4. Impact of initial conditions on monsoon evolution

4.1 Experiment design

The experiments discussed in this chapter test the hypothesis for land-atmosphere interaction outlined at the end of Chapter 2. They consist of four-member ensemble simulations, which are used to analyze the behavior of the modeled East Asian summer monsoon. In these ensembles, data are produced daily to facilitate analysis of the intraseasonal variability of the East Asian summer monsoon, including the East Asian land-sea precipitation dipole. Each ensemble member is started at May 1, using a random selection of four May 1 initial conditions from a 20-year simulation of climate by the GEOS-1 GCM coupled to the Mosaic LSM, forced using the Reynolds annual cycle climatology for global SSTs (Reynolds, 1988). In the experiments, the SSTs are also set to this climatology, since we wish to isolate the role of the land surface in the variability of the East Asian monsoon, without considering non-linear land surface feedbacks resulting from anomalous SSTs forcing.

We investigate the portion of the hypothesis related to land surface processes and the energy and hydrologic cycles. Additional information can be obtained on the effect of the May 1 initial state on the subsequent development of the summer monsoon, by performing separate experiments where different aspects of the initial climate state are varied. Here, different elements of the initial state will be varied in three ensembles.

A summary of the ensemble experiments discussed in this chapter is found in Table 4.1. FILS refers to a four-member ensemble with the same initial land surface state, including all surface and sub-surface moisture reservoirs, and land surface temperature. The initial land surface condition is taken as the mean of four May 1 initial states drawn at random from a 20-year interactive land-atmosphere experiment, using the same GCM and LSM at the same spatial resolution, and with the same land physical parameterizations. FISM is the same as FILS, except that only the initial soil moisture is the same at the outset. For VISM, the initial climate state is taken from one year of the 20-year land-atmosphere simulation, but the initial soil moisture states are taken from the four years for which the mean land surface state in FILS is calculated. Finally, a four-member ensemble is run as in FISM, except that the evaporability of the land surface is set to the May 1 to September 30 mean monthly climatology for the 20-year climate simulation.

The rest of this chapter will include a description of the model climatology (Section 4.2), and a comparison to the observed climatology. Section 4.3 will discuss variability in the Asian monsoon and East Asian monsoon introduced by initial atmospheric versus land surface conditions. In section 4.4, composited wet and wet minus dry phase land surface and dynamical fields will be examined. Finally, section 4.5 will summarize the results and offer conclusions.

Table 4.1:
Summary of ensemble experiments for this chapter, describing which initial conditions are varied (columns two, three, and four, where V and F indicate differing and identical initial states among ensemble members, respectively) and whether land-atmosphere coupling is present (column five, where Y indicates land atmosphere coupling). Column 1 gives the ensemble experiment acronym.


soil moisture

Ts, snow mass, canopy interception

Atmospheric conditions/ dynamics

Land-atmosphere coupling (Y/N)?

















not applicable

F (Ts and snow only)



4.2 Model Asian monsoon climatology

The following model climatology is taken from the control simulation FILS, and can be compared to the observed climatology discussed in Section 2.1. We examine the precipitation advance, and the upper and lower tropospheric circulations, for May, June and July.

Two important points must be kept in mind regarding the climatology of the model versus the observations: observations include the effect of large-scale anomalous sea surface boundary forcing such as El Niño, while the simulations use the annual cycle climatology of the global sea surface temperatures. This may render the mean precipitation patterns different in the model from the observations, regardless of how well the model simulates the Asian monsoon. This is especially true for the Asian monsoon, because its response to El Niño type forcing is not linear; that is, the anomalies for cold events are not mirror images of those for warm events. Additionally, the observations and the simulations are subject to sampling error because of the limited sample size. The actual mean climatology will differ from that of a sample drawn from the population for which the climatology is developed.

4.2.A Precipitation

Figure 4.1 shows mean model rainfall for the Asian monsoon region for May, and the one month change in rainfall from May to June and from June to July. This figure illustrates the climatological monsoon advance over the East Asian monsoon region, and can be compared to Figure 2.3. In May, the pre-monsoon rains reside over south China, the South China Sea, and Southeast Asia. In the Indian monsoon region, heavy rains exist over the Arabian Sea and southern quarter of the Indian subcontinent.

Figure 4.1: The May (a), May to June change (b), and June to July change (c) in mean monthly precipitation from the control ensemble FILS. Areas in (a) with precipitation greater than 6 mm(day)-1 are shaded. Contour intervals in (a) are 1, 2, 4, 6, 8, and 10. In (b) and (c), areas with increased precipitation are shaded. Contours are in 1 mm(day)-1 increments from -5 to 5, with the zero contour highlighted in bold.

The change in monthly mean rainfall from May to June is striking. Precipitation increases over the Indian subcontinent (2-5 mm day-1), southeast Asia (1-2 mm day-1), and China (1-2 mm day-1, especially between 30°-35°N). The oceanic regions south of India, the Philippines, and the South China Sea all have increased rainfall (1-2 mm day-1). While its geographical placement is north of observations, the precipitation increase over China can be considered equivalent to the Mei-Yu onset. From June to July, we note reduced rainfall over China (1-3 mm day-1), and South China and Arabian Seas (1-2 and 1-6 mm day-1, respectively). The mean July drying over China is similar to long-term precipitation observations, though once again displaced northward. Mean monthly precipitation increases over Southeast Asia, Japan, Korea (all 1-2 mm day-1), and the Indian subcontinent (1-5 mm day-1) and oceanic regions to its south (1-5 mm day-1).

4.2.B Upper tropospheric circulation

Figures 4.2a-c show the mean 200 hPa circulation for the Asian monsoon region for May, June, and July, to be compared to Figs. 2.2a-c from Chapter 2. In general, the model simulates the month to month changes in the upper troposphere associated with the May to July monsoon advance. In May (Fig. 4.2a), the upper tropospheric anticyclone associated with large scale convective heating is centered over southeast Asia and the adjacent portions of the Bay of Bengal, with a zonal ridge axis extending to southern India to the west and the Philippines to the east. The mid-latitude westerly jet stream is still active to its north, with upper tropospheric easterlies limited to regions south of 10°N. In June (Fig. 4.2b), the upper tropospheric high shifts northward, with centers over northern India and south-central China. The mid-tropospheric westerlies to its north weaken, and the monsoon easterlies shift northward and increase. This is consistent with the development of convective heating from land-based large-scale precipitation over the Asian land mass (Fig. 4.1b).

Figure 4.2a-c: The 200 hPa circulation from 40°-140°E and from 10°S-50°N for May (a), June (b), and July (c). Wind vectors are scaled as in the arrow below each panel. Units are meters per second.

By July (Fig. 4.2c), the Tibetan high is fully developed and has centers straddling the Tibetan plateau, as the monsoon rains over India become fully developed. Where rainfall decreases over China, this seems to relate to the reduction of synoptic scale forcing from baroclinic (extratropical) activity, as evidenced by the weak upper tropospheric flow as the Mei-Yu activity found in June weakens and shifts northward (Fig. 4.1c).

4.2.C Lower tropospheric circulation

Figures 4.2d-f show the mean 850 hPa circulation for the Asian monsoon regions for May, June and July, and can be compared with Figs. 2.2d-f. Here, we note the development and northward expansion of the lower tropospheric monsoon westerlies as the monsoon rains move over the south Asian land mass. In May (Fig. 4.2d), we find cross-equatorial flow from the south Indian Ocean to the Arabian Sea and the southern quarter of India, where the monsoon rains reside in the model mean (Fig. 4.1a). There is lower tropospheric divergence to the north over most of India, with flow from the dry desert regions to the northwest. Over Southeast Asia, a strong west to southwesterly lower tropospheric flow converges with southerlies and southeasterlies to the east. This convergence region is marked by heavy pre-monsoon rains (Fig. 4.1a). The Pacific subtropical high axis is located at about 30°N, with southeasterlies covering the western Pacific.

In June (Fig 4.2e), the monsoon westerlies increase in speed and shift further north over the south Asian continent, consistent with the development of land-based convective rains. Over India, there is a subtle but an important northward shift in the monsoon westerlies, and an increase in moisture convergence from the Arabian Sea. This accompanies the increased precipitation over India from May to June (Fig. 4.1b). Convergence in the lower troposphere increases over the SCS. A cyclonic circulation develops over this region and to the east, consistent with the development of the summer monsoon trough. Over China, convergence also increases as the mean monthly rains increase, as the lower tropospheric flow between the Pacific subtropical ridge at 30°N and the monsoon trough to its south increases.


Figure 4.2d-f: Same as for Fig. 4.2a-c, but at 850 hPa.


By July (Fig. 4.2f), the monsoon trough over the SCS and eastward becomes fully developed. The cyclonic flow over India and the Bay of Bengal gains additional curvature. Over China, however, there is a decrease in lower tropospheric moist inflow from the surrounding oceans as the monsoon trough over the SCS and the Philippines strengthens, concurrent with decreased rainfall in that region. Southeasterlies from the ocean increase further north, over Japan and adjacent Asian mainland, as the Pacific subtropical high shifts to 40°N. While some climate features are displaced poleward from observed, we conclude that the general Asian monsoon and East ASM model climatology is adequate for this research.


4.3 Initial condition effects on energy and hydrology

We now examine simulated land surface variability and the effect, if any, of the initial state on the subsequent East Asian monsoon development. The next section describes how the initial land surface state is measured.

4.3.A Land surface energy and hydrologic state measurement

Land surface energy state is represented by the land surface temperature Ts. Ts is in turn determined by the net energy forcing at the land surface, and reflects the surface temperature required to bring the land surface into instantaneous energy balance, through the upward long wave radiative flux, after we take into account all other radiative and turbulent fluxes:




where F LW is the upward long wave flux, FLW is the downward long wave flux, Fnet-SW is the net solar or short wave flux, LHF is the latent heat flux or evaporation, SHF is the sensible heat flux, and G is the downward heat flux into the ground, all at the land surface. The energy and hydrologic cycles are coupled through evapotranspiration.

The land surface hydrologic state is described by amount of water in its hydrologic reservoirs. Water from the atmospheric branch of the hydrologic cycle available for surface and sub-surface soil moisture change (SM) is in turn determined by precipitation P minus evaporation E (P-E forcing), less runoff R and canopy interception I:


SM = (P-E) - R - I (4.2)


As discussed in Chapter 2, however, the land surface may become hydrologically uncoupled from the overlying atmosphere under certain conditions. If the soil is wet enough that evaporation can proceed unimpeded by vegetation resistance and/or the canopy is wetted, then only the energy available serves as a limiting function to evapotranspiration.


4.3.B East Asian monsoon region soil moisture and surface temperature

As discussed at the beginning of this section, the soil moisture reflects the hydrologic state of a region, and the surface temperature its energy state. The following discussion describes the ensemble mean and variability of these two states over the East Asian monsoon region.

Figure 4.3 shows the time series of area-averaged East Asian root zone soil moisture from each simulation in the three control experiments where the land-atmosphere coupling is active. The surface soil layer, with a very small moisture capacity, responds similarly to the root zone. Time series from the VISM experiments show the sensitivity of the evolving East Asian hydrologic state to varying initial global soil moisture, but the same initial conditions otherwise. These data also test whether an identical initial atmospheric state will cause the initially varying East Asian hydrologic state to converge over time. On the other hand, the same data from FISM are shown to check the sensitivity of the East Asian soil moisture evolution to initial atmospheric state, snow mass, and surface temperature. Finally, the data from FILS tests sensitivity of East Asian soil moisture evolution to the initial atmospheric state alone. Figure 4.4(a) shows East Asian soil moisture evolution for the VISM ensemble. The synoptic scale forcing, which is controlled by the initially identical large-scale and regional circulations, dominates at first, and the initial East Asian soil moisture states trace the same general pattern from their differing initial states for about 2 to 3 weeks. As the general circulation begins to 'feel' the differing initial soil moisture conditions at the lower boundary, the area-averaged soil moisture responds to evolving differences in the P-E forcing among the ensemble members, and the initially similar trajectories begin to diverge. This effect becomes significant by late June, when anomalies for individual years begin to change sign relative to the ensemble mean. At individual grid points, the hydrologic variables at the land-atmosphere interface (precipitation and evaporation) respond almost immediately to the differing initial soil moistures (not shown). Coupling of the hydrologic and energy balances through evaporation results in almost immediate changes to surface temperature and sensible heat flux at the grid point spatial scale. Area-averaging over the East Asian domain, however, smoothes out evolving grid point scale differences in precipitation and evaporation which tend to cancel each other until the large scale forcing begins to force region-wide differences.

In VISM (Fig. 4.4(a)), we begin and end with two wet and two dry ensemble members. However, two years change sign relative to the VISM mean during the course of the summer monsoon (year 4 becomes dry after being initially wet, while year 3 becomes wet after being initially dry). Additionally, while beginning and ending as the driest of the four simulations, year 1 reaches the highest area-averaged soil moisture level of all the simulations in the first half of July. Since a wet state can become dry and a dry state wet, this indicates that the evolution of the hydrologic state is chaotic and independent of the initial global soil moisture.

Comparing panels (a), (b), and (c) in Fig. 4.4, we note that the ensemble mean hydrologic state from the relatively dry FISM, remains dry relative to VISM and FILS. Additionally, the ending hydrologic states for the initially dry FISM experiments are all drier than ending states for the VISM experiments. Recall that FILS was started from mean conditions, including soil moisture, while VISM was started from the 1 May, year 1 initial state, which was relatively dry over East Asian. Thus, continental scale soil moisture anomalies may persist beyond the monsoon season, which has implications for interannual variability in model soil moisture. The ending hydrologic states for VISM versus FILS seem to have about the same variance; VISM has a larger variance than FISM. This suggests that from an initial mean global land surface state, anomalies develop in response to internal variability of the persistent anomalies over time. We note that in both FISM and FILS (Fig. 4.4(b) and (c)), the initially constant root zone soil moistures show area-averaged differences of as much as 3% of saturation after the first simulation day. The differing initial atmospheric states; in particular air temperature, near-ground specific humidity, precipitation input, and cloudiness; are the causes for the initial divergence in hydrologic state. We note that East Asian regional soil moistures begin to diverge further from the initial synoptically induced variance during the last half of May for FISM, and in mid-to-late June for FILS. The later divergence in FILS is consistent with the additional control over soil moisture that fixing the full initial land surface state provides.

Figure 4.3: Time evolution of area-averaged East Asian root zone soil moisture for (a) varying initial soil moistures, identical initial states otherwise, (b) same initial soil moisture, varying initial state otherwise, and (c) same initial land surface state (all land variables). The four ensemble members for each experiment are indicated by the gray scale in the legend at the top of the figure. Heavy black lines indicate the ensemble mean root zone soil moisture. Units are percentage of volumetric saturation of the soil layer.

The evolution of the East Asian land surface state is sensitive both the initial global circulation and the initial global land surface state. With the initial land surface condition being fully constrained in experiment FILS, the differing initial general circulations result in variability comparable to VISM within two simulation months. Since the results from FISM diverge more quickly than those from FILS, the East Asian soil moisture state is less sensitive, at first, to changes in atmospheric P-E forcing when the initial land surface temperature, canopy moisture content, and snow mass are fixed, along with the soil moisture. Ultimately, the precipitation variability induced by the different atmospheric states, forces the initially identical land surface hydrologic states, as measured by soil moisture storage, to diverge.

Figure 4.4 shows the evolution of area averaged East Asian monsoon region land surface temperature Ts for the four ensembles and their members. On cursory inspection, VISM, which started with differing initial soil moisture but identical conditions otherwise, shows less variance generally than the other two ensembles with land-atmosphere coupling. Of the remaining two ensembles, FISM shows more variability than FILS. This is the result of the greater number of constraints placed on the initial land surface state in FILS than in FISM. At first, the ensemble variations in Ts are controlled by synoptic scale disturbances, as reflected in the large and differing swings in Ts for each ensemble member. Later during June to August (JU), Ts anomalies from the ensemble mean, correlate both to anomalies and rapid changes in regional soil moisture. For example, year 4 in FILS is consistently cooler and wetter in JJA than the ensemble mean. Year 3 in FISM is consistently warmer and drier.

4.4 Simulated intraseasonal East Asian summer monsoon variability

As discussed in previous sections, intraseasonal variability is of many temporal and spatial scales in the East Asian monsoon region, all of which influence the variability of the East Asian summer monsoon. To analyze the intraseasonal variability, a time series of smoothed ensemble average precipitation (with 30 day centered boxcar smoothing function applied) is calculated. Anomalies from the ensemble average are then calculated for each ensemble member, ignoring the first simulation month (May) to eliminate equilibration effects.

Then, time series of area averaged anomalous rainfall over the East Asian mainland and the adjacent ocean regions are determined for each ensemble member. We define 'land' as all land grid points from 22°-34°N and from 100°-130°E. 'Ocean' points are those grid boxes designated as ocean from 10°-18°N and from 100°-130°E in the GEOS-1/Mosaic coupled model. We create an East Asian Precipitation Dipole (EAPD) time series by differencing the precipitation anomaly time series over ocean from that over land. This results in a measure of the anomalous tendency for land precipitation, after removing the mean seasonal cycle of precipitation.

Figure 4.4: Same as Fig. 4.3 for the East Asian land surface temperature.

The time series are then smoothed to their five day centered average values, using a five-point centered boxcar filter. The standard deviation of the smoothed EAPD for each ensemble is then calculated. Six or more consecutive days with EAPD greater than plus or minus one standard deviation from the ensemble mean constitute a 'wet' or 'dry' case, respectively. The wet and dry cases are then composited for each ensemble to determine the regional characteristics of the dipole. Land surface variables and dynamical variables for wet versus dry cases will be compared.

An additional four member ensemble was run with the land hydrologically uncoupled (see Appendix B) from the atmosphere (ensemble A). The results for this ensemble were similar for precipitation and circulation anomalies to the coupled simulations and will not be shown. However, the implications of this similarity will be discussed in the concluding section of this chapter.

4.4.A The EAPD

Figure 4.5(a)-(d) shows the unnormalized EAPD values from June 1 through September 30 for each ensemble member of the FILS control. The other ensembles, including the uncoupled A ensemble, show similar behavior and are not shown. The standard deviation for the full ensemble is highlighted with a bold line on either side of the zero anomaly axis. Standard deviations of smoothed EAPD, for all ensembles, are shown in Table 4.2. Differences in EAPD variability are not statistically significant between the ensembles.

Table 4.2:
Sample standard deviation of smoothed (see text) EAPD for each ensemble experiment. Units are mm day-1. Ensemble experiment acronyms are as found in the main text.


Sample standard deviation (mm/day)










Figure 4.5: Smoothed time series of EAPD (see text) for the FILS ensemble members. Plus and minus one standard deviation from the centered average 30-day seasonal mean difference between land and ocean precipitation(for the entire ensemble) is indicated by the solid bold lines. The time series was smoothed with a 5-day centered boxcar smoother before the ensemble standard deviation from the seasonal mean was calculated. Units are mm day -1 for the ordinate and calendar day for the abscissa.



From the ensemble time series, 4 wet and 5 dry cases were chosen from FILS ensemble, 1 wet and 1 dry case from the VISM ensemble, and 2 wet and 5 dry cases were chosen from the FISM ensemble, for a total of 7 wet and 11 dry cases from all coupled ensembles. The cases are summarized in Table 4.3.


Table 4.3:
Wet ( bold) and dry (italics) EAPD cases from the mean daily solar radiation forced ensembles (font style indicates sign of EAPD phase). Dates are simulation days, year numbers give the ensemble member, from which the cases are taken. Criteria for selection of wet and dry cases can be found in the text.


Beginning in June

Beginning in July

Beginning in Aug.



7/21-26/2, 7/31-8/5/4



6/22-7/4/1, 6/6-11/2, 6/5-12/3

7/3-8/2, 7/7-28/3

8/10-15/1, 8/26-31/2


6/25-7/6/1, 6/22-30/2, 6/20-27/3, 6/30-7/5/3, 6/17-7/3/4

7/11-23/2, 7/22-27/3

8/13-19/1, 8/7-23/3

Composite data was weighted by the number of days spent in one phase for each case. Results are discussed below, with wet composite anomalies compared to the dry composite anomalies.


4.4.B EAPD time series analysis

Variance power spectra for EAPD were calculated for the members of all ensembles. The 30 day smoothed ensemble mean EAPD was removed for the variance calculation, before the spectral decomposition. Results for the FILS ensemble are shown in Figure 4.6. Spectra for the other ensembles are statistically indistinguishable from those of FILS, and are not shown.


Figure 4.6: FILS time series (left) and power spectra (right) for land precipitation for each ensemble member. Four year 30-day running mean for East Asian land precipitation was removed to calculate the autocorrelation function. Units for abscissa on left are day-1 , and (mmday-1)2 for abscissa on right (spectra were not normalized by the variance). Ordinates are simulation day and oscillation period on the left and right, respectively.



Three of the four ensemble members have a broad spectral peak, significant at the 95% confidence level as determined from red noise spectra with the same one-lag autocorrelation function (Gilman et al., 1963), in the 13-40 day range. However, we find no consistent placement of peak in either amplitude or period. This is confirmation of the chaotic nature of the EAPD oscillation noted in Lau and Bua (1998, in print) for monthly mean East Asian summer monsoon data.

The lack of distinguishing characteristics for the coupled versus uncoupled ensembles is not surprising if we consider the physics of land-atmosphere interaction with moist soils. As mentioned previously (Chapter 2), if the soil is so moist that evapotranspiration from vegetation can take place unimpeded (as is the case in these experiments), the land surface becomes hydrologically decoupled from the overlying atmosphere (Delworth and Manabe 1988, 1989). Then, the ensembles with land-atmosphere interaction behave similarly to the uncoupled ensemble. The resulting regional energy-hydrology schematic is somewhat different from that proposed in our original hypothesis (Figure 2.10) and will be discussed at greater length in section 4.4.


4.4.C EAPD wet and wet minus dry phase composites

In the next several sections, the composite wet and dry phase anomalies will be described, in terms of the land surface and atmospheric variables discussed in the hypothetical relationship in Chapter 2. Unless otherwise stated, wet and dry phase composites will group together all the coupled ensembles (FILS, FISM, VISM) to increase the total sample size for the composite. The A ensemble wet and dry phase anomalies, will not be discussed, since the land surface hydrologic forcing is small, and the dynamical and radiative anomalies are similar to those found in the coupled ensembles.

4.4.C.1 Precipitation anomalies

The first surface field we will examine will be the composite wet and dry phase anomaly fields of precipitation for the combined FILS, FISM, and VISM ensembles (Figure 4.7). The East Asian monsoon region precipitation anomalies for each ensemble considered individually are similar (not shown). Amplitude of the land-sea precipitation signal is greater in two of the three coupled ensembles than in the uncoupled ensemble (not shown). Each ensemble has a distinct land-sea precipitation dipole signal over East Asia and adjacent ocean waters to the south and east. The magnitude of composite seasonal anomalies over land and adjacent oceans is from 1 to 3.5 mm day-1 for the combined ensembles, with the East Asian spatial patterns reversed for wet versus dry. Shaded areas indicate anomalies greater than one standard deviation away from the seasonal mean for precipitation at that grid point.

Figure 4.7: Wet phase (a) and dry phase (b) precipitation anomalies for the FILS ensemble. Positive anomalies are indicated with solid contours. Negative anomalies have dashed contours. The zero contour is highlighted in bold, and the contour interval is 0.5 mm day-1. Areas with anomalies of greater than one standard deviation from the seasonal mean of all coupled ensembles are shaded. A nine-point (distance weighted) Cressman filter has been applied to the data.

4.4.C.2 Other hydrologic anomalies

We separate our analysis of other land surface variables into hydrologic and energy composites, to examine the behavior of the wet phase and dry phase hydrologic and energy cycles. Figures 4.8 and 4.9 show composites of coupled ensemble wet and dry phase hydrologic variable anomalies (evaporation and root zone soil moisture respectively) over the East Asian Monsoon Region.

Figure 4.8: Same as Fig. 4.7 for evaporation. Contour interval is 0.1 mm day-1.

Figure 4.9: Root zone soil moisture anomalies for wet and dry phases. Units are percentage of volumetric saturation. Contour interval is 2%. Negative regions are dashed, positive regions are shaded with solid contours, and the zero contour is highlighted.

Composite wet phase positive evaporation anomalies (Fig. 4.8) are consistently less than 1 mm day-1. over the East Asian monsoon region land mass, northern SCS, and southeast Asia, with dry phase anomalies of similar magnitude and opposite sign over the same regions. Anomalies of opposite sign from the East Asian monsoon region land are found over the southern SCS and adjacent maritime continent during the dry phase of the EAPD. Smaller anomalies of opposite sign are also found over Manchuria. During the dry phase, similar anomaly patterns of opposite sign are found.

The composite volumetric root zone soil moisture (RZSM) anomaly in both phases is consistent with the precipitation found above (Figure 4.9). The values of RZSM found in both wet and dry phases is sufficient almost everywhere to prevent vegetation stress from lack of moisture. Once more, the regions to the north and south of the East Asian monsoon region show anomalies of opposite sign to those found in the East Asian monsoon region. The implications of these high East Asian monsoon region soil moisture levels will be further discussed in section 4.5.

4.4.C.3 Surface energy cycle anomalies

The wet and dry anomaly latent heat flux (LHF) and sensible heat flux (SHF) are shown in Figure 4.10 and 4.11, respectively. Positive LHF anomalies during the wet phase are as large as 15 Wm-2 over East and Southeast Asia, and the adjacent oceans. In the dry phase, LHF is up to 18 Wm-2 less than average over East Asia and the immediately adjacent ocean waters. LHF is less in the monsoon trough region around 10°N and in the subtropical ridge at 25°-30°N. This is the result of reduced wind speeds in the monsoon trough region in the wet cases; circulation differences will be discussed below. Land north of the primary EAPD region also has reduced evaporation during the wet phase.


Figure 4.10: Same as Fig. 4.8 for latent heat flux. Units are W m-2. Contour interval is 3 W m-2.


Figure 4.11: Same as Fig. 4.10 for the sensible heat flux.



The SHF during the wet phase is very small over much of East Asia, demonstrating the dominance of LHF in the turbulent transport of heat upward from the surface in the East Asian region, and the contention that the land surface acts much like a sea surface during the wet phase. SHF ranges from slightly less than zero to 50 Wm-2 (not shown). Consistent with the inverse relationship usually found between LHF and SHF, the wet phase shows anomalously negative SHF in the EAPD region of up to 18 Wm-2 . North of this region, anomalously high SHF values appear, consistent with the inverse LHF/SHF relationship and the LHF pattern. During the dry phase, positive SHF anomalies are found over the East Asian monsoon region of as much as 21 W m-2, again with anomalies of the opposite sign to the north over Manchuria.


4.4.C.4 Cloud radiative forcing anomalies

Figures 4.12 and 4.13 show the anomalous wet and dry absorbed surface short wave radiation (ASWS) and outgoing long wave radiation (OLR), respectively. During the wet phase, ASWS is 5 to 20 W m-2. less than normal over the East Asian land mass. Anomalously positive values are noted to the north and south, particularly in the vicinity of the monsoon trough over southeast Asia, the SCS and eastward into the western Pacific. The dry phase anomaly map for ASWS shows positive anomalies of up to 15 W m-2 over the EAPD area, and negative values to the north and south. There is also a general decrease over the Indian Ocean of smaller magnitude. These results correspond well to the precipitation differences noted previously, and result from changes in convective cloud cover.

OLR values over the East Asian land mass during the wet phase are from 5 to 25 W m-2. As with ASWS, the dry phase values are consistent with the placement of the precipitation differences, with similar difference magnitudes between the East Asian monsoon region and regions to the north and south, especially in the monsoon trough region.

Figure 4.12: Same as Fig. 4.11, for observed surface absorbed short wave radiation. Contour interval 5 W m-2.

Figure 4.13: Same as Fig. 4.12 for outgoing long wave radiation.

4.4.C.5 Regional circulation anomalies

Figure 4.14 show the wet and dry phase circulation anomaly fields for the 200 hPa vector wind. Regions where the wind vector exceeds one standard deviation from the combined ensembles mean are shaded. In the wet cases, the East Asian land mass is dominated by an anomalous anticyclone, part of a larger anomalous anticyclone which covers the whole of South Asia with its axis at 25°-30°N. The upper tropospheric anticyclone results from the dynamical response to large scale convective heating. Anomalously strong tropical easterlies dominate to the south of 20°N. Anomalous upper tropospheric troughs are found east of the Caspian Sea and over northern China. In the dry phase anomaly maps of 200 hPa vector wind indicate a more localized response to shifting of precipitation from land to ocean. The east Asian trough is weakened, as indicated by the anomalous anticyclonic circulation over north China. The cyclonic anomaly over southern China results from reduced convective heating, as a result of the shift in the locus of convection from the mainland. Wind anomalies are in excess of 4 ms-1 in both the wet and dry phase. Note that in the dry phase, less of the region shows wind anomalies of greater than one standard deviation from the mean.

Figure 4.14: Vector wind anomaly at 200 hPa for (a) wet and (b) dry phase of EAPD over the Asian Monsoon Region. Scale of anomaly is found at the lower right of each panel. Units are ms-1. Shaded regions have zonal and meridional wind speeds of greater than one standard deviation from the mean zonal and meridional wind speeds.

Figure 4.15 shows wet phase anomaly winds for the lower troposphere (850 hPa). Anomalous anticyclonic circulations are found over the SCS and southeast Asia, and southwest of the Indian subcontinent. An elongated anomalous cyclonic circulation is found over central China to Korea. The result is stronger than normal monsoon westerlies over the northern Indian Ocean, and over southeast Asia and the Chinese mainland, as far north as Korea.

The anomalous anticyclone over the SCS may indicate a weakening of the climatological monsoon trough typically found from the Bay of Bengal eastward over southeast Asia into the SCS and western Pacific. Such weakening is consistent with the previously discussed changes in atmospheric branch of the hydrologic cycle, and in cloud forcing, found in this area.

Dry phase anomalies at 850 hPa, as at 200 hPa, are more local in scope and indicate a strengthened monsoon trough over the SCS and western Pacific. Monsoon westerlies are shifted south from their normal position to over the central and southern SCS. Moist air feeding into the Chinese mainland now is redirected further south. The East Asian monsoon region land mass is now dominated by anticyclonic anomalous circulation, with anomalous northeasterlies from Korea to southeast Asia. Anomalous 850 hPa wind magnitudes in excess of 2 ms-1 are found in both phases of the EAPD over East Asia.

Figure 4.15: Same as Fig. 4.14 for 850 hPa wind vectors.

4.5 Discussion and conclusions

We have examined a series of ensemble experiments, designed to determine the effect of initial atmospheric and land surface conditions on the EAPD and on the East Asian summer monsoon in general. Under the normal range of hydrologic and energy states in the coupled ensembles, there is little difference between the behavior of the EAPD, and the East Asian summer monsoon in general, with or without land-atmosphere coupling. An EAPD exists regardless of land-atmosphere interaction. Its amplitude is somewhat increased, perhaps, by land-atmosphere coupling, but its basic nature is independent of such coupling.

We find that the hydrologic and energy variables are consistent with the hypothesis in Chapter 2 (Fig. 2.10). The hydrologic cycle during the wet phase is characterized by increased total column moisture convergence over land and increased moisture divergence over the adjacent seas, as evidenced by P-E patterns (not shown). Interestingly, the slight evaporation increase noted over land is the result of two larger offsetting components in the total evaporation; the result of decreased evapotranspiration by plants, more than offset by increased canopy interception evaporation (not shown), consistent with the importance of canopy evaporation discussed in Scott et al., 1995. The energy cycle indicates slightly cooler land surface temperatures (not shown), resulting from slightly increased evaporation and reduced absorbed short wave radiation because of extensive convective cloud cover. Decreased net upward long wave radiation and SHF act as negative feedbacks on this cooling. Cloud radiative forcing during the wet phase indicates reduced OLR because of cold convective cloud tops during the wet phase of the EAPD.

The continental scale circulation observed during the wet phase of the EAPD indicate well developed anomalous anticyclonic circulations in the upper troposphere over South China and the Tibetan Plateau, consistent with the middle to upper tropospheric heating taking place as the result of the summer monsoon convection. The extratropical jet is displaced well to the north in its summertime position, though the upper tropospheric trough downwind of the Himalayan orography, along and east of the East Asian coastline, is enhanced during the wet phase. In the lower troposphere, the wet phase of the EAPD is characterized by increased cyclonic flow and convergence over East Asia. The monsoon trough over the Philippine and South China Seas is weakened or shifted north, redirecting moisture to the Chinese mainland.

While figures for a fixed evaporation ensemble simulation were not shown, they indicate that:

1. An EAPD exists even with fixed land evaporability,

2. The amplitude of anomalies for precipitation and circulation for the wet and dry phases are similar, and

3. The land surface variables do not indicate any particular spatial or temporal anomalies (not shown).

This leads to the most important point from this section: Land-atmosphere coupling is not required to force the EAPD, even though it tends to enhance its amplitude. We find chaotic behavior for the EAPD in both coupled and uncoupled ensembles. Under typical East Asian summer monsoon soil conditions, then, we modify the hypothesis presented in Chapter 2 (Figure 4.16).

Under this "moist soils" scenario, the largest portion of precipitation reaching the land surface will either go into canopy storage or run off, as the ground cannot absorb much additional moisture. As a result, soil moisture is little changed. Also, evaporation is not controlled by the land surface, since vegetation resistance is small at soil moisture levels above field capacity, and moisture recycling from the vegetation canopy is significant. Precipitation rates are controlled by regional moisture convergence, which in turn is controlled by radiative and dynamic processes, rather than land surface soil moisture anomalies. The existence of the EAPD does not require land surface interaction for its existence. It is an intrinsic mode of atmospheric variability, controlled by the interaction of internal dynamics with the convective/ radiative forcing. Fluctuations in moisture convergence thus take place between two climatologically favored regions for convection, the Chinese mainland and the SCS. Research by Ferranti et al (1997) has shown that the primary mode of intraseasonal precipitation variability found in the Indian and Southeast Asian monsoon regions, is also caused by interaction of radiative/convective forcing and internal dynamics.

Figure 4.16: Schematic of land-atmosphere hydrologic coupling, this time for moist soils. Since the soils are moist enough to effectively allow evapotranspiration at the potential rate, this effectively decouples the land surface from the overlying atmosphere. Thus, precipitation throughfall from the vegetation canopy to the land surface either becomes part of the soil moisture reservoirs or leaves the area as runoff. Under this' moist soils' scenario, precipitation variability over land is controlled mostly by dynamical and radiative processes. Symbols are as in Figure 2.10.


While we conclude that the EAPD is an intrinsic internal mode of variability for the East Asian summer monsoon, it is possible that the East Asian Monsoon can be significantly modulated in its amplitude and phase by changing the energy and boundary forcings. In Chapter 5, we will discuss the effect of the diurnal cycle of land-atmosphere hydrologic coupling and solar radiation on the East Asian monsoon and EAPD. Chapter 6 will discuss the effect of initial drought conditions over the Eurasian continent on the Asian monsoon, East Asian monsoon and EAPD.