2. The observed broad scale Asian Monsoon and regional monsoons

2.1 The large scale Asian monsoon

As atmospheric measurements have become available from a number of different sources (surface observations, radiosondes, airplane observations, satellites, and intensive observation periods) in this century, the climatology and variability of the observed large-scale Asian monsoon have become better understood from a dynamical perspective. The ensuing discussion on monsoon phenomenology will be based on what this understanding is to date.

The seasonal cycle of the large-scale monsoon is, at first order, regulated by the differential heating of land and ocean surfaces, resulting from their widely differing heat capacities. The meridional seasonal migration of the latitude of maximum insulation, leads to seasonally reversing land-sea temperature contrast. This, in turn, leads to seasonally reversing meridional temperature gradients and tropospheric circulations, and migrating large-scale precipitation regions.

 2.1.A Asian monsoon climatology

In the ensuing discussion, please refer to Figure 2.1 for locations of the regions being discussed. Before summer monsoon onset, south Asia is dominated by the dry winter monsoon circulation, characterized by lower tropospheric northeasterlies and mid to upper tropospheric westerlies (Figure 2.2a for May). The lower tropospheric circulation results from high surface pressures over the relatively cold extratropical Asian land mass and low pressures over the relatively warm Indian and tropical Pacific Oceans, while the middle to upper tropospheric circulation results from the prevailing north-south thermal gradient through the troposphere. Large scale convective precipitation is found near and south of the equator, over the Indian Ocean, the maritime continent, and Australia, in association with the Intertropical Convergence Zone (ITCZ) (e.g., Lau and Li 1984).

Figure 2.1: Geography of global scale and monsoon scale regions mentioned in text. Top panel gives global scale regions, and bottom panel gives monsoon region locations.

During boreal spring, observations and GCM studies indicate that the Tibetan Plateau, which has an average elevation of about 5 km above sea level, plays a crucial role in the seasonal reversal of the monsoonal circulation (e.g., Li and Yanai, 1996; Yanai and Li, 1994; Vernekar et al., 1995). Once its cover of snow is removed in spring, the Tibetan Plateau becomes an elevated sensible heat source for the middle and upper troposphere (Li and Yanai 1996, Yanai and Li 1994). As a result, the middle and upper troposphere over south central Asia are heated more vigorously during the spring than the surrounding areas. By late May, the meridional temperature gradient over Tibet becomes reversed, with temperatures higher over the south Tibetan Plateau than further south over India and the Indian Ocean (Li and Yanai 1996). Winds in the upper troposphere reverse from westerly to easterly in response to the change in meridional temperature gradient, and the so-called Tibetan high develops in the upper troposphere as the atmosphere adjusts hydrostatically to the warming at the elevated lower boundary. During this time, the southerly and southwesterly Somali Jet develops off the coast of east Africa (
Fig. 2.2d and 2.2e). A broad cross-equatorial flow in the lower troposphere from the South Indian Ocean subtropical high toward the south Asian land mass commences. The summer monsoon trough develops over the Bay of Bengal, adjacent southeast Asia, and eastward over the South China Sea and Philippines. This surface pressure feature marks the position of the northward shifting ITCZ, and is the main focus for large-scale moisture convergence and convection. The advance of precipitation from the oceans onto the land is marked by sudden onset, and a change in location of regional moisture convergence from the now relatively cool oceans to the south of the Asian land mass to the heated land mass itself (Fig. 2.2d through 2.2f). Preceding the change in the meridional location of moisture convergence is the possible development of symmetric instability, which arises from advection of negative potential vorticity across the equator from the southern to the northern hemisphere with the seasonal meridional wind reversal (e.g., Tomas and Webster 1997; Krishnakumar and Lau 1996, 1997).

Over the south Asian land mass, sensible heating from below, with increasing moisture convergence from the surrounding oceans, results in a favorable environment for organized convection. The development of such landlocked organized convection is important for the continued development and maintenance of the Asian monsoon, as the large-scale latent heat release resulting from convection maintains and expands the initial thermal gradient set up by sensible heating of the Tibetan Plateau (e.g., Webster et al., 1997; Li and Lanai 1996).

Figure 2.2a-c: The 200 hPa circulation from 40 to 140E, and from 10S-50N, for (a) May, (b) June, and (c) July. Wind vectors are scaled by the arrow below each panel. Units are meters per second.



At the height of the summer monsoon, the reversed temperature gradient initially found south of the Tibetan Plateau spreads across much of the eastern hemisphere from the western Pacific to Africa. The spatial scale of the Tibetan High shows a concomitant increase in breadth and strength, with extensive easterlies through the mid and upper troposphere to its south (Webster et al. 1997, see Fig. 2.2c). These features are maintained through the boreal summer by convective heating over the Asian monsoon region and Africa. Over East Asia, the lower tropospheric southwesterlies advance as far north as Japan, Manchuria, and Korea, with attendant rains (Lau et al. 1988, Lau and Li 1984, see Fig. 2.2f). The rains in these more northerly latitudes reflect combined forcing from both synoptic scale extratropical and convective/tropical processes.

The behavior of precipitation over the south Asian land mass during the summer monsoon season is characterized by active (wet) versus break (relatively dry) phases with periods of 10-20 and 30-60 days. The former period has been associated with land surface processes (Webster 1983, Srinivasan et al. 1993). The latter period is related to oscillations in the ITCZ associated with the west-to-east and south-to-north movement of the Madden-Julian oscillation and the associated regional Hadley cells (Gadgil et al. 1992).

By late August, monsoon rains begin to retreat from the south Asian land mass from northwest to southeast. The withdrawal of monsoon rains from the continent to the surrounding oceans is more gradual than their onset. As the Asian land mass cools, the meridional temperature gradient between the Tibetan Plateau and surrounding regions once more reverses, and supports a subtropical westerly jet in the middle and upper troposphere. As the Eurasian land surface cools, the Siberian high begins to build over the northern part of the Asian continent while the ITCZ and subtropical high over the Indian Ocean retreat to the south. As a result, the lower tropospheric southwesterlies once more become northeasterly, bringing an end to lower tropospheric moisture convergence over the south Asian land mass. Generally, by the end of October, rainfall is once more limited to the oceanic regions south of Asia, and the winter monsoon circulation regime is reestablished.


Figure 2.2d-f: Same as Fig. 2.3a-c except for the 850 hPa circulation in (d) May, (e) June, and (f) July.


Any physical process which alters either the amplitude or phase of the seasonal cycle of the global scale land-sea temperature contrast can be a mechanism for interannual global scale monsoon variability. This involves one or more components of the land-atmosphere-ocean system, and feedbacks among them. SST variability takes place mostly on large spatial and long time scales on the order of two to seven years (Quasi-Biennial Oscillation [QBO], ENSO), and affects the large-scale divergent flow in the tropics through displacement of large-scale precipitation regions and the associated latent heating. In contrast, land surface processes which may alter monsoon behavior take place on a multiplicity of time scales from hourly (e.g., surface fluxes of heat and moisture) to intraseasonal (e.g., soil moisture anomaly decay time scales), to interannual (e.g., Eurasian scale snow mass anomalies). Spatial scales of land surface variability range from a kilometer or less (e.g., vegetation heterogeneity) to continental scale (e.g., snow mass).

2.1.B SST variability and the Asian monsoon

Early studies of the large scale Asian monsoon did not have the benefit of upper air radiosonde observations, satellite data, and the like, and relied on surface rainfall measurements to determine Asian monsoon intensity. The longest and most reliable time series of monsoon rainfall are found for the Indian subcontinent, and have been extended back to 1845 (Sontakke et al., 1993). Walker (1910) attempted to relate Indian monsoon rainfall (IMR) to global phenomena, in an attempt at monsoon prediction. In the process (Walker 1910), he identified the Southern Oscillation (SO), the interannually swaying sea-level pressure perturbation across the southern tropical Pacific and Indian Oceans. He used statistical techniques to link precursory and current climate anomalies to the SO and IMR (Walker 1923, 1924, Walker and Bliss 1932). While his attempts to accurately predict IMR using, these methods were ultimately unsuccessful, he was the first to view climate anomalies in a hemispheric framework, and to link the Asian monsoon and SO.

The link between the SO and equatorial Pacific SSTAs was not made until the 1960s. However, once that link was hypothesized by Bjerknes (1966, 1969) and established by Rasmusson and Carpenter (1982), investigation could begin on the interaction between the coupled El Niño (EN)/Southern Oscillation phenomenon, or ENSO, on the Asian monsoon. Rasmusson and Carpenter (1983) established a connection between EN and IMR, and a number of subsequent investigators have confirmed this work to varying degrees. This study was extended globally by Ropelewsky and Halpert (1987, 1992) with respect to both seasonal precipitation and surface temperature anomalies.

Webster and Yang (1992, hereafter WY) undertook an observational study of ENSO-Asian summer monsoon relationships, using large-scale dynamical measures of interannual Asian monsoon variability determined from the M1* index discussed in chapter 1, rather than the more capricious IMR index, which has strong local land surface influences. WY found that significant anomalies in the general circulation were associated with variability in equatorial Pacific SSTs, and that summer circulation anomalies were preceded by anomalies of a similar sign for as long as two or three seasons prior to the summer monsoon of concern (the so-called precursory signal), and were related to ENSO. The dynamical relationships WY found were more clear than those for IMR. Yang et al. (1996) extended this study using the Goddard Laboratory for Atmospheres (GLA) model and found that the precursory signals found in the observations are also found in simulations with observed SSTAs.

Soman and Slingo (1996) used the UGAMP climate model to determine the relative effect of SST anomalies in the tropical Pacific warm pool (western Pacific) and cold tongue (eastern Pacific) on the evolution of the Asian monsoon. They found that the SSTAs in the western Pacific were required to reproduce the monsoon anomalies observed. The onset of the Asian monsoon was particularly affected, with delayed onset associated with the cool western Pacific SSTAs associated with the warm phase of ENSO events. This is counterintuitive to what one might expect, since cool western Pacific SSTs increases land-sea contrast and thus should result in a stronger monsoon. The circulation anomalies induced by the decrease in equatorial SST gradient across the Pacific may overcome the influence of land-sea temperature contrast in determining the ultimate Asian monsoon behavior.

In summary, positive SSTAs over the equatorial eastern Pacific have a negative impact on the broad scale Asian monsoon, through zonal shifting of the regions of maximum tropical precipitation and diabatic heating from convective precipitation processes. Asian monsoon circulation anomalies can be measured through broad-scale indices of area-averaged circulation such as the M1* index of WY.

The relationship between the broad-scale Asian monsoon index and regional precipitation, however, is less clear, because precipitation processes take place on smaller temporal and spatial scales than the broad scale Asian monsoon. As a result, regional precipitation, such as IMR, is less well correlated with Pacific SSTAs than the broad scale monsoon.

Both physical and dynamical measures of monsoon variability are of interest. The dynamical, broad scale Asian monsoon teleconnects to circulation anomalies for long distances downstream from the Asian land mass (Lau 1992), and thus can have an impact on North American summer climate. The regional rainfall is of interest because it is this rainfall that has economic, agricultural, and social implications for the peoples of South Asia.

2.1.C Land surface variability and the Asian monsoon

Since the large scale Asian monsoon is driven by the seasonal cycle in land-sea temperature contrast, many researchers have examined the sensitivity of the Asian monsoon to variability in the Asian land surface. The most obvious large scale influence of the warming of the Eurasian land mass in spring is the extent and depth of continental scale Eurasian snow cover and snow water content (or snow mass). Blanford (1884) drew a connection between Himalayan snow depth and the subsequent summer IMR. A number of subsequent observational studies (e.g., Walker 1923, 1924; Hahn and Shukla 1976; Dickson 1984) have linked anomalous Asian snow cover with the strength of the subsequent Asian monsoon. Yang (1996) notes a precursory signal in snow mass, which may indicate that positive Eurasian scale snow mass anomalies are related to warm ENSO events. This would imply that Eurasian snow mass anomalies may act as a bridge between ENSO and the Asian monsoon, mediated through the hydrologic anomalies over the Eurasian continent resulting from excess snow mass.

Recent modeling studies using idealized snow mass anomalies have linked such anomalies to Asian monsoon strength. Barnett et al. (1989) and Vernekar et al. (1995) found from numerical experiments that hydrological (rather than radiative) effects of large snow mass anomalies were key factors in Asian monsoon variability. Doubling of climatological snow mass over Eurasia points in the antecedent winter resulted in a weak Asian monsoon the following summer, while halving the Eurasian snow mass resulted in a strong Asian monsoon the following summer. While Barnett et al. (1989) invoked soil moisture anomalies on a continental scale, Vernekar et al. (1995) specifically attributed the weakened Asian monsoon to anomalously low sensible heat fluxes over the Tibetan Plateau. These fluxes reduced the meridional temperature gradient (and thus the upper tropospheric easterlies) between that region and the Indian Ocean. In an observational study using the European Center global analyses, Li and Yanai (1996) found that the strength of the Asian monsoon was sensitive to sensible heating anomalies over the Tibetan Plateau in the prior boreal spring. Additionally, they found that negative upper tropospheric temperature anomalies of hemispheric scale, centered over the Tibetan plateau, were associated with weak large-scale monsoons, and positive temperature anomalies with strong large-scale monsoons. They cited the earlier snow mass studies above in suggesting that hydrologic anomalies resulting from continental scale snow mass might be the reason for the anomalous sensible heat fluxes.

It should be noted that the snow cover-monsoon relationship has been difficult to prove decisively. While Walker (1923, 1924) used Himalayan snow cover in his regression equations for Indian summer monsoon prediction, the Walker and Bliss study of 1932 deleted the snow cover variable from the equations, citing both poor measurements and the weakening of snow cover-monsoon correlation. Additional difficulties arise in treating the full Asian monsoon as one system. For example, Yang and Xi (1994) found that observed rainfall over China as a whole shows no relationship to observed Eurasian snow cover. However, they did find significant sub-regional relationships; northern and southern China summer monsoon precipitation were positively correlated to Eurasian snow cover, while western, central and northeastern China showed negative correlations. As discussed in section 2.2, each of these regions is affected in a unique way by the climatology of the East Asian monsoon, from a monsoon onset to maintenance to withdrawal. These regional relationships suggest that the regional rainfall correlations of Chinese rainfall with boreal winter/spring Eurasian snow mass are consistent with alteration of the seasonal cycle in the East Asian monsoon.

2.2 The East Asian Monsoon

2.2.A Climatology

Figure 2.3 shows the transition in rainfall over the Asian monsoon from pre-monsoon to a monsoon onset. The East Asian monsoon summer rainfall climatology is associated with two significant transitions in the movement of a northeast-to-southwest oriented rain belt (Lau 1992 and references therein). In May (the pre-monsoon period), the rains occupy an area over southern China and southeast Asia. This is caused by the interaction of tropical moisture with the subtropical jet and the polar front which are found around 30°-35°N latitudes.

Figure 2.3: Advance of Asian monsoon region precipitation form (top) May to (middle) June to (bottom) July. Units are mm day-1, with positive regions shaded with solid contours, negative regions clear with dashed contours, and the zero contour highlighted. Contour inteval is 1 mm day-1.

The first monsoon transition (Fig. 2.3b) takes place during June, and is associated with an onset of the Mei-Yu rains over central China and the Baiu-Yu or plum rains in Japan. The second jump occurs in July (Fig. 2.3c), and corresponds to the commencement of the summer monsoon in northeastern China and Korea. During this time, central China dries out. By late July and August, a different regime takes place, with mean rainfall maxima progressing from south to north through China with an approximate 20 day period (Lau et al., 1988, not shown). This regime may result from the relative dominance of land surface processes during high summer compared to synoptic scale variability, because of reduced baroclinic instability once the summer season is established. Land driven precipitation variability has been shown to have a 10-20 day period (Webster 1983, Srivasan et al. 1993).

Each transition in monsoon rainfall in east Asia is associated with a shift in the regional mean circulation (Fig. 2.2). The mean 200 hPa Tibetan high (Figs. 2.2a-c), the 850 hPa monsoon westerlies (Figs. 2.2d-f), and the western Pacific subtropical high shift northward as the monsoon season progresses. The sudden transition from the pre-monsoon to Mei-Yu rains is marked by a shift of the east Asian zonal wind maximum at 200 hPa from 35°N to 40°N, and of the axis of the Pacific subtropical high from 15°N to 25°N (Lau, 1992 and references therein). The commencement of northern China rains, and the drying out of central China, is again marked by a sudden transition in the 200 hPa wind maximum, from 40°N to 45°N and a shift in the subtropical high from 25°N to 30°N.

2.2.B The East Asian Land-Sea Precipitation Dipole

Analysis of a number of different observational data sets, derived from different sources, indicates that there exists an east Asian land-sea precipitation dipole (EAPD). In the ensuing discussion, mean monthly data are analyzed from a number of these data sets, using one point correlation of South China Sea precipitation anomaly (SCSPA) time series to the remainder of the Asian monsoon region (40°-140°E, 2°-50°N). The SCSPA is calculated over all oceanic grid points within 110° to 120°E and 10° to 20°N for each data set. In some cases, the spatial resolution of the data set will result in a domain slightly smaller than the one cited here .

Additionally, these rainfall data sets must be used with caution. The reanalyses shown (Figures 2.4 and 2.7) have interactive land surfaces, but the rainfall is predicted by the model rather than assimilated from observations. Thus model biases in rainfall will result in biases in the land surface soil moisture, which will feed back onto the precipitation, resulting in further errors. For the precipitation products derived from remote sensing, blended, or rain bucket observations, there are inaccuracies resulting from problems with remote sensing algorithms and inferring aerial coverage of precipitation from point sources of data.

Figure 2.4 shows the SCSPA time series, and the June one-point correlation map of Asian monsoon region precipitation to that time series, for the National Centers for Environmental Prediction (NCEP) reanalysis (Kalnay et al., 1997) from 1982 to 1993. The NCEP reanalysis, at 2.5° latitude by 2.5° longitude resolution, uses interactive land surface hydrology in its assimilation of the atmospheric forcing data, and hence is more useful for comparison purposes than reanalyses using mean land surface climatology. We note the dipole structure between the SCS and the Chinese mainland to the north during June, with negative correlations over east-central China as large as 0.8. Absolute values greater than 0.55 are statistically significant at the 95% confidence level, using Fisher's z statistic.

Figures 2.5, 2.6, and 2.7 illustrate the SCSPI time series and one point correlations found between the SCS precipitation index and the precipitation over the Asian monsoon , obtained using the 4° latitude by 5° longitude, MSU precipitation data (Spencer, 1993), the 2.5° latitude by 2.5° longitude blended GCPI index of Huffman et al. (1995), and the 2° latitude by 2.5° longitude GEOS-1 reanalysis, respectively. All data were taken from 1980 through 1989 for direct comparison with the initial GEOS-1 climate model experiments. Correlations of greater than 0.63 are 95% significant, again using Fisher's z

Figure 2.4: The NCEP reanalysis, 1982-93 South China Sea seasonal anomaly time series (in mm day-1), for the region defined in the text (top panel), and the one point correlation (dimensionless units) in June to seasonal anomaly in precipitation for all grid points from 40-180E and from 10S to 50N. The contour interval is 0.2 in the bottom panel, with positive values shaded and negative values dashed. Statistical significant correlation (95% level)exists for absolute values greater than 0.55.

Figure 2.5: Same as Fig. 2.4, except using the Microwave Sounding Unit (MSU) rainfall estimates from 1980-1989.

Figure 2.6: Same as Fig. 2.4 except using the blended Global Climate Precipitation Index (GCPI) of Huffman et al. (1995) from 1980-1989.

Figure 2.7: Same as Figure 2.4, except using the GEOS-1 reanalysis from 1980-1989.

statistic. In the East Asian monsoon region, the correlation maps are similar, with the meridional land-sea precipitation dipole evident in all data sets. This is in spite of different spatial resolutions and analysis methods for determining the amount of rainfall. Each method has its shortcomings, but the consistency in results among all four data sets indicates that the signal is quite robust.

Figure 2.8 shows the first three EOF modes of variability for climatology of high cloudiness (a precipitation surrogate) found by Kang et al. (1997) from the International Satellite Cloud Climatology Project (ISCCP) data set (Rossow and Schiffer, 1991). The differences from the mean May to August values for high cloud were used for the EOF calculation. The first two modes (51.9% and 19.9% of total climatological variance, respectively) represent the climatological advance of the monsoon from ocean to land; in particular, the second mode represents, in part, the Mei-Yu precipitation advance discussed earlier in this chapter. The third mode (9.9% of total climatological variance), however, is the dominant intraseasonal mode of variability, and seems to indicate a seesaw pattern in the climatology of precipitation during the monsoon season between south Asian land and the adjacent oceans. Note that the pattern is qualitatively similar to that of the one point correlation maps in the east Asian monsoon region.

Figure 2.8: First three leading eigenvectors and associated time series obtained from the EOF analysis of the 5-day mean climatological variation of high cloud fraction from May to August. Positive and negative values are denoted by solid and dashed lines, respectively. Arbitrary units. From Kang et al. (1997).

Kang et al. (1997) also calculated extended (or moving window) EOFs (EEOFs) for the same data set, after removing the seasonal cycle of high cloudiness from May through August (
Fig. 2.9). The first two EEOFs (together explaining 32.5% of the total intraseasonal variance) are similar, and may represent a dominant fixed frequency since the respective principal component time series are in approximate quadrature with each other. The mode was active not only during onset, but later in the monsoon season, indicating that a land/sea competition for precipitation exists during the entire monsoon season. Over a 20-day period, the spatial pattern reverses sign. The Kang result lends additional support that the EAPD is a physical mode of the East Asian monsoon. The amplitude of the mode shows interannual variability as well.

Figure 2.9: First and second modes of the extended EOF of climatological intraseasonal component of high cloud fraction with time lags from -10 days to +10 days. (A)-(c) are the first eigenvector components for each time lag, (d)-(f) for the second eigenvector, and (g) the time series associated with the first and second eigenvectors, denoted by solid and dashed lines, respectively. Arbitrary units.


2.2.C East Asian Monsoon Failures

The behavior of intraseasonal variability has been shown to determine the overall precipitation anomaly for a summer monsoon season. For example, variability in so-called 'active' and 'break' monsoon behavior is associated with interannual variability in total ISM rainfall (e.g., Gadgil and Asha, 1992). Monsoon failures in India and southeast Asia are associated with frequent breaks in monsoon rainfall. Over East Asia, however, the behavior of the monsoon is more complex because of more overt extratropical influences, and is described in more detail below.

Deficient East Asian monsoon precipitation is generally associated with the skipping of one of its sudden transitions (cf. section 2.2.A), or by shorter than normal residence time in one of the phases of monsoon progression (e.g., Park and Schubert 1997, regarding the 1994 drought in China). This is evident from the similarity in the spatial structure of variability in Chinese monsoon precipitation at temporal scales from pentadal to monthly to interannual (e.g., Shen and Lau, 1995). Such drought may be associated with internal dynamical processes (Park and Schubert, 1997) or anomalous SST forcing (Shen and Lau, 1995). Additionally, since the land surface is important in determining the response to global scale circulations at smaller scales, it may play a role in determining the character of the seasonal march of the East Asian monsoon. The land surface state is also important in determining the precipitation intensity while the monsoon is in a given phase of its seasonal advance, since static stability (and thus convective precipitation) is strongly affected by energy and moisture fluxes at the land-atmosphere boundary.

Nineteen ninety four was the last of a series of drought years over East Asia (Kogan, 1997), but was not marked by significant precursory SST anomalies in the Pacific basin (Park and Schubert 1997). This leaves internal dynamics or land surface processes as causes for the 1994 East Asian drought. Park and Schubert (1997) suggests that the cause of the 1994 drought was dynamical in nature, and involved development of the summertime circulation in June rather than July.

The cause of this appeared to be the development of an anomalous blocking upper tropospheric anticyclone over the western Eurasian continent (Park and Schubert, 1997, Fig. 5), resulting in a downstream response over East Asia. This anomalous anticyclone has the effect of advancing the summer monsoon large scale circulation (Park and Schubert, 1997, Figs. 7 and 14). As a result of this acceleration of the seasonal cycle, the Mei-Yu and jump was largely missed during the monsoon season. However, the persistence of drought from previous years, as evidenced by remote sensing of vegetation density (NDVI, Kogan, 1997), suggests that deficient soil moisture in the land surface may also have been the cause of this particular drought, or at least contributed to its severity. Studies by previous researchers have shown that mid-latitude drought in North America is sensitive to the soil moisture state (e.g., Atlas et al., 1993), which amplifies any tendency to drought resulting from remotely forced circulation anomalies. It is possible that in East Asia as well, the global scale dynamics and the continental scale land surface state may strongly interact to produce interannual anomalies. How this interaction takes place has not been investigated.

2.3 The land surface/large scale circulation coupling

2.3.A Land surface energy and hydrologic cycles

The reflectivity or albedo of the land, and the overlying clouds, control the amount of incident solar radiation which is absorbed by the land surface. The absorbed short wave radiation is then partitioned into different components. This partitioning can be summarized through the land surface energy equation:


where Rn is net radiation, S is total solar radiation incident on the surface, is the surface albedo, FLW is the downwelling long wave radiation, Ts is the surface temperature, is the surface emissivity, is the Stefan-Boltzmann constant, H is the sensible heat flux (conduction of heat to the air), LE is the latent heat flux (energy release to the air by evaporation), L is the latent heat of vaporization, E is the moisture flux into the air, G is the heat flux into the ground, and Ph is the energy used for photosynthesis by plants.

Elements which control the amount of energy partitioned into latent heat flux are vegetation, soil moisture, and surface roughness. Over time scales longer than a day or so, Rn mostly consists of sensible and latent heat fluxes. The ratio of the sensible to latent heat flux (the Bowen Ratio) can be a significant determinant of the resulting climate, because of the different nature of the two fluxes. Under moist conditions, much of the net surface radiation is transported upward by latent heat flux, while if the land surface is dry, much of the net surface radiation goes into sensible heat flux. The difference in climate results from the more efficient transport of latent heat than sensible heat, though adiabatic cooling, condensation and subsequent latent heat release and diabatic heating in precipitation processes. It will later be shown (Chapter 6) that the amount of latent energy in the planetary boundary layer is the primary determinant for convection over the East Asian monsoon region.

The land surface hydrologic cycle is controlled by water inputs from the atmosphere and outputs through percolation into deep soil moisture reservoirs, runoff of excess liquid water through surface and sub-surface water transports, and interception by the vegetation canopy. This can be described by the following hydrologic balance equation:


where P is the precipitation input, SM is the local change in soil moisture, I is the interception of precipitation by leaves in the vegetation canopy, R is the runoff through surface and sub-surface liquid water pathways, and Sn is accumulation of snow mass. Note that the land surface energy and hydrologic cycles are linked through the moisture flux E.

2.3.B Vegetation morphology and physiology

The flux of moisture from the land surface is strongly controlled by vegetation morphology, the size and shape of plants, the canopy they create over the land surface, and resistance to evaporation from subsurface soil moisture reservoirs provided by leaves (Sellers et al., 1986). The canopy can contain a relatively small amount of water from precipitation. The excess continues down to the land surface as throughfall. While the amount of water instantaneously held in this reservoir is only a few mm, its availability for immediate evaporation into the atmosphere increases its importance to the hydrologic balance. Results from modeling experiments concerning sub-grid scale precipitation persistence, especially in regions dominated by convective precipitation like the monsoon regions, indicate that the hydrologic cycle is sensitive to the way precipitation persistence is parameterized, and thus to the canopy interception reservoir itself (Scott et al. 1995). Observational studies have found that up to half the LE comes from canopy interception in the Amazon rain forest (Shuttleworth et al., 1988) during the rainy season. The effect of such water recycling is to short-circuit the connection between the atmosphere and the surface and sub-surface soil moisture reservoirs, and reduce land-atmosphere hydrologic coupling (Scott et al. 1995).

The essence of land-atmosphere hydrologic coupling is contained in the behavior of the small holes in the canopy leaves called stomates, which open and close to regulate the transfer of moisture from the leaves to the atmosphere, in response to environmental stress. Stomates, while covering only about 1% of the total leaf area, totally control the transpiration of water from the sub-surface soil moisture reservoirs. Energy for plant metabolism is provided by short wave and near-infrared radiation (photosynthetically active radiation or PAR). Environmental stresses are provided by air temperature, leaf water potential deficit (related to surface and root zone soil moistures), and ambient vapor pressure deficit stress (Jarvis 1976). Absent stress, the availability of PAR will determine the evapotranspiration rate. Modeling the biosphere in terms of stomatal control will be discussed in greater detail in Chapter 3.

2.3.C Precipitation processes and the diurnal cycle

Remote sensing, surface based observations, and computer modeling studies have determined that there are distinct land precipitation oscillations in the tropics. Over land, the diurnal cycle of precipitation is clearly linked to surface evaporation or evapotranspiration over vegetated surfaces. Where the afternoon precipitation maximum dominates, the diurnal cycle of precipitation is dominated by solar heating and its effect on static stability. Where and when soils are moist, latent heat flux is enhanced through heating of the land surface and the provision of PAR for plant metabolism, as discussed in the previous section. Local moistening of the planetary boundary layer (PBL) takes place as the turbulent flux of latent heat from the land surface increases. Increases in PBL internal energy are dependent on the heating of the boundary layer by sensible heat. Other factors that affect diurnal cycle in precipitation include atmospheric dynamics, local land surface variability, and radiative forcing (e.g., Gray and Jacobson, 1977).

Examining the impact of the diurnal cycle of various aspects of the land surface makes sense in light of the diurnal variability of precipitation in the Asian monsoon and East Asian monsoon regions, and the importance of convective heating in the development and maintenance of the Asian monsoon and East Asian monsoon. This will be discussed further in Chapter 3, in terms of experiment design considerations.

2.3.D General circulation interactions with land surface processes

The general circulation may be considered part of the forcing mechanism for the land surface processes. This forcing takes places through cloudiness and precipitation induced by the larger scale circulation. Through cloudiness and precipitation, the slowly varying general circulation exerts a control over the more rapidly varying energy and hydrologic inputs.

The land surface energy and hydrologic cycles may feed back onto the regional circulation and to the large scale circulation beyond. The nature of these feedbacks is in dynamical response to the tropospheric heating caused by latent heat release due to convection. Large scale convection in turn is influenced by the amount of moist static energy in the planetary boundary layer, which is controlled in part by land surface processes. The land surface may reinforce or damp the tendency for wet or dry conditions produced by the general circulation. These mechanisms are discussed further in the next section.

2.4 A hypothesis on land coupling to the large scale circulation

Based on the discussion of land-atmosphere coupling above, and what we observe of the East Asian summer monsoon precipitation variability, I summarize the discussion from the previous sections in a hypothesis of how the continental scale land surface may interact with the general atmospheric circulation in the East Asian monsoon region. Land surface physics suggests that land surface energy and hydrologic cycles are sensitive to vegetation control over evaporation (and thus vegetation type) and the characteristics of the underlying soil which determine the soil moisture holding capacity. I have also discussed the importance of partitioning of surface heat flux on climate, and the possible influence of the general circulation on the land surface hydrologic and energy cycles (precipitation minus evaporation, or P - E). The following discussion of land surface energy and hydrologic cycle modalities explains the hypothesis.

2.4.A The cool, wet land surface mode

For the cool, wet regime, evapotranspiration (and free evaporation from a wet vegetation canopy) provides a significant moisture input into the PBL. The canopy and surface are kept relatively cool, also, which prevents the temperature from increasing to the point where temperature stress on the vegetation becomes important. Convection typically can take place through buoyant (or other) lifting, with the result being enhanced deep convection and precipitation, and heating through a considerable depth of the troposphere. The middle and upper tropospheric heating from the large scale convection tends to reinforce the regional moisture convergence, upward motion, and precipitation, through creating its own convergent/divergent circulation couplet in the troposphere. Figure 2.10 is a schematic of the processes involved in the maintenance and damping of the cool, wet regime. Damping mechanisms are indicated by the negative feedback (NF) links.

Moistening through moisture convergence takes place on synoptic to seasonal time scales, and result from circulation changes initially induced by land-sea temperature contrast. These circulation changes are subsequently enhanced by large scale convection which reinforces the lower tropospheric moisture convergence. Land-sea temperature contrast, as modulated by the energy and hydrological state of the land surface, is a negative feedback to the hot, dry regime on synoptic to intraseasonal time scales.

Figure 2.10: Schematic showing elements of land-atmosphere coupling over the East Asian monsoon region, involving the fast response of the hydrologic and energy cycles and the influence of the large scale circulation induced by the slowly varying part of the Asian monsoon -SO system (e.g., ENSO). The signs of the anomalies indicate the situation for a cool, wet cycle, with the large scale forcing favoring general upward motion over the Asian monsoon region. Negative feedback links are denoted by the symbol NF. The sign of the change should be reversed after each NF link (e.g., LW and SW effects resulting from changes in Ts damp further changes in Ts, and changes in Ts damp further changes in precipitation because of land-sea contrast). For the hot, dry cycle, the feedback mechanisms are the same, but with all the signs reversed, including the arrows for the large scale forcing. From Lau and Bua (1998).

2.4.B The hot, dry land surface mode

On the other hand, if the latent heat flux is restricted because of lack of available water for evaporation, more incoming solar energy will go into heating the land surface and thus sensible heat flux will be higher. An additional positive feedback on the reduced evaporation is provided by the increased surface and canopy temperature, since at high temperatures photosynthesis is reduced, thus further increasing vegetation resistance to evapotranspiration. Even though the resulting land surface and PBL temperatures will be higher with deficient soil moisture, it may be more difficult to generate moist convection, since moisture for condensational heating is also deficient. The moist static energy in the PBL is less, despite the higher PBL temperature. Turbulent exchange between the PBL and free atmosphere can also mix down drier air from higher levels in the troposphere, thus exacerbating an already dry regime. Convection and precipitation will be suppressed, and incoming solar radiation will reach the surface relatively unimpeded, thus further heating the land surface (a positive feedback). The increased short wave radiation reaching the surface is partially offset by increased sensible and upward long wave fluxes. These negative feedbacks result from increased surface temperature and reduced lower tropospheric moisture (which allows more long wave radiation to escape to space because of reduced 'greenhouse effect'), and increased vertical temperature gradients which enhance the turbulent flux of sensible heat. The long wave and sensible heat fluxes damp the surface temperature increase, and act on short time scales to bring the surface energy budget to a new, though higher, balance. Unless the boundary layer moistens through advection from elsewhere, the hot, dry regime will continue.

An experimental design to test the above hypothesis will be described in Chapter 3.