Upper tropospheric stratiform clouds associated with deep convection are important to global radiation budgets and to cloud-radiation feedbacks on climate variability and change. Several recent observational studies indicate that vertical wind shear is an important factor affecting stratiform cloud fraction and cloud overlap. This study further examines wind shear effects on cloud properties (including cloud fraction and cloud optical depth) and associated top of atmosphere (TOA) and surface radiative fluxes, using observations from the Tropical Ocean Global Atmosphere program's Coupled Ocean Atmosphere Response Experiment (TOGA COARE) experiment and long-term satellite measurements. Wind shear affects cloud-radiative fluxes, through both the cloud fraction and optical thickness, in a strong and systematic way. In typical convecting conditions, shear-induced additional cloudiness can reduce outgoing longwave radiation (OLR) by 10s of Wm−2, implying longwave radiative changes on the order of 10% of the total latent heating. Such cloud also reflects shortwave radiation, reducing surface downward flux (energy input to the ocean) by 10s of Wm−2. Current climate models lack these effects.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Cloud radiation feedback is well recognized as a key uncertainty in predicting any potential future climate change [Intergovernmental Panel on Climate Change, 2001]. The upper tropospheric stratiform clouds (anvil and cirrus clouds) associated with tropical deep convection are important to the earth's radiation budgets because of their large longwave (LW) and shortwave (SW) cloud radiative forcing (CRF [Ramanathan et al., 1989] Interestingly, the seasonal-mean LW CRF and SW CRF nearly cancel each other in current climate, but this delicate balance between two big terms is sensitive to the stratiform cloud properties, and breaking of the balance is observed when stratiform cloud properties change, for example, during a strong El Niño event [Cess et al., 2001], and in the tropical intraseasonal oscillation [Lin and Mapes, 2004]. Climate change modeling studies also show that, in different models, changes in stratiform cloud properties lead to different changes in LW CRF and SW CRF, because of their associated cloud fraction feedback [e.g., Hansen et al., 1984] and cloud optical depth feedback (e.g., Charlock ; Somerville and Remer ; see review by Del Genio and Wolf ).
 Although the SW CRF and LW CRF nearly cancel, SW cooling acts mainly at the surface while LW warming acts mainly on the atmosphere. The surface SW cooling is important for regulating the tropical sea surface temperature [e.g., Ramanathan and Collins, 1991]. The atmospheric LW warming, together with vertical gradients in its heating profile [e.g., Webster and Stephens, 1980; Ackerman et al., 1988], play an important role in tropical large-scale circulations such as the Hadley circulation [e.g., Randall et al., 1989], the Walker circulation [e.g., Bretherton and Sobel, 2002] and the tropical intraseasonal oscillation [e.g., Lin and Mapes, 2004]. Therefore, parameterization of stratiform clouds is an important aspect of General Circulation Models (GCMs).
 The art of cloud parameterization in GCMs has evolved steadily in complexity from prescribed cloud properties, to diagnostic schemes, to prognosed schemes (see reviews by Randall  and Fowler et al. ). Current schemes often use prognostic equations to predict various cloud and precipitation variables (such as cloud ice, rain, or snow), and include parameterizations of a large number of microphysical pathways between these categories. After that, fractional cloudiness can be predicted using either the diagnostic method (see review by Xu and Krueger ) or the statistical method (see review by Tompkins ). The predictors often employed are grid-scale thermodynamical and microphysical properties, such as relative humidity, total water content, convective mass flux, large-scale mass flux, and precipitation.
 In addition to the grid-scale thermodynamical and microphysical properties, observations indicate that vertical wind shear is another important factor affecting stratiform cloud fraction [Saxen and Rutledge, 2000] and cloud overlap characteristics [Mace and Benson-Troth, 2002; Hogan and Illingworth, 2003], yet shear dependence is not included in any climate model we know of, and evidently doesn't arise automatically (Figure 7).
 When evaluating a satellite rainfall retrieval algorithm using data from TOGA COARE, Saxen and Rutledge  found that for a given rainrate, the coverage of cold clouds (with cloud top temperature < 235 K or cloud top pressure < 275 hPa) increases with vertical wind shear. Using cloud radar data from a midlatitude station, Hogan and Illingworth  found that strong wind shear decreases the correlation length of clouds between different vertical levels, suggesting that wind shear decreases the cloud overlap and hence increases total cloud cover.
 As an example, visible satellite images for two scenes with similar mean rainrate (as estimated either from radar or from heat and moisture budgets) but quite different vertical wind shear are shown in Figure 1. The strong shear case (Figure 1b) is clearly associated with more cirrus coverage than the weak shear case (Figure 1a). In addition to this increased thin cirrus cloud, precipitating stratiform clouds also contribute a larger fraction of total rainfall in conditions of stronger shear [e.g., Lin et al., 2004].
 The above studies serve to raise the following questions:
 (1) How much does the wind shear affect the total cloud fraction and cloud top height, which is important for the cloud fraction feedback?
 (2) Does the wind shear also affect the cloud optical depth?
 (3) Is the wind shear effect detectable in the associated radiative fluxes?
 The purpose of this study is to address the above questions using observations from the TOGA COARE experiment and long-term satellite measurements. Datasets used are described in section 2. The wind shear effects are analyzed in section 3. Summary and discussions are given in section 4.
 Datasets used in this study include TOGA COARE data and long-term satellite data. The variables used include clouds, radiative fluxes, surface precipitation, and wind profiles.
 The TOGA COARE data are area-averages for the Intensive Flux Array (IFA) covering the 4-month period from November 1992 to February 1993. The clouds come from the 3-hourly International Satellite Cloud Climatology Project (ISCCP) D1 data [Rossow and Schiffer, 1999]; The TOA and surface radiative fluxes come from the 3-hourly data by Krueger and Burks , with the TOA fluxes from the GMS-derived broadband fluxes [Minnis et al., 1995], and the surface fluxes from averages of measurements at five surface stations. The surface precipitation comes from the 10-minute radar-estimated rainfall map [Short et al., 1997], which cover about 60% of the IFA area. The wind profiles come from the 6-hourly upper air soundings [Ciesielski et al., 2002]. All datasets are averaged to daily data and we use only the days without any missing data. The long-term measurements are 15 years (1979–1993) of daily mean (5475 days) OLR from Advanced Very High Resolution Radiometer (AVHRR [Liebmann and Smith, 1996]), precipitation from Microwave Sounding Units (MSU) [Spencer, 1993], and wind profiles from the NCEP/NCAR reanalysis [Kalnay et al., 1996]. We use the data averaged over a 10 degree by 10 degree region (5N–5S, 150–160E) in the western Pacific warm pool, which covers the TOGA COARE IFA.
3. Wind Shear Effects on Clouds and Radiative Fluxes
 First we consider wind shear effects on the TOGA COARE IFA clouds and radiation fluxes. We stratify the data by the strength of 150–700 hPa wind shear, and compare the lower and upper quartiles. Figure 2a shows the scatter diagram of ISCCP total cloud fraction versus radar-estimated rainrate. The weak shear quartile (open circles) and strong shear quartile (filled circles) appear as distinct regimes. For a given value of rainrate, strong shear increases the cloud fraction by more than 20%. The increase is mainly in high and middle clouds (not shown), resulting in an increase of averaged cloud top height (Figure 2b). Besides increasing the cloud fraction and cloud top height, wind shear also increases the cloud optical depth (Figure 3).
 The effects of shear on stratiform cloud properties should be detectable in the associated radiative fluxes. To test this possibility, Figure 4 shows the scatter diagram of GMS-derived broadband OLR versus rainrate. As expected, OLR generally decreases (note inverted scale of y axis) with increased rainrate. However, for a given value of rainrate, strong wind shear decreases the OLR by 10s of Wm−2, implying a radiative heating difference in the atmosphere which is more than 10% of the total latent heating for a moderate area-averaged rainrate of 0.3 mm/hr.
 In addition to the LW effect, wind shear also increases the TOA reflected SW flux by 10s of Wm−2 (Figure 5a), and decreases the downward SW flux at the ocean's surface, which is measured by independent ground-based instruments (Figure 5b by a similar amount).
 Comparison of Figure 4 and Figure 5a indicates that the shear-induced changes in TOA OLR and reflected SW flux largely cancel out. Therefore, the effect of shear is to repartition radiative heating between the atmosphere and the ocean, not to change the overall column radiative heating.
 Although a systematic relationship is clearly indicated by Figure 4 and Figure 5, the TOGA COARE data provide a too small sample for reliable quantitative assessment. To bolster the case, we also analyzed the long-term measurements of OLR from AVHRR, precipitation from MSU, and wind profiles from NCEP/NCAR reanalysis. The resulting joint dependence, binned into a 5 × 5 array of precipitation and shear categories, is shown in Figure 6. Contours slope up to the left, indicating that OLR decreases both with increased rainrate and with increased wind shear. For a given value of rainrate, strong shear reduces the OLR by about 20 Wm−2, which is consistent with the TOGA COARE results (Figure 4).
 Next we compare the observational results with a climate model – the NCAR Community Atmosphere Model CAM2, which uses a prognostic cloud microphysics scheme [Rasch and Kristjansson, 1998] and cirrus cloud fraction proportional to convective mass flux [Collins et al., 2000]. We used four years (1985–1988) of daily mean (1460 days) outputs from an AMIP-type simulation. The data were averaged along the equator (between 5N and 5S) with a zonal resolution of 10 degree longitude. To make the sample size comparable to that of the above long-term observational data, we used five 10 degree by 10 degree grids from 130E to 180E in the western Pacific warm pool. Figure 7 shows that, in contrast with observation (Figure 6), the CAM2 OLR does not change substantially with vertical wind shear.
4. Summary and Discussions
 In summary, wind shear strongly affects cloud-radiative fluxes, through both the cloud fraction and optical thickness. The effect is strong and systematic, with wind shear reducing OLR by 10s of Wm−2, implying radiative changes on the order of 10% of the total latent heating. Shear-induced cloud also reflects SW radiation, reducing surface downward flux (energy input to the ocean) by 10s of Wm−2.
 Our results raise the following questions for future studies:
 (1) What are the physical mechanisms behind the wind shear effect? Is it mainly due to increased cloud cover at one level or decreased cloud overlap between different levels? What causes the increase of cloud optical depth?
 (2) What are the impacts of the shear effect on climate and climate change?
 We plan to study these questions using more detailed long-term observations from the Atmospheric Radiation Measurement (ARM) sites, and outputs from GCM experiments.
 This study benefits from the discussions with Chris Bretherton, Dick Johnson, Dave Randall and Ed Zipser. Steve Krueger kindly provided the TOGA COARE radiative fluxes data. Jim Hurrell and Adam Phillips kindly provided the NCAR CAM2 outputs. The research described here was supported by NSF grants ATM-0336790, ATM-0097116 and ATM-0112715, and by NOAA OGP CMEP.