Aerosal effect on cloud: smoke of burning in amazon decrease drop size of cloud \cite{Kaufman_1993}. In dry season of Amazon, the cloud optical property is weakly depedent on metrological variationsCloud top temperature from IR is related to cloud types.Brightness temperature and albedo of cloud could be useful for classifying cloud typesTopography and Cloud\cite{Wang_2013}circulation and the development process of thunderstorm were analyzed. Simulation results indicated that the valley breeze prevailed after sunrise and converged on the mountaintop, which might provide uplift airflow and water vapor conditions for the formation of initial convection; third, a TOPO experiment (see Table 2) revealed that the complex topography (the Yellow Mountain) played a significant role in determining the amount and locations of the precipitation. On flat terrain, the main effects were local disturbance weakening without strong topographic convergence and lifting of wind and water vapor associated with the valley wind. In contrast, the CNTL showed stronger vertical mixing, raising the moisture, increasing potential temperature disturbance, and accumulating convective instability energy, which was conducive to the growth of convective clouds; finally, other three sensitivity tests with uniform grassland (GRASS), mixed forests (FOREST), and bare soil (DESERT) were conducted. Among the impacts of land use changes, both the thermal and momentum transport were significant for the localized thunderstorm. When covered by grassland, there were less sensible heating and lower moisture, leading to the PBL height decreasing and vertical lifting weakening, which tended to cause more stable atmosphere and less rainfall on the mountaintop. When covered by mixing forests, only small dif- ferences presented in simulated meteorological fields (e.g., wind fields, moisture, cloud water mixing ratio, precipitation, and other fields). In DESERT experiment, the latent heating was more important in influencing the process of thunderstorm. There were less latent heating and lower accumulated water vapor compared to other experiments, causing vertical lifting weakening, stability of atmosphere increasing, and precipitation reducing.It shows topography and land cover contribute to cloud cover in mountains.Cloud and precipitationWhile a linear relationship between fractional coverage and rainfall amount shows considerable scatter at the smallest scale, there is much better correspondence at the larger scales, with linear correlation coefficients often exceeding 0.8 \cite{Richards_1981}.Take life cycle effect into account to study rainfall and cloud relaitonship, as there is a lag effect. Negalecting life cycle for individual cloud leads to serious errors for precipitation estimation. Cloud temperature lower than a threshold. At 0.5 degree, the cloud and rainfall sctter is poor, at 2.5 degree, it becomes stronger. \cite{Chen_1994}Cloud and Precipitaion processes in large scale model are classified as sub-grid scale mechnisms that need to be parameterized. Shallow convective cumulus, independent of deep convective cummulusDifferent types of landscape variability, such as topograpy, land-surface moisture, albedo, and vegetation, can significantly affect the local flow by redisributing solar energy into turbulent sensible and latent heat fluxes at the ground surface.A number of observational and theoretica lstudies suggest that convective rainfall seems to be associated with increases in vegetation and with with variation in surface characterstics on scales rangin from 10km to large fractions of continents.Differential heating between wet and dry land could generate mesoscale sea-breeze-like circulations. Convergence formed near the boundary between wet and dry regions and updrafts formed at the circulation front, which could result in enhanced cloud development.When a strong sea-bressze-like mesoscale circulation is generated, precipitation falls mostly in the region close to the circulation front. Otherwise, the horizontal distribution of precipitation is determined by turbulence and, therefore, appears to be random.The effect of land-surface moisture on precipitation appears to be complicated by at least two factors: (1) the intensity of mesoscale circultions and (2) the amount of water vapor available in the atmosphere. In general, the upward motion in mesoscale circulations is stronger than thermal cells induced by turbulances. From this numerical experiment, it appears that the distribution and intensity of shallow cumulus precipitation is strongly affecte by the spatial distribution of land-surface moisture. Mesoscale circulations and water vapor availablity, both dependent upon the land-surface moisture pattern, affect significantly the development of shllow convection. The distribution of precipitation appreas to be determined by the dominance of either random turbulent motions or organized mesoscale circulations.Mesoscale perturbations of vertical velocities were slightlly reduced by the moderate background wind, and perturbations of temperature remain practically unchaRossbynged. However, strong background winds larger than 1- m/s in the direction of the mesoscale flow tend to eliminate the temperature gradient generated by the wet-dry land contrast, as a result, the associated mesoscale circulation.Strong background wind increases the turbulent sensible heat flux due to increased mixing. Depending on the atmosphric conditions, a significant variation in the land-surface moisture can produce either an increase, a decrease, or almost no change in the simulated cloud amount. This implies that if the entire land surface is relatively dry, the presence of a moist atmosphere does not necessarily increase finall. In a moist atmosphere, the maximum rainfall is obtained in the case where the land surface is a wet land. In general, precipitation rate is primarily determined, for a given atmospheric condtion, by two factors: (1) the intensity of upward velocity and (2) the local supply of water vapor. The vertical drafts can be either initiated by the random turbulent eddies or created in the convergence region of organized mesoscale circulations at the circulation front. It is interesting to note that the maximum global mesoscale heat flux and precipitation first increase iwth the wavelength of dry land, reaching the largest varlues in the 100-km-wide dry land case, and decrease with a further increase of the size of the dry patch. The local Rossby radius of deformation varies between 80 and 140 km (depending on the time of the day and the elevation in the PBL) in our simulations. indicated that the largest mesosclae circulation should be obtained for a wavelength of forcing corresponding to the local Rossby radius of deformation.\cite{Heiblum_2013}Daytime FCu cloud fields similar to those seen in Figs. 1 and 2 can be observed in several other locations around the globe, such as central Africa during most of the year or northeast America and Siberia during the boreal summer. The common denominator in all cases is their preferred formation over dense, large-scale forests during stable meteorological conditions, when formation of more developed clouds is suppressed.Hence, the evapotranspiration properties of the land cover vegetation are tightly linked to the dynamics of the boundary layer and the shallow Cu clouds, which com- monly cap the boundary layer.Deforested areas in the Amazon (either pasture or crop- land) usually display higher sensible heat and lower latent heat fluxes in comparison with the forested areas, which in turn can enhance the growth of the boundary layer during the day and favor the formation of larger convective clouds (Fisch et al., 2004). Moreover, surface heterogeneities often result in local mesoscale breezes which can also affect low- level convergence patterns and cloud formation (Rabin et al., 1990; Souza et al., 2000).Generally there exists a preference for shallow Cu formation over densely forested rather than deforested areas (located around the southern boundaries of the Amazon Basin; see Fig. 1) (Cutrim et al., 1995). However, most observational studies, including those listed above, have focused exclusively on deforested pockets within forested areas, showing a clear preference for shallow Cu formation over deforested areas (Cutrim et al., 1995; Chagnon et al., 2004; Wang et al., 2009) and for deep convective cloud formation over the surrounding forested areas (Wang et al., 2009). Moreover, studies showing preference for shallow Cu over deforested land cover are confined mainly to the southwest regions of the Amazon, which are highly deforested and experience very stable meteorological conditions during the dry season.Using NOAA- NCEP Global Data Assimilation System (GDAS) analysis data (Saha et al., 2006; Parrish and Derber, 1992), we exam- ined the spatial patterns of various meteorological parame- ters (J-A-S averages). The two parameters that were found to best reflect the spatial variance of the FCu fields are the geopotential height (HGT) at 700 hPa (see Fig. 5a) and the relative humidity (RH) at 850 hPa (see Fig. 5b). High geopotential height at 700 hPa (pres- sure levels of850–500 hPa give similar results) indicates upper level subsidence, adiabatic warming and drying, and is associated with the SASH (Figueroa and Nobre, 1990). Relative humidity at 850 hPa corresponds to the mean cloud base height (based on ceilometer measurements), and is essential to cumulus formation.Previously discussed cloud features such as FCu field preference over the northern part of region, and preference for deep convective clouds in high RH areas (i.e., coastal and northwestern areas), are seen for all four years. The relatively high chance for deep convective clouds in the northwest part of the study region may be due to variable terrain and com- plex topography in that region as well (see Fig. 1). Water bodies such as the Amazon River, Atlantic Ocean, and lakes clearly inhibit all types ofcloud formation. The southern part of the region experiences little or no clouds at all, due to the dry and stable meteorology in that area. For the forest land cover, we see a positive dependence ofpFCu on EVI. This dependence is especially strong for the lower EVI values. The increase in pFCu then saturates at a moderate value ofabout EVI=0.54. For the non-forest land cover, the dependence of pFCu on EVI is somewhat different. For low EVI values (EVI<0.48), there is a strong positive dependence (similar to that seen in forest land cover), but for higher values of EVI >0.48 there is a clear decrease of pFCu with EVI.As seen in Fig. 7b for the NA region, pFCu over forest land cover is lowest close to the boundaries with other land cover types, then increases sharply with large variance at short distances (<5 km), and is relatively constant at distances larger than ∼10 km.We divided the data into two subsets: (i) forest data within 10km of other land cover types (Fig. 7c), which we assume in- cludes the bulk of mesoscale circulation effects, and (ii) for- est data further than 10km away from other land cover types (Fig. 7d), which we consider to be free of mesoscale circu- lation effects., ..., Hence, we can assume that in transition zones between forest land cover and water/non-forest land covers, a superposition of deep forest and non-forest depen- dencies on EVI takes place. Generally, for a given EVI value the further the distance from water/non-forest land covers, the higher the chance of observing FCu fields., ..,. Clearly, the well-being and productivity of the deep forest promotes the formation ofFCu fields.We can speculate that the FCu fields correspond to a specific solution of Rayleigh–Benard thermal convec- tion over land (or specifically cloud streets, as discussed in Sect. 1), since the basic physical settings are similar over the Amazon and ocean surfaces, namely a homogeneous warm surface and a moist boundary layer with a well-defined inversion layer. Hexagonal open convection cells have al- ready been simulated over tropical land in the western Pacific (Saito et al., 2001). The fact that vegetation properties control both surface fluxes and boundary layer depth (h) to a large degree, and that the Rayleigh number (Ra) is highly dependent on that depth (proportional to h4), suggests a physical link between forest and the cloud fields formed above.Vegetation and cloud\cite{Cutrim_1995}It may be counterintuitive to expect more shallow cumulus clouds over forested lands (e.g., Salati and Vose 1984), but if deforestation actually increases the coverage of shallow cumulus clouds, as was observed following the harvest of wheat in Oklahoma, in Rondonia during the dry sea- son the increase in surface albedo that accompanies deforestation (e.g., Bastabel et al. 1993) could be augmented by a cloud-forced increase in atmospheric albedo. In this work, we tested the hypothesis that the clearing of the rain forest in Amazonia tends to increase local daytime cumulus cover.The algorithm runs on the University of Wisconsin's Man-computer Interactive Data Access System (Suomi et al. 1983). It applies a threshold of relative albedo, derived from the visible image (Garand 1988), to distinguish clouds from the land or water surface. Next, it applies a threshold of temperature, derived from the companion infrared image, to distinguish deep convective clouds and middle or high clouds from shallow clouds. Then the algorithm applies a minimum threshold to spatial contrast in albedo, which it calculates along scan lines of the visible image, to distinguish cumulus clouds from stratus clouds. Second, overthe period of a month the land surface imposes a strong signature on daytime cumulus clouds. Third, in some situations surface water and vegetative cover may be more important than elevation in controlling the development of cumulus clouds. Fourth, evidence is emerging that in Rondonia by 1988 deforestation was causing an increase in dry-season afternoon fair-weather cumulus clouds. To our knowledge this is the first time such evidence has been presented.our observation of more shallow cumulus clouds over a deforested surface would sug- gest numerical and observational analysis of the co- gent processes and more realistic treatment of boundary layer cumulus clouds in simulations using general circulation models. Finally,\cite{Rabin_1996}Apart from the suppressive effecto f the Great Lakes, early observation osf portions of the MississippBi asin[ e.g.,O livera nd Oliver,1 963;A ndersone t al., 1966, pp. 34-36 and 169-170] implied that relatively gentle topography can affect the distribution of these clouds.A minimum threshold of albedo distinguishes clouds from most land surfaces. A minimum threshold of temperature, roughly corresponding to the daytime height of the planetary boundary layer, excludes deep convective, middle, and high clouds. Change in albedo along a scan line distinguishes a cumulus cloud by spatial coherence from a layer cloud such as stratus. A "warm" cloud pixel is considered to be cumuliform if albedo changes by 10% or more along the three-pixel line segment ending with the cloudy pixel. Maximum thresholds of albedo and temperature can be used to exclude some warm and bright surface features, such as sand dunes or sun glint. In the algorithm all five thresholds are variablesOne of the most notable features in Figure 2 is the nearabsence of cumulus clouds over Lake Michigan. Enhanced cumulus frequency along the shoreline accentuates the lake "hole." Values range from less than 2% over the lake to as high as 8% near the shore when averaged over 0.5 ø x 0.5 ø latitude-longitude grid cells. Animation of images of cumulus frequency from hour to hour reveals an expansion of the cumulusr ing aroundt he lake, especiallyo n the easts ide.Weaker minima lie over smaller water bodies such as Wisconsin's Lake Winnebago, Oklahoma's Lake Eufaula and certain rivers (e.g., the Mississippi on the western border of Illinois, the Arkansas and the Tennessee in northern Alabama). In the summer during the day, water bodies would tend to conduct less heat to the atmospheret han surroundingla nds( e.g., seeC hena ndA vissar [1994]).Large bodies can foster lake breezes[ Eichenlaube t al., 1990; Simpson, 1994], which would reinforce the suppressive effect of diminished sensible heat flux over the water surface and could generate a ring of on-shore cumulus clouds. In aggregate, many strong lake breezes can yield an expanding cumulus ring like that described above.Noting a lack of clouds over the Missipine delta in individual satellite scenes, Oliver and Oliver [1963] suggested that differential friction between the plain and highlands may generate sinking air near the plain. Gibson and Vonder Haar [1990] observed cloud minima in other valleys of the south- eastern United States, mostly smaller than the delta, and at- tributed them to upslope circulations forced by differential heatingIn general, cumulus frequency tended to decrease as NDVI increased. The decrease is stronger in the drought year, 1988. Simulations [e.g., Rabin et al., 1990; and Chen and Avissar, 1994] suggest that this difference could have resulted in part from a drier atmospheric boundary layer in 1988.It is this range of NDVI, where vegetation is more sparse, that a dry upper soil layer (such as in July 1988) leads to elevated surface temperature and sensible heat flux and reduced evapotranspi- ration (e.g., see Nemani et al. [1993]). Indicative of the increased spatial contrast of the surface fluxes during July 1988, the variance in monthly mean clear sky temperature (inferred from GOES 7 across the Illinois Plain subregion) exceeded that of July 1987 by 40%.Second, over the period of a month the land surface within the Mississippi Valley imposes a distinctive signature on day- time cumulus clouds. Cumulus frequency is modulated by even modest elevation features. Locally, slope and aspect may also modulate cumulus frequency. Overall, cumulus frequency tended to peak early in the afternoon. Third, at times, soils and vegetative cover may be more important than elevation in controlling the development of cumulus clouds. This was the case over southern Illinois during the drought month of July 1988. There, cumulus frequency tended to be inversely associated with plant cover and plant- available soil moisture.\cite{Gibson_1990} Cloud variations in the Southeast USCloud formed by strong convection have high, cold tops. These cumuliform clouds relect considerable light and are, therefore, bright in the visible specturm. By way of contrast, clouds formed by weak convection, though they may be bright, have warm cloud-top temperatures. Additionaly, cirus anvil clouds, though they may have cold tops, relect less light as compared to an overshooting cumlonimbus top. Using these characteristics, we determined convective cloudiness by using a combination of brightness of the visible image and the cloud-top temperatures as indicated by the brightness of the infrared image.Vertical motions, and therefore convergent flow of the air mass, are essential to initiate and maintain convective activity. However, convergent flow must be offset by subsiding air and divergent flow in other locations. The waters of even the smallest bays and inlets of the Gulf of Mexico have cloud minima relative to the influrnce of many rivers emptying into the Gulf.Meso r scale (2-20km), and micro alpha scale (200m to 2km)These satellite data show evidence of upward vertical motions along the entire shorelin with the maximum located over peninsulas and minium over the near-shore areas adjacent to bays and inlets. During those times when a sea-to-land breeze occurs, divergence development is limited ot water-covered areas and is most noticeable over bays and inlets. The increase in cloudiness along the entire shooreline suggests that the sea breeze induce convergence over the land areas and seldom, if eve, causes divergence, regardless of whether the breeze is associated with peninsulas or land areas adjacent to inlets. The sea breeze-induced coverence ins the primay effect and shore line shape is the secondary effect. Thus, shoreline shape modulates the sea breeze-induced convergence, but not to the extent that it results in divergence. Note too, that all bays and inlest, regardless of size, show a strong cloud minimu during daylight hours,THe magnitude of the frequency of deep convection depends on the shape or the coastline. There is a sclae relationship between coastline configuration and the frequency of convection. Convergence development over small peninsulas due to the sea breeze is not strong enough to produce deep convection, but it is stronger over larger peninsulas. The frequency of cumuliform clouds can be enhanced by vertical motions of scales that at least as large horizontally as the clouds. Small peninsulas slightly enhance vertical air motions and therefore enhance only clouds of small horizonal extent. Large peninsulas enhance vertical motions of a larger scale and therefore enhance of the frequency of cumuliform clouds of both larger horizontal and vertical extent.As the day beings the cloudy areas are over low terrain and over the Gulf. Areas which are favorable for cloudiness later in the day are over the land that is adjacent to bodies of water and over steep or high terrain. The valley-to-mountain breeze induces divergent flow and consequent suppression of cloudiness over the adjacent lower terrain. When water temperature is less than the air temerpature the near-surface air layer is stabilized and clouds do no develop. Small bodies of water such as narrow rivers and small ponds show no significant effect on cloud formation while larger rivers and larger lakes can be important. This is due to the sclae of the individual cloud elements as compared to the water body. The fresh bodies of water in the study area were small compared to convective cloudiness and, therefore, had no evident effect on deep convection. Even the smallest inlets and bays adjacent to major water bodies suppress cloud development over the water, but sea breeze-induced cloud maxima appears over the adjacent land. Small peninsulas enhance cloud development associated with sea-breeze convergence, but peninsulas must be relatively large in order to have a significant effect on convective activiety. Relating vertical wind shear and atmospheric stability to the frequency of convection of the various intensities should be particularly productive.\cite{Lynn_1998} Cloud and land patch size, on deep moist convection.It is not surprising that observations suggest the existence of landscape- generated mesoscale1 circulations associated with land- scape patches defined by these discontinuities (Voron- stov 1963; Burman et al. 1975; Rabin et al. 1990; Bou- geault et al. 1991; Mascart et al. 1991; Kimura and Kuwagata 1993; Mahrt et al. 1994; Cutrim et al. 1995; Doran et al. 1995; Koch et al. 1997). These circulations are generated by contrasts in surface heat fluxes that can occur, for example, between relatively wet and dry ground.Analytical and numerical studies have shown that landscape-generated mesoscale circulations can assume the form ofstrong sea breezes [e.g., Anthes (1984); Dalu et al. (1991); Enger and Tjernstro¨m (1991); Pielke et al. (1991); Avissar and Chen (1993); Chen and Avissar (1994a); Doran et al. (1995); Lynn et al. (1995a)], and that circulations associated with patches with length scale similar to the local Rossby radius of deformation are most intense. Most importantly, observational results suggest that mesoscale circulations generated by landscape patches can affect the formation of shallow clouds (Rabin et al. 1990; Cutrim et al. 1995).There was a 256-km dry patch in the center of the domain in experiment 1 (Fig. 3) adjacent to two 128-km wet patch- es; noting that the lateral boundary conditions were pe- riodic, it was surrounded by a wet patch of the same size. The domain of experiment 2 consisted of two dry and wet patches, each 128 km in length (Fig. 4). In experiment 3, experiment 4 (Fig. 5), experiment 5, and experiment 6, a random number generator was used to specify the length of individual patches, subject to the constraint that no patches were more than twice the average. The development of mesoscale circulations in re- sponse to differential heating of the atmosphere by land surface patches has been discussed in detail in prior work (e.g., Pielke et al. 1991; Avissar and Chen 1993; and Lynn et al. 1995a). For this reason, it is noted only that mesoscale circulations formed in experiments 1–6 along the edges of the dry land surface patches, when these patches were greater than about 10 km. These circulations, which resemble sea-breeze-like fronts, moved inward over the center of the dry patches, interacting first over the smallest of the patches, and then over the largest. Thus, the timescale of these circulations was, in general, least over the smallest patches and big- gest over the largest patches. It varied from about one- third of a day over the smallest patches to more than half a day over the largest patches.This was because there was a collision of sea-breeze-like fronts over the 128-km patches in experiment 2 at the optimum time (when the convection was most strongly forced by the sea-breeze- like fronts), which led to relatively intense updrafts and the formation ofdeep clouds (not shown). The magnitude of the mean wind was also important. The weakest and most discrete rain cells occurred in experiment 9, in which there was a very slow propa- gation speed of about 1.7 m s⫺1. In contrast, more intense and longer lasting rain cells were obtained in ex- periment 10. In experiment 10, the propagation speed was about 5.6 m s⫺1 , and as explained by (Chen et al. 1994): flow over the cold pool along sea-breeze-like fronts, and hence lifting along them, is more significant at larger propagation speeds.In these studies, it was found that the strongest mesoscale circulations and rainfall are generated over patches similar in size to the local Rossby radius of deformation. This is because the forcing by the land surface corresponds well with the preferred scale of the circulations, as determined by the local Rossby radius.Thus, the results obtained in this study seem to point to the importance of ro on deep moist convection when this convection is forced by local contrasts in the surface fluxes (patches). It indicates the scale at which an in- teraction among opposing sea-breeze cells is likely to occur at an optimum time, leading, perhaps, to robust, deep, moist convection (e.g., experiment 2).Moreover, background wind also had an important impact on the rainfall obtained at any one patch size. Clearly, the standard deviation obtained at any patch length was relatively large when compared to the mean rainfall at that patch length. In addition, tabular results (not shown) indicated that there was no simple rela- tionship between the background wind profile and the mean rainfall. These results were in contrast to those obtained by Chen and Avissar (1994b) for shallow convection; they found that accumulated rainfall was largest when the background wind was small.In experiments 1–5, between 66% and 75% of the rainfall occurred in association with CLT-type clouds. This was because of the direct forcing ofthe sea-breeze- like front on these clouds. The highest percentage of CMT-type clouds was obtained in experiment 2. This was because of the collision ofopposing sea-breeze-like fronts over the center of the dry patch at the optimum time (see above). In fact, the percentage of CMT was higher in experiments 3–5 than in experiment 1, owing also to a collision of sea-breeze-like fronts in them. However, the percentage in each was less than in ex- periment 2 because the convection occurred before the optimum time.The largest CAPE evolved above the wet patch in experiment 1, while the smallest CAPE evolved over the dry patch. Thus, the largest potential for deep convection occurred over the wet ground, as is consistent with the results obtained in a one-dimensional study by Segal et al. (1995). How- ever, the most rainfall did not occur where the CAPE was largest. Instead, the heaviest rainfall occurred along the sea-breeze-like fronts—where the CAPE was of in- termediate values.\cite{Pielke_2001} review paper, contain useful equtions Changes in land-surface properties are shown to influence the heat and moisture fluxes within the planetary boundary layer, convective available potential energy, and other measures of the deep cumulus cloud activity.Segael ta l. [1995]s howed th t with the same value of RN, with a smaller Bowenr atio, the thermodynamic potential for deep cumulusc onvection increaseSincel and-waterc ontrastsp ermit the developmenot f sea breezes, which focus thunderstorm development overi slandsa nd coastalr egionsin the humidt ropicsa nd in humid middle and high latitudes during the summer [e.g., see Pielke, 1984, Figure 12-13], it would be expected that similarv ariationsi n surfaceh eatinga ssociatedw ith landscappea tternsw oulda lsop roducem esoscalcei rculationso f a similarm agnitudeO. ther papersw hichd escribe the influenceo f seab reezeso n thunderstormasr e thoseb y Pielke [1974] and Pielke et al. [1991b].Dalu et al. [1996] used a linear model to conclude that the Rossby radius defined in (A41) is the optimal spatial scale for landscapeh eterogeneitiest o produce mesoscalefl ows.A vissara nd Pielke [1989],H adfielde t al. [1991], Shen and Leclerc [1995], Zeng and Pielke [1995a, 1995b], Wang et al. [1997], and Avissat and Schmidt [1998] also explored the issue of the size of landscape patchiness that is needed before the boundary-layer structurei s significantlya ffecteda nd a mesoscalec irculation is produced. Consistent with these conclusions, Segal et al. [1997] found that cumulus clouds are a minimum downwind of mesoscale-size lda kes duringt he warm season as a result of mesoscale-induced subsidenceo vert he lake and the resultants uppressioonf zi.Pan et al. [1996] concludedt hat increasesin soil moisture enhanced local rainfall when the lower atmospherew as thermodynamically unstablea nd relatively dry but decreased rainfall when the atmosphere was humid and lacked sufficient thermal forcing to initiate deep cumulus convection.These results illustrate that the effect of landscape evaporation and transpiration on deep cumulonimbus convection is quite nonlinear. These opposing effects further explaint he apparentc ontradictionb etweent he results reported by Lyons et al. [1996] and Pielke et al. [1997], as discussede arlier in this section.While increased moisture flux into the atmospherec an increase CAPE, the triggering of these deep cumulus clouds may be more difficult since the sensible heat flux may be reduced. The depth of the planetary boundary layer,f or example will be shallower if the sensibleh eatf lux is less. Other studies which explore how vegetation variations organize cumulus convection include those by Anthes [1984], Vidale et al. [1997], Liu et al. [1999], Souza et al. [2000], and Weaver et al. [2000].D.irmeyearn dZ eng[ 1999] concluded that evaporation from the soil surface accounts for a majority of water vapor fluxes from the surface for all butt hem osth eavilyfo resteda reas, where transpiration dominate.\cite{Wang_2000}The simulation results suggest that the synoptic forcing, in terms of atmospheric stability and background horizontal wind, dominates during the rainy season; synoptic conditions were so favorable to moist convection that the added effect of surface heterogeneity was negligible. During the dry season, a noticeable impact of mesoscale circu- lations resulting in enhancement of shallow clouds was simulated; the mesoscale circulations also triggered scattered deep convection that altered the spatial distribution of precipitation.Deforestation at smaller scales (250 km) may reduce local rainfall but may not affect the large-scale circulations. This result was confirmed by the modeling study of Eltahir and Bras (1994), which investigated the climate impact of deforestation in the Amazon over a limited region. At even smaller scales (25 km) deforestation is expected to trigger mesoscale circulations with rising motion over deforested areas.The linear theory of Wang et al. (1996) predicts that stable stratification strongly inhibits the development of the mesoscale circulation and hence drastically reduces mesoscale transport of momentum, heat, and moisture. Mesoscale fluxes, in the linear analysis, are important only in an environment ofneutral stratification and weak synoptic wind. The intensity of the mesoscale circula-tion is shown to be proportional to the standard deviation of the turbulent sensible heat flux at the surface. In the lower atmosphere, the thermal variability of the landscape at the smaller length scales is more efficient in triggering dry convection, and, at higher altitudes, the atmospheric dynamics are more sensitive to thermal forcing at longer length scales. The atmosphere from bottom to top behaves as a low-pass filter to the thermal variability of the landscape.In general, the numerical solutions of nonlinear mesoscale circulation show qualitative agreement with those from the linear analytical the- ory. The stability of the atmosphere was identified as the dominant hurdle to the forcing of mesoscale cir- culations by land surfaces with random heterogeneity. The mesoscale circulation could reach 5-km height in an atmosphere with neutral stratification but was suppressed down to a layer not deeper than 200 m near the surface in a stable atmosphere.The results of Wang et al. (1998) indicate that land surface heterogeneity with length scales near a few tens of kilometers is effective in forcing mesoscale circulations. The form of the distributions of the surface heterogeneity is not critical as long as it is characterized with similar variability and length scale.Their findings include 1) the land surface imposes a strong signature on daytime cumulus clouds over the period of a month, 2) surface vegetation cover could be more important than elevation in controlling the development of cumulus clouds, and 3) the defor- estation in Rondoˆnia was causing an increase in dry- season afternoon shallow cumulus clouds. In a midlat- itude environment, Rabin et al. (1990) presented a case of significant impact of spatial variation in landscape on cumulus cloud formation over a flat terrain in a rel- atively dry environment. They further studied the atmospheric conditions under which the role of land sur- face was important. They concluded that the effect of heterogeneous land surface on shallow cumulus clouds is noticeable when the synoptic forcing (fronts and other large scale disturbances) is weak.Moist convection in the atmosphere results from processes over a wide range of scales. Synoptic low pres- sure disturbances (⬃103 km) are responsible for exten-sive cloudiness and precipitation. At local scales (⬃100 km), diurnal development of turbulent flow in the planetary etary boundary layer could also trigger moist convec- tion, leading to scattered cumulus clouds. Mesoscale circulations induced by land surface heterogeneities, oc- curring at a scale in between (⬃101 –102 km), arguably may provide an additional mechanism for triggering moist convection. The mesoscale circulation mechanism is highly dependent on the magnitude and length scale of the heterogeneity of the land surface as well as on the synoptic conditions. In an absolutely stable atmosphere, mesoscale circulations are inhibited; hence, they make no contribution to triggering of moist convection. In a conditionally unstable atmosphere, in which there is a small energy barrier near the surface, the turbulent flow within the planetary boundary layer can trigger moist convection even over a homogeneous land sur- face. Under the above conditions, the mesoscale circulations play little role in triggering convection. Three climatic regimes have been identified over Rondoˆnia, Amazonia. A typical dry-season day is char- acterized by a stable sounding profile with little convective available potential energy (CAPE). The wet (rainy) season in the region is characterized by an unstable sounding with relatively large CAPE and small energy barrier. Between the dry and wet seasons, there is a period of time over Rondoˆnia when the synoptic disturbances are relatively weak. This transition time is called break period. The basic features of the mesoscale heat flux are dif- ferent from the corresponding turbulent heat flux, par- ticularly in its vertical distribution. Unlike the turbulent heat flux, which has a maximum at the surface and decreases upward, the mesoscale heat flux increases with height from zero at the surface, reaches a peak at a certain level (depending on the atmospheric stability profile, usually at about 1 km, or the 900-hPa level), then decreases to a negative maximum with a smaller magnitude, eventually vanishing with height. It also has a well-defined periodicity in time that follows the di- urnal heat cycle of the land surface. These features will be used to detect the signals of mesoscale circulations and to describe its role in triggering convection.Apparently, the development of PBL did not trigger convection. Nevertheless, there was convincing evidence that mesoscale circulations were induced over the deforested domain. Lack of afternoon convection indicates that the mesoscale circulations were not strong enough to overcome the energy barrier in the sounding (Fig. 4) to trigger convection. The sounding profile shows that the most unstable (i.e., high- est equivalent potential temperature) level is at ⬃840 hPa. The mesoscale circulations, however, were weak near this level (Fig. 3b), at which the mesoscale heat flux was vanishing.This experiment was repeated using a weakened synoptic forcing by reducing the magnitude of the synoptic wind by an arbitrary factor of 10 (the large- scale convergence and/or divergence was reduced ac- cordingly). The corresponding simulated rainfall and afternoon cloudiness are shown in Figs. 5 and 6. The rainfall patterns in the deforestation and control runs were qualitatively similar, although the locations of the sporadic rain cells were somewhat different. Note that they did not resemble the land cover patterns. Enhanced cloud was simulated over the deforested locations (Fig. 6b), however. The clouds were observed at levels between 900 and 800 hPa. Enhanced cloudiness over the deforested areas also occurred in the afternoon of the second and third simulation day (results not shown). The mesoscale circulations mea- sured by the the mesoscale heat flux (Fig. 7) were stronger under this weakened synoptic wind condi- tion.The enhanced afternoon cloudiness con- sisted of low-level cumulus clouds at the 800-hPa level. The low-level cumulus clouds produced little precipitation, as seen in Fig. 5, which shows that the sparse rainfall cells do not overlap with the locations of the clouds (Fig. 6). These results suggest that de- forestation may enhance afternoon cloudiness but contributes little to local rainfall. Rain season: Active deep convection and in- tensive rainfall were obtained in both the deforestation and the control runs. The distributions ofthe high clouds and rainfall from the control and deforestation runs were nearly identical. Shallow clouds did not occur in the rainy-season simulations. Moreover, they had no simi- larity to the patterns of land cover. Variability in land cover condition clearly played no role in shaping the distribution of cloudiness and rainfall.The reduction in the cloudiness and rainfall presumably was due to the weakened large-scale ascending motion ofthe air and reduced con- vergence of moisture.Dry season: In particular, precipitation from the deforestation run (Fig. 18b) had a broader coverage than that from the control run (Fig. 18a) during the first afternoon, when localized deep convection occurred over the deforested region and led to precipitating high clouds (discussed below). The organization of the rainfall (Fig. 18b) re- flects a land cover signature with the locations of rain cells slightly shifted toward the northwest relative to the deforested areas because of the prevailing wind. As in the case of the break period under weakened synoptic wind condition, an enhancement of shallow cumulus clouds (near the 800-hPa level) (Figs. 19 and 20) was simulated in the afternoon of the first day of simulation over the deforested areas. Enhanced high cu- mulus clouds (near the 300-hPa level) (Fig. 21) were also simulated over the partially cleared areas, reflecting the pattern of rainfall shown in Fig. 18b. These precip- itating high cumulus clouds, contrary to the nonprecip- itating shallow clouds, contributed local rainfall (Fig. 18b). The majority of the en- hanced shallow clouds were simulated over the defor- ested patches as expected, following closely the patterns of land cover. They were short-lived clouds, lasting about 4h starting from 1400 LT on the first day, and from 1500 LT on the third day.\cite{Durieux_2003}Another approach consists in analysing the cloud cover, given that long-time series of satellite observations are now available. Using the monthly mean outgoing long-wave radiation (OLR) data from the NOAA polar-orbiting satellites and monthly rainfall totals at Belem and Manaus in the past 15 years, Chu, Yu, and Hastenrath (1994) found indications of a slight rainfall increase (related to an increase in convection) associated with deforestation over almost all of the Amazon basin (see Fig. 1). Also, modelling simulations by Avissar and Liu (1996) and observations during the ABRACOS experiment by Dias and Regnier (1996) show that the surface inhomogeneity caused by the interruption of the tropical rain forest by patches of pasture induces differences in the surface sensible heat fluxes. These differences can affect surface fluxes and generate mesoscale circulation in (and even above) the boundary layer, which in turn may trigger cumulus convection and increase the recycling of water (Dalu, Pielke, Baldi, & Zeng, 1996; Pielke et al., 1998). Observations in South Amazonia have shown that defores- tation can result in an increase in dry season afternoon fair weather cumulus clouds (Cutrim, Martin, & Rabin, 1995). Thus, the increase in surface albedo that accompanies deforestation could be augmented by a cloud-induced increase in atmospheric albedo.we selected the following parameters: total cloud cover (TCC), cloud top pressure (CTP), cloud vertical optical thickness (VOT), high-level cloud cover (HCC), middle-level cloud cover (MCC), and low-level cloud cover (LCC). Combining cloud top pressure and optical thickness makes it possible to distinguish cirrus clouds (high cloud top pressure and low optical thickness), very important in the radiation budget, from convective clouds (high cloud top pressure and high optical thickness), responsible for most ofthe precipitation in the tropics. High- level clouds are statistically related to the convection and will be used as a proxy for the convective clouds variability. TCC is a combination of three cloud types that are defined by the range of the cloud top pressure: low (CTP>680 hPa), middle (680>CTP>440 hPa), and high (CTP< 440 hPa). The difference between TCC and HCC gives the low-level and middle-level cloud cover representing the majority of the cloudiness but very weakly linked to the precipitation.The TCC minima are associated with El Nin˜o Southern Oscillation (ENSO) events whereas TCC maxima are associated with La Nin˜a Southern Oscillation (LNSO) events.The studied deforested areas always have less cloud cover and less precipitation than the reference areas on the interannual time scale; however, this can be explained by climatic differences due to their different locations.For all pairs and all criteria, these data are consistent, with a dry season climate more affected by deforestation than the wet season climate.Below a deforestation threshold (typically 10%), cloud cover does not show any significant differences with respect to that of a forested region. Above this threshold, some changes appear during the dry season with a significant decrease in convection, together with a significant increase in low-level clouds (mainly made up of shallow cumulus clouds). Above a second deforestation threshold (around 20%), the climate change also concerns the wet season with a significant increase in convection while convection during the dry season continues to decrease and the preponderance of shallow cumulus clouds stabilizes. Above this last threshold, the climate appears to show more seasonal contrasts.\cite{Ray_2003} Southwest AustraliaBased upon AVHRR data near 1430 LT, Lyons et al. [1996] found that sensible heat fluxes are larger over native vegetation than over agricultural regions in all seasons; latent heat fluxes were higher over native vegetation during summer and over agriculture during winter. Lyons et al. [1993] suggested the higher sensible heat fluxes and the resultant vigorous mixing in the boundary layer to be responsible for the observed preferential cloud formation over native vegetation. On the other hand, the aircraft observations during late September-early October show that latent heat fluxes are significantly larger than sensible heat fluxes during daylight hours over agricultural areas (Bowen ratios usually >1.5); over native vegetation latent heat fluxes are greater than sensible heat fluxes in the morning and late afternoon hours, but then smaller than sensible heat fluxes (Bowen ratios often <0.67) during the heat of the day.The cloud detection technique used in this investigation specifically targets boundary layer cumulus cloud fields, which in general are strongly coupled to local surface processes. Varying surface conditions on the two sides of the bunny fence should result in different cumulus cloud fields, but not upon large-scale frontal cloudiness. The cloud detection scheme used in this study is effective in eliminating the larger cloud systems associated with frontal activity. To determine the cloud number density, the segmentation technique of Kuo et al. [1993] is applied where unique identifying numbers are assigned to pixels belonging to the same cloud. The segmentation algorithm also calculates the location of the individual cloud centroid. Details are given in the work of Nair et al. (submitted manuscript, 2002).However, by 1500 LT cumulus clouds cover nearly all of the native vegetation (Figure 6d), while the barren agricultural areas remain cloud free. There is a complete reversal of cloud cover patterns between August and December–January for the agricultural and native vegetation regions. This clearly demonstrates that land use contributes to the development of cloud cover in these regions. Recall that these regions have no other differ- ences, other than vegetation cover. The differences in topography are small, and there is no evidence of frontal activity in these cases.During the summer months ofDecember and January, cumulus clouds occur with higher frequency and are spa- tially more extensive in occurrence over native vegetation during the afternoon hours (Figure 8).It shows that during the summer months of December 2000 and January–March 2000 cumulus clouds occur with higher. During the winter (agricultural season) months of May–September, cumulus clouds occur more frequently over agricultural areas. \cite{Chagnon_2004} Two statistics that relate shallow cumulus cover to deforestation are examined: the monthly average difference in fractional cloud cover between the deforested and the forested surfaces (hereafter referred to as cloud density difference (CDD)); and the number of days in a month when the fractional cloud cover over the deforested surface exceeds that over the forested surfaces (hereafter referred to as the exceedence count (EC)).Thus, there is an overwhelmingly significant association between shallow cloud cover and deforestation patterns; this association is a symptom of persistent change in the climatology of shallow clouds.we find here that the results are slightly modulated by the lower-tropospheric (i.e., 700–850 mb layer-average) wind speed at 14h45 UTC, in that they occur more often under calm conditions (Figure 3c). However, significant enhancements of shallow clouds still occur under strong wind conditions at 14h45 UTC, and the frequency of significant shallow cloud en- hancement is independent of wind speed at 17h45 UTC and 20h45 UTC (figure not shown). Strong wind shear has a suppressing influence on the cloud contrast at all times of the day (Figure 3d). Still, significant cloud contrasts occur under strong background wind shear, which shows the resilience of the phenomenon.For all seasons, we observe a monotonic decrease in the frequency of monthly significance as previous wetness increases. Rainfall tends to equalize the thermal contrast created by deforestation; under dry conditions the trees have a greater access to water due to their deep roots and can thus continue to transpire while the deforested areas dry out more quickly and thus heat up. Thus, the spatial distribution of shallow clouds seems to be bimodal: one peak over topographical features and another over forest cover feature. Conversely, the mesoscale circulations created by defores- tation may serve to trigger precipitation from these shallow clouds, thereby potentially cancelling the effect of biomass burning aerosols.\cite{Sato_2007}\cite{Wang_2009} It is known that this type of land cover change causes land breeze that may enhance cloudiness (15–18), leading to the notion that clouds follow deforestation. Past works by the authors (16) and others (17) have shown that these ‘‘land breezes’’ are effective at scales of heterogeneous land cover on the order of 1-10 kilometers.Shallow clouds are defined as those with cloud-top brightness temperature (CTBT) >280 K, and deep clouds with CTBT <240 K. Out of the 103 months, 86 months had more shallow clouds over deforested surfaces than forest surfaces, and 29 months had more deep clouds over deforested surfaces than forest surfaces. It is evident that the atmosphere over the forest surfaces was consistently more (conditionally) unstable than over the deforested surfaces, i.e., a greater CAPE for the forest site than that for the pasture site. Meanwhile the values of CINE at the 2 sites are comparable. The net radiation was at approximately the 500W m⫺2 level at the forest site, significantly higher than the approximately 350Wm⫺2 at the pasture site. Latent heat flux was at approximately the 150W m⫺2 level at the pasture site, whereas it was at the approximately 340W m⫺2 level at the forest site. The measurements from the surface stations (Movie S1) have shown nearly equal surface air temperature at the 2 sites, but substantially higher surface humidity at the forest site. As a result, surface wet-bulb temperature, ?w, was also significantly higher (⬇4K)at the forest site than at the pasture site due to higher humidity. Because the humidity control of ?w, hence CAPE (24), is a characteristic behavior of an oceanic atmosphere (25), the Amazon may be viewed as a ‘‘green ocean’’ (26).Theoretically, the level of neutral buoyancy (LNB) would be the maximum cloud height. The shallow clouds (Fig. 3) did not reach the LNB (Movie S1). which indicates that the large scale atmospheric subsidence during the dry season suppressed the convective activity, leading to the heights of clouds much lower than the LNB.Hence we argue that the (thermal) lifting mecha- nism is the dominant factor in the development of convection in the dry-season environment. The orographical lifting mecha- nism is important during wet season (27), but plays a secondary role in dry season (14). Lack of lifting mechanism over the forested surfaces leads to suppression of convection even though ample potential energy for convection is available. This line of reasoning is consistent with the observations (28) showing significantly weaker sensible heat flux over the forest than over the pasture assuming that the boundary layer turbulence is the dominant lifting mechanism to initiate convection.Therefore, it is reasonable to argue that the atmospheric instability condition over the Savannah was more like that at the forest site than the pasture site. If this is true, large CAPE combined with stronger sensible heat flux (relative to that of the surrounding forest) would provide a favorable environment for the development of deep clouds. Even though the exact surface conditions were unknown at the Savannah, the reality that deep clouds only occurred over this forest-like surface is consistent with the statistical analysis presented earlier.The rarity of deep clouds reflects that there is the lack of lifting mechanism to initiate convection over the forest area, contrary to the lack of sufficient energy to support deep convection over the pasture. The CAPE and the vertical distribution of moisture in the cloud environ- ment determines whether the clouds are shallow or deep once convection is initiated, while the lifting mechanism determines whether clouds occur, shallow or deep, which we discuss below.Although the lack of shallow clouds over the forested area can be explained by the shallow boundary layer relative to the LCL, the boundary layer turbu- lence alone cannot explain the behavior of the shallow clouds over the deforested area.The observational evidence strongly supports the argument that the dominant lifting mechanism was due to the mesoscale circulations instead of the boundary layer turbulence. Lack of lifting mechanism due to mesoscale circulations over the uniform forest is believed to be responsible for few shallow clouds over this study period as the boundary layer turbulence did not initiate convection even though there was plenty of CAPE. We also contend that the deep clouds over the Savannah were due, in part, to its less uniform land cover compared with the neighboring forest. All of the evidence points to the ther- mally induced mesoscale circulation as a major, if not the only (27), lifting mechanism for convection during dry seasons in the deforested Amazon.An open question is whether the Amazon as a green ocean now will be converted into an area with more continental characteristics as the ex- panding deforestation makes the land cover more uniform. There must be a limit to the length scale of land cover hetero- geneity above which the lifting mechanism is lost and the enhancement of shallow clouds over deforested areas will dis- appear. An analysis of scales is a subject of future research.\cite{Garcia_Carreras_2010} Use Isoprene Variations in these fluxes can then potentially impact the dynamics and growth of the PBL, the development of convective clouds via changes in, for example, the convective available potential energy (CAPE) [Segal et al., 1995], and therefore the surface radiative fluxes and amount and distribution of precipitation [Pielke, 2001].At the mesoscale, heterogeneities in surface characteristics, and thus surface fluxes, can lead to sharp gradients in PBL temperatures which may force local circulations anal- ogous to ‘‘sea breezes’’ (see Segal and Arritt [1992] for a comprehensive review).These, as well as theoretical studies such as Baldi et al. [2008], show that the necessary length scale of the heterogeneities for these circulations to develop must be approximately 10–100 km, although at low latitudes the impact is maximized for the lowest length scales within this range [Wang et al., 1996]. For smaller length scales, turbulent eddies during the day will tend to destroy these temperature gradients before a circulation can develop (e.g., as shown in the work of Linden and Simpson [1986] in laboratory experiments).High-resolution modeling of opposing sea breezes over Cape York Peninsula in Australia [Goler and Reeder, 2004] as well as laboratory experiments on gravity currents [Simpson and Britter, 1980] show that the presence of a head wind leads to a shallower and more coherent circulation, given by stronger potential temperature anoma- lies and sharper gradients between the airflows in the mixed layer. A tail wind, on the other hand, will act to deepen the circulation and reduce the thermal and wind gradients (see also discussion by Parker [1998]).land surface–induced mesoscale circulations will generally be unresolved by general circulation models, these studies suggest that their impact can be significant for the correct prediction of convective initiation and, thus, precipitation.Despite the considerable number of modeling studies corroborating the presence and significance ofland surface– induced mesoscale circulations, there are few observational studies demonstrating their existence because of real life heterogeneities not being as well defined as in the modeling studies as well as the difficulty of decoupling any local, land-induced effects from the synoptic flow.It seems that higher evapotranspiration over the high isoprene- emitting shrubland, as opposed to the low isoprene-emitting cropland, leads to increased latent heat fluxes, an associated decrease in sensible heat fluxes, and thus to cool PBL temperature anomalies. These temperature anomalies also seem to persist throughout the season. It is proposed that the temperature gradients around these temperature anomalies then initiate land surface–induced flows, or vegetation breezes,More importantly, in practically all cases the peak in the temperature gradient coincides with a convergence zone. Thus, the convergence zones lie over the boundaries between the different vegetation types. For example, the divergence over the cool patch between 11.90°N and 12.05°N leads to convergence zones in the northern edge of the vegetative boundary (dotted line at 11.82°) and more weakly in the southern edge, at 12.05°NThe spectral coherency can be regarded as a spectral equivalent to the correlation between two data sets. The cospectral analysis computes coherency as a function of length-scale, with a maximum value of 1 indicating perfect coherency, and 0 meaning the two data sets are not coherent with each other at that length-scale.The relationship between po- tential temperature and wind speed is very significant down to wavelengths of 8 km (wave numbers less than 0.125). There is a second peak between 4 and 5 km, which could be either because of the effect of turbulent eddies [Taylor et al., 2007] or boundary layer rolls (observed by Marsham et al. [2008] when comparing observations with large eddy model simulations). Figure 8 shows a bias toward more cloud in mid- afternoon (1.5–3.0 h), between 11.1°N and 11.8°N, with two peaks at 11.4°N and 11.55°N. This is consistent with the enhancement of cloud cover over the cropland, for example, when comparing to the region of reduced isoprene emissions between 11.2°N and 11.6°N. The match with the land surface temperature data, a more accurate reflection of the mean land surface impact throughout the season, is also consistent. The largest peak in the land surface temperature is between 11.35°N and 11.6°N, located in the center of the cropland, and the largest cloud cover lies just within the edges of this temperature anomaly. Earlier in the afternoon, at 1 h, there seem to be two smaller peaks at 11.2°N and 11.6°N, particularly noting that the background should be negative, rather than 0. This may be related to the initiation of clouds along the edges of the cropland boundaries. This study shows observations of a mesoscale orga- nization of the winds persisting over 2 h, with areas of convergence which control patterns in the formation of cumulus congestus clouds. The organization is attributed to PBL temperature anomalies caused by variations in sensible heat flux at boundaries between forest/shrub and cropland.
