INVESTIGATION OF CLOUDS AND CLOUD RADIATIVE FORCING ON THE WINDWARD SIDE OF THE MADAGASCAR MOUNTAIN CHAINS

Clouds affect the radiative energy balance of the earth–atmosphere system by reflecting and trapping the radiation. The cooling occurs over the earth by reflecting the incoming solar radiation and warming by trapping the outgoing longwave terrestrial radiation. In this paper an attempt has been carried out to understand the clouds and cloud radiative forcing over the windward side of the Madagascar mountain chain. The study was carried out using the Clouds and Earth’s Radiant Energy System (CERES) data June – September from 2000 to 2016. Over the windward side, clouds tend to cool whereas on the leeward side, clouds tend to warm marginally. During this period, peak value of shortwave cloud forcing and the longwave cloud forcing are −45 Wm−2 and +15 Wm−2 respectively. Generally, the clouds are restricted to low level in the windward side. We also examined the association between the cloud radiative forcing and cloud physical properties such as cloud optical depth, cloud, cloud top temperature and cover amount. The cloud optical depth (−0.74 correlation value) and cloud cover amount (−0.51 correlation value) show better correlation with net cloud radiative cooling. The surface pressure of the Madagascar is also correlated with the net cooling over the windward side.


INTRODUCTION
Clouds are composed of liquid droplets and frozen ice particles suspended in the atmosphere.They form due to the saturation of water vapour in the air when the water vapour is cooled to its dew point.Clouds affect the radiation balance of the earth-atmosphere system (Rossow, 1996).Clouds block a fraction of the earth emitted longwave radiation from escaping to space and thereby causing a warming.On the other hand, clouds reflect a fraction of the incoming shortwave solar radiation back to space and thereby causing a cooling.The net effect of the clouds is either warming or cooling of the earthatmosphere system depending upon the balance between longwave warming and shortwave cooling.Cloud macro-(cloud top height, cloud cover, cloud top temperature, etc.) and micro-physical (liquid-ice phase, cloud droplet size, etc.) properties play an important role in cloud-radiation interaction (Sathiyamoorthy et al., 2011).
Influence of the clouds on the earth's radiation budget can be studied using cloud radiative forcing (CRF) parameters.CRF is computed by finding the difference between clear-sky and total-sky radiative fluxes (Charlock and Ramanathan, 1985;Sathiyamoorthy et al, 2004).Several studies made attempts to estimate the net radiative effect of clouds with satellite measured radiation budget data (Hartmann, et al., 1986;Ramanathan, et al., 1989;Sathiyamoorthy et al., 2004;2007;2011).With the advent of satellite era, it is possible to measure the earth radiation budget at the top of atmosphere.
The concept of studying the role of clouds on earth radiation budget from satellite data started from the Earth Radiation Budget Experiment (ERBE, Barkstrom, 1984) followed by Scanner for Radiation Budget (SaRaB) and Clouds and the Earth's Radiant Energy System (CERES) instruments.Kiehl and Ramanathan (1990), studied the Indonesian region during April 1985 using ERBE data and noted that there is a near balance between shortwave cloud radiative forcing (SWCRF) and longwave cloud radiative forcing (LWCRF) over the deep convective regions.Pai and Rajeevan (1998) examined the variation of net cloud radiative forcing (NCRF) in the tropical Indian ocean region using ERBE data and they suggested that the variation in cloud radiative forcing is strongly correlated with changes in high cloud amount but weakly correlated with changes in low or middle level cloud amounts.Many of the earlier studies on the NCRF in the tropics had focused their attention on the warm pool region and Asian monsoon region.Kiehl (1994) suggested that tropical deep convective clouds neither cool nor warm in an average sense, i.e., cooling due to shortwave reflection is nearly balanced by warming due to longwave absorption.But Rajeevan and Srinivasan (2000) noted that the Indian summer monsoon clouds exert a net cooling effect which otherwise is unique in the tropical belt.They suggested that large amount of high level clouds found over this region with high optical depth may be the reason for the radiative cooling.Sathiyamoorthy et al (2004) suggested that the increased amount of high level clouds over the monsoon region may be due to blowing-off of deep convective cloud tops by upper tropospheric wind shear.
When winds encounter a mountain, they may be forced to ascend and thereby produce clouds due to adiabatic cooling if the air is sufficiently moist (Grossman and Durran, 1984).They are called as orographic clouds.Atmospheric stability plays an important role in shaping the orographic clouds.If the atmosphere is stable, orographic clouds may wrap around the mountain in the windward side and it's top.On the other hand, if the atmosphere is unstable, orographically lifted clouds may grow deeper.Orographic clouds provide rainfall in the windward side of the mountain.The clouds thin out and dissipate in the leeward side of the mountain due to the atmospheric subsidence motion.Over the tropics, it is observed that orographic clouds tend to exert a net radiative cooling.No major efforts have been taken to understand the characteristics of theses clouds and their radiative forcing.The island nation of Madagascar in the southern Indian ocean has a mountain chain running across the island from south to north (Fig. 1).The mountain chain encompasses of Tsaratanana Massif in the north, Ankaratra Massif in the central parts and Ivakoany Massif in the south.During June to September months, clouds form over the eastern side of this mountain chain and adjoining east coast.In Fig. 2, visible channel imagery of MODIS onboard Terra satellite for a typical day during June-September months (10 September 2016) is shown.On this Day, clouds are seen over the east coast of Madagascar and the eastern parts of mountain chain.The leeward is free from clouds.These clouds are found to exert a net cooling along the east coast of Madagascar.As Madagascar mountain chain is isolated from nearby land or orographic features of Africa, it is an ideal place to study the characteristics of the orographically generated clouds and their radiative forcing.

DATA AND METHODOLOGY
The study region is Madagascar and its surrounding oceanic region bound between 30 • S -12 • S latitudes and 35 • E -55 • E longitudes.Global 30 Arc-Second Elevation (GTOPO30) Digital Elevation Model (DEM) data available at 30 km spatial resolution is used to study the geographical features of the Madagascar.International Satellite Cloud Climatology Project (ISCCP, Rossow et al. 1996)

Low level atmospheric circulation over the Madagascar
In Fig.  Findlater,1969;1974).As the moisture laden easterly trades encounter the Madagascar mountain chain, lifted-up mechanically over the eastern parts of the mountain chain and adjoining east coast region during June to September months.anically and form orographic clouds over the eastern parts of the mountain chain and adjoining east coast region during June to September months.

Cloud Physical Properties
The

Cloud Radiative Forcing
In Fig. 7, 9 year (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008) average SWCRF, LWCRF and NCRF during June to September months are shown for the study region.The LWCRF is about 5-10 Wm -2 over the study region.As the orographic clouds are present at lower levels and the difference in the temperature between cloud top and the surface is less, they exert weak longwave warming effects.But the magnitude of SWCRF is comparatively more along the eastern (windward) side of the Madagascar mountain chain and adjoining east coast and less over the leeward side.The SWCRF is as low as -45 Wm -2 over the windward side.As the magnitude of SWCRF is more than the LWCRF, the NCRF becomes negative.The NCRF is as low as -35 Wm -2 along the eastern parts of the mountain and the coastal area.This analysis suggests that the orographic clouds of the Madagascar exert a peak radiative cooling of about -35 Wm -2 during the June to September months.Comparison of NCRF and COD images suggest that the spatial pattern of NCRF is closely matching with the spatial pattern of COD.Both have higher magnitude over the windward side and lower magnitude over the leeward side.We speculate that the increased COD windward side may be the reason for the incresed NCRF.
In Fig. 8, the scatter plot between SWCRF and LWCRF is shown for rectangular region bounded between -23ºS to -12ºS, 47ºE to 51ºE (East box).This plot is made using monthly CERES flux data of June to September months during 2000-2008.The scatter plot suggests the existence of imbalance between SWCRF and LWCRF over the east box region.The SWCRF varies between zero to -80 Wm - 2 whereas LWCRF varies between 0 to 30 Wm -2 .This causes the imbalance between shortwave cooling and longwave warming by clouds over the east box.This imbalance leads to the net cooling by clouds.The ratio between SWCRF and LWCRF is 2.74.To further explore the relationship between COD and CRF components, scatter plots between monthly COD and CRF component (LWCRF, SWCR and NCRF) are plotted for the east box during June to September months of the 9year study period (Fig. 9).
In Figure 9    The correlation between SWCRF and COD is -0.72.LWCRF is not showing any notable association with COD as evident from the scatter plot and the flat linear fit line.The correlation between COD and LWCRF is 0.17.As the SWCRF is exhibiting association with COD, NCRF is also associated with COD.The correlation between COD and NCRF is -0.74 which is statistically significant at 99% confidence level.
So the scatter plots suggest that CRF components are related to the changes in CCA and COD over the east box.

Relative Influence of Cloud Cover and Cloud Optical Depth on CRF Components
Previous analysis suggested that both CCA and COD influence the CRF components.In this section, the relative influence of monthly CCA and COD on monthly CRF components are analysed over the east box region during June to September months of 2000-2008 period.The CRF components are plotted for different CCA and COD bins for the east box for JJAS months for the 9-year analysis period (Fig. 11).This figure suggests that the magnitude of the SWCRF increases as the magnitude of both COD and CCA increases.
Figure 1.Elevation (m) map of Madagascar prepared using GTOPO-30 DEM data 3, 9 year (2000-2008)  average low-level (975hPa) winds during June-September months over Madagascar and its surrounding (31ºE -55ºE, 8ºS-30ºS) are shown.At this level, winds are stronger at the eastern and northern parts as compared to the western and southern parts of the study area.Winds are easterly or southeasterly on the eastern and northern parts.These strong winds are part of the southern hemispheric trade winds which upon reaching east African mountain chain strengthen and cross the equator and develop as Indian summer monsoon low level jet stream ( 9-year average (2000-2008)  monthly total cloud cover amount and cloud top pressure obtained from ISCCP cloud data during June to September (JJAS) months over the Madagascar and surrounding region (now onwards referred as the study region) have been shown in Fig.4 (a-b).The total cloud cover amount is about 40-60 % over the eastern side (windward side) while it is less than 30% over the western side (leeward side).The cloud top pressure is more (i.e., height is less) over the windward side of the Madagascar mountain chain.Cloud top pressure ranges from 550 hPa -700 hPa over the windward regions whereas it is less than 550 hPa over the leeward side.Further, 9-year average low, middle and high cloud cover amounts during June to September (JJAS) months are also shown from ISCCP cloud data in Fig.5.These figures suggest that the clouds over the Madagascar during JJAS is mostly low clouds with peak low clouds (>30%) along the eastern parts of Madagascar.From Fig.3and Fig.5, it is clear that the trade winds carry moisture from the Indian Ocean and upon approaching Madagascar mountains, they support cloud formation due to mechanical lifting and associated cooling.

Figure 3 .
Figure 3. Nine year (2000-2008) average 975 hPa wind (m/s) during June-September months from ERA data over the Madagascar and its surrounding

Figure 6 .
Figure 6.Nine year (2000-2008) average (a) pressure vertical velocity (Pa/s) in shades, zonal velocity (m/s) contour and vectors generated using zonal and vertical velocities during June to September along 18ºS latitude.(b) Cloud optical depth (unit less) averaged during June to September of the nine-year study period In Fig.6b, nine year (2000-2008) average cloud optical depth during JJAS is shown for the study region.The optical thickness of the clouds is more on the windward side and low on the leeward side.On the windward side, the COD is as high as 5.5.Blocking of the trade winds carrying moisture by the Madagascar mountain chain and associated increase in cloud water droplets may be the reason behind the increases in COD on the windward side of the mountain.As the subsidence is suppressing the cloud vertical growth, cloud droplets may not be able to grow bigger due to collision and coalition process.As the smaller water droplets are effective in reflecting incoming solar shortwave radiation, they may increase the COD.

Figure 9
Figure 9 Scatter plots between monthly mean cloud cover amount and SWRF/LWCRF/NCRF are shown for the east box during June to September of 2000-2008 period.The correlation between SWCRF and cloud cover amount (CCA) is -0.63.LWCRF showing a positive correlation of 0.58.The correlation between CCA and NCRF is -0.52.

Figure 9 .
Figure 9. Scatter plots between monthly mean cloud cover amount (%) with SWCRF, LWCRF, NCRF (Wm - 2 ) from the CERES over the east box during June to September months of 2000-2016 For example, at any given CCA value, magnitude of SWCRF increases as the COD increases.Similarly, for any given COD value, magnitude of SWCRF increases as CCA increases.No clear relationship is found between LWCRF and CCA/COD.As the NCRF is the sum of SWCRF and LWCRF, it is also influence by COD as well as CCA.At any particular COD or CCA value, the magnitude of the NCRF increases if the other parameters (COD for CCA) is varied.This analysis confirms that the net cooling observed over the windward sides of Madagascar mountain chain is jointly caused by COD and CCA.

Figure 11 .
Figure 11.Variation of SWCRF, LWCRF and NCRF (Wm -2 ) at diffent COD and total cloud cover (%) bins prepared using 9 year (2000-2008) CERES and ISCCP cloud data during June-September months 4. CONCLUSION Causes of formation of low level Orographic clouds on the wind ward side of the Madagascar mountain chain during June to September months is studied using 9 year ISCCP cloud data and ERA-interim reanalysis data.When the southern hemispheric moisture laden trade winds encounter the Madagascar mountain chain, they are forced to ascend and result in cloud formation due to adiabatic cooling.These clouds are unable to grow taller due to the presence the subsidence above 700 hPa.Cloud covers and cloud optical depth are more over the windward side of the Madagascar mountain than the leeward side.These orographic clouds are exerting a net radioactive cooling as high as -35 Wm -2 .Close matching in the spatial pattern between COD and CRF components suggest that the radiative cooling is closely associated with COD.Relative influence of COD on CRF components are also examined.CCA influences all three components of CRF whereas COD influences SWCRF and NCRF.