization for general circulation models. J. Geophys. Res.,92, 1009–1016.

  • Helfand, H. M., and J. C. Lebraga, 1988: Design of a nonsingular level 2.5 second-order closure model for prediction of atmospheric turbulence. J. Atmos. Sci.,45, 113–132.

  • Hide, R., J. O. Dickey, S. L. Marcus, R. D. Rosen, and D. A. Salstein, 1997: Atmospheric angular momentum fluctuations during 1979–1988 simulated by global circulation models. J. Geophys. Res., (D), 102, 16 423–16 438.

  • Houghton, J. T., L. G. Meria Filho, B. A. Callander, N. Harris, A. Kattenberg, and K. Maskell, 1995: The Science of Climate Change: Intergovernmental Panel on Climate Change. Cambridge University Press, 572 pp.

  • Huffman, G. J., and Coauthors, 1997: The global precipitation climatology project (GPCP) combined precipitation datasets. Bull. Amer. Meteor. Soc.,78, 5–20.

  • Kiehl, J. T., and V. Ramanathan, 1990: Comparison of cloud forcing derived from Earth Radiation Budget Experiment with that simulated by NCAR Community Climate Model. J. Geophys. Res.,95, 11 679–11 698.

  • Krishnan, R., and M. J. Fennessy, 1997: GCM simulation of the intraseasonal variability in the Indian summer monsoon. Center for Ocean–Land–Atmosphere Studies, Rep. 40, 54 pp. [Available from COLAS, 4041 Powder Mill Road, Suite 302, Calverton, MD 20705-3106.].

  • Lau, K.-M., and S. Yang, 1996: Seasonal variation, abrupt transition, and intraseasonal variability associated with the Asian summer monsoon in the GLA GCM. J. Climate,9, 965–985.

  • ——, Y. C. Sud, and J.-H. Kim, 1996: Intercomparison of hydrologic processes in GCMs. Bull. Amer. Meteor. Soc.,77, 2209–2227.

  • Lord, S. J., W. C. Chao, and A. Arakawa, 1982: Interaction of a cumulus cloud ensemble with the large-scale environment. Part IV: The discrete model. J. Atmos Sci.,39, 104–113.

  • Miller, R. L., and A. D. Del Genio, 1994: Tropical cloud feedback and natural variability of climate. J. Climate,7, 1388–1402.

  • Moorthi, S., and M. J. Suarez, 1992: Relaxed Arakawa–Schubert: A parameterization of moist convection for general circulation models. Mon. Wea. Rev.,120, 978–1002.

  • Ou, S. C., and K. N. Liou, 1995: Ice microphysics and climate temperature feedback. Atmos. Res.35, 127–138.

  • Randall, D. A., Harshvardhan, D. A. Dazlich, and T. G. Corsetti, 1989: Interaction among radiation, convection, and large-scale dynamics in a general circulation model. J. Atmos. Sci.,46, 1943–1970.

  • Rasch, P., and J. E. Kristajansson, 1998: A comparison of the CCM3 model climate using diagnosed and predicted condensate parameterizations. J. Climate,11, 1587–1614.

  • Rood, R., cited 1999: Algorithm theoretical basis document for Goddard Earth Observing System Data Assimilation System (GEOS DAS) with a focus on version 2. [Available online at http://dao.gsfc.nasa.gov/subpages/atbd.html.].

  • Rossow, W. B., and R. A. Schiffer, 1991: ISCCP cloud data products. Bull. Amer. Meteor. Soc.,72, 2–20.

  • ——, and L. C. Gardner, 1993: Cloud detection using satellite measurements of infrared and visible radiances for ISSCP. J. Climate,6, 1403–1420.

  • Santer, B. D., K. E. Taylor, and L. C. Corsetti, 1995: Statistical evaluation of AMIP model performance. Proc. First International AMIP Scientific Conf., Monterey, CA, World Climate Research, 13–18. [WMO/TD 732.].

  • Schlesinger, M. E., and J.-H. Oh, 1993: A cloud-evaporation parameterization for general circulation models. Mon. Wea. Rev.,121, 1239–1248.

  • Slingo, A., and J. M. Slingo, 1988: The response of a GCM to cloud long wave radiative forcing. I: Introduction and initial experiments. Quart. J. Roy. Meteor. Soc.,114, 1027–1062.

  • Slingo, J., 1987: The development and verification of a cloud prediction scheme for the ECMWF model. Quart. J. Roy. Meteor. Soc.,113, 899–927.

  • Smith, R. N. B., 1990: A scheme for predicting layer clouds and their water content in a general circulation model. Quart J. Roy. Meteor. Soc.,116, 435–460.

  • Spencer, R. W., 1993: Global oceanic precipitation from MSU during 1979–91 and comparison to other climatologies. J. Climate,6, 1301–1326.

  • Sperber, K. R., J. M. Slingo, P. M. Inness, and W. K.-M. Lau, 1996:On the maintenance and initiation of the intraseasonal oscillation in the NCEP/NCAR reanalysis and the GLA and UKMO AMIP simulations. Rep. 36, 60 pp. [Available from PCMDI, Lawrence Livermore National Laboratory, P. O. Box 808, L-264, Livermore, CA 94550.].

  • Sud, Y. C., and A. Molod, 1988: The roles of dry convection, cloud-radiation feedback processes and the influence of recent improvements in the parameterization of convection in the GLA GCM. Mon. Wea. Rev.,116, 2366–2387.

  • ——, and G. K. Walker, 1993: A rain evaporation and downdraft parameterization to complement a cumulus updraft scheme and its evaluation using GATE data. Mon. Wea. Rev.,121, 3019–3039.

  • ——, and G. K. Walker, 1999: Microphysics of clouds with the Relaxed Arakawa–Schubert Scheme (McRAS). Part I: Design and evaluation with GATE Phase III data. J. Atmos. Sci.,56, 3196–3220.

  • ——, W. C. Chao, and G. K. Walker, 1991: Contributions to the implementation of Arakawa–Schubert cumulus parameterization in the GLA GCM. J. Atmos. Sci.,48, 1573–1586.

  • Sundqvist, H., 1988: Parameterization of condensation and associated clouds in models for weather prediction and general circulation simulation. Physically Based Modelling and Simulation of Climate and Climatic Change, M. E. Schlesinger, Ed., D. Riedel, 433–461.

  • ——, 1993: Inclusion of ice-phase of hydrometeors in cloud parameterization of large-scale and meso-scale models. Contrib. Atmos. Phys.,66, 137–147.

  • Takacs, L. L., A. Molod, and T. Weng, 1994: Goddard Earth Observing System (GEOS) general circulation model (GCM) version 1. NASA Tech. Memo. 104606, Vol. 1, 97 pp. [Available from NASA/Goddard Space Flight Center, Greenbelt, MD 20771.].

  • Tiedtke, M., 1993: Representation of clouds in large-scale models. Mon. Wea Rev.,121, 3040–3061.

  • Waliser, D. E., K.-M. Lau, and J.-H. Kim, 1999: The influence of coupled sea surface temperatures on the Madden–Julian Oscillation: A model perturbation experiment. J. Atmos. Sci.,56, 333–358.

  • Wang, W., and M. E. Schlesinger, 1999: The dependence on convection parameterization of the tropical intraseasonal oscillation simulated by the 11-layer UIUC atmospheric GCM. J. Climate,12, 1423–1457.

  • Weng, F., and N. Grody, 1994: Retrievals of cloud liquid water using Special Sensor Microwave/Imager (SSM/I). J. Geophys. Res.,99, 25 535–25 551.

  • Xue, Y. K., P. J. Sellers, J. L. Kinter III, and J. Shukla, 1991: A simplified biosphere model for global climate studies. J. Climate,4, 345–364.

  • Zhao, Q., and F. H. Carr, 1997: A prognostic cloud scheme for operational NWP models. Mon. Wea Rev.,125, 1931–1953.

  • Zhou, J., Y. C. Sud, and K.-M. Lau, 1996: Impact of orographically induced gravity wavedrag in the GLA GCM. Quart. J. Roy. Meteor. Soc.,122, 903–927.

  • Fig. 1a. Zonal average planetary albedo (%, circles) and OLR (W m−2, squares) emulated in McRAS II vs ERBE data for the 2-yr (1987–88) average during the simulated periods: DJF (top) and JJA (bottom). Filled symbols are for McRAS II, plain line is for control, and hollow symbols are for ERBE data.

  • Fig. 1b. Monthly time series from 1 Jan 1987 through 1 Jan 1991, for planetary albedo (%) and simulated in McRAS II and simulated minus ERBE data. Verification data were available for two years only.

  • Fig. 1c. Same as Fig. 1b except for OLR (W m−2). Verification data were available for three years only.

  • Fig. 1d. Same as Fig. 1a except for shortwave and longwave CRF (W m−2).

  • Fig. 2a. Time series of zonal average monthly precipitation (mm day−1) and its 4-month mean: (top) McRAS II simulation; (bottom) Huffman et al. (1997) data. All 4-yr averages are thick, winter (JFM) average is dotted, and summer (JAS) average is dashed. An additional thin line (top) is for 4-yr mean of the control simulation.

  • Fig. 2b. Simulated vis-à-vis observed precipitation (mm day−1) for all four winters (JFM).

  • Fig. 2c. Simulated vis-à-vis observed precipitation (mm day−1) for all four summers (JAS).

  • Fig. 3a. Zonal average total cloudiness for all four years (top) High-level clouds; (middle) middle-level clouds; and (bottom) total clouds. McRAS II (SSM/I) fields appear in the left (right) panels.

  • Fig. 3b. Total cloudiness for one DJF for McRAS II (top), ISSCP data (middle), and SSM/I data of Ferraro et al. (1996) (bottom).

  • Fig. 3c. Total cloudiness for one JJA for McRAS II (top), ISSCP data (middle), and SSM/I data of Ferraro et al. (1996) (bottom).

  • Fig. 4a. Annual cycle of zonally averaged total-column cloud water in 10−2 g kg−1, averaged over the oceans for McRAS II (top), ISCCP data (middle), and SSM/I data (bottom).

  • Fig. 4b. Total in-cloud water in 10−2 g kg−1 for one DJF for McRAS II (top), and SSM/I data of Ferraro et al. (1996) (bottom).

  • Fig. 4c. Total in-cloud water in 10−2 g kg−1 for one JJA for McRAS II (top), and SSM/I data of Ferraro et al. (1996) (bottom).

  • Fig. 5.

    Madden–Julian (30–60 day) oscillation as seen in the 200-hPa velocity potentials (106 kg s−1) filtered with high-and low-pass filters for (a) McRAS II, (b) the control case,

  • Fig. 5.

    (Continued) and (c) ECMWF analysis. The data are averaged for 10°S–10°N and filtered to retain 20–100-day periods.

  • Fig. 6a. Cloud water production and dissipation tendencies in 10−3 g kg−1 day−1 in McRAS II for a single JJA simulation. (left) Cloud water generation tendency by (a) stratiform cloud processes, (b) moist-convective processes, (c) boundary layer eddies, and (d) all cloud water generation processes. (right) Cloud water dissipation tendencies by (e) cloud munching; (f) CTEI; (g) cloud advection, subsidence, and entrainment by downdrafts; and (h) all cloud water dissipation processes.

  • Fig. 6b. Same as (a) except for cloud fraction (% day−1).

  • Fig. 7a. Zonal and seasonal mean cloud fraction differences for one DJF (top) and one JJA (bottom) due to no CTEI (thick lines) and no cloud advection (thin lines). Differences are generated as a sensitivity simulation minus McRAS II.

  • Fig. 7b. Influence of CTEI and CA on CRF error (McRAS II minus ERBE data) for DJF (top) and JJA (bottom). Solid (dashed) lines for SW (LW) CRF. The simulations are distinguished by filled circles (McRAS II), hollow squares (no CTEI), hollow circles (no CA), and plain line (without Cahalan correction)

  • Fig. 8a. Simulated in-cloud water (10−3 g kg−1) binned from 60°S to 60°N as a function of temperature for a single JJA simulation. Line distinctions are McRAS II control (thick), lmin = zero (thin), lmin = 100 (dash), lmin = 0.5 × lcrit (dotted), and lcrit = 333.33 (thick chain).

  • Fig. 8b. Zonal average all-sky cloud water (10−3 g kg−1) for a single DJF McRAS II simulation. Line distinctions are McRAS II control (thick), lmin = zero (thin), lmin = 100 (dash), lmin = 0.5 × lcrit (dotted), and ISCCP data (thick dash).

  • Fig. 8c. Zonal average temperatures (K) for DJF (top) and JJA (bottom). ECMWF analysis (thick), McRAS (thin), lmin = zero (dotted), and lmin = 100 × 10−3 g kg−1 (dash).

  • Fig. 9.

    (a) Influence of lcrit on the cloud water substance (10−3 g kg−1) as a function of temperature. McRAS (thick), lcrit = 333.33 (thin), and lcrit = 1000 (dash). (b) Same as Fig. 1d except for the influence of the choice of lcrit (10−3 g kg−1). Line types correspond to those in Fig. 9a.

  • Fig. 10.

    Precipitation differences (mm day−1), 1987–88, for JAS period as (a) simulated by the GEOS GCM with McRAS and (b) in GPCP data analysis of Huffman et al. (1997). Shaded regions show statistical significance at 90% and 95% levels in the ensemble set of four simulations.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 313 46 6
PDF Downloads 126 35 0