6. The effect of initial Eurasian soil moisture on the East Asian summer monsoon



6.1 Experiment design

It has been shown by researchers that the Asian monsoon is sensitive to the snow mass over the Eurasian continent. In the reported results, the importance of snow mass is not in its reflective properties, but in its role as a land surface moisture reservoir affecting interannual variability of the Eurasian soil moisture (e.g., Barnett 1989, Yasunari 1991, Vernekar et al., 1995). As discussed earlier (section 2.1.B.), research has shown that the effect of anomalously wet Eurasian soils is to weaken the large-scale Asian monsoon, because of reduced land-sea thermal contrast. Anomalously dry Eurasian soils have the opposite effect on the Asian monsoon, because of increased land-sea thermal contrast. However, the effect of regional land-atmosphere interaction on regional monsoons such as the East Asian summer monsoon was not discussed in these studies. We will concentrate our efforts in this chapter on East Asian summer monsoon land-atmosphere interaction and its relationship to the larger scale general circulation.

Specifically, we examine the sensitivity of the East Asian summer monsoon to initial drought over the Eurasian continent, defined here as the region from 0°-180°E, 6°-74°N (heavy outline in Figure 6.1). The initial Eurasian soil moisture (ESM) was reduced in two experiments, each with four ensemble members. In the first, designated the ZISM ensemble, initial ESM was set to zero. A second ensemble, designated WISM, was run with ESM set to the total wilting level for each mosaic tile in each grid square, over the Eurasian domain. Appendix B explains the method used for the total wilting level calculation. In the drought experiments, the snow mass over Eurasia is left unchanged, to not involve initial snow mass variability in the analysis of monsoon evolution. The amount of snow mass on May 1 is shown in Figure 6.2. Examination of the time series of May 1 to June 30 Eurasian snow mass for the control (FILS) versus drought (WISM, ZISM) experiments indicates that there is little variability in the melting of snow mass during that time (not shown). Therefore, we can assume that snow mass has little effect on the level of ground wetness in the WISM and ZISM ensembles. We also note that the May 1 initial soil moisture state makes little if any difference is the rate of removal of area-averaged, May 1 snow mass (not shown).

Figure 6.1: World map showing large scale regions discussed in this chapter.

 

 

Figure 6.2: Initial snow mass water equivalent for the drought ensemble experiments. Contour interval is 30 mm of water equivalent.

 

 

In addition, much of the snow melts in all the experiments runs off because of a problem in modeling the deep soil temperature in Mosaic (Robock et al., 1995). The method of parameterization for heat flux to and from the soil, keeps the soil frozen longer than observed. Much of the water locked in the snow cover then becomes unavailable for percolation into the soil (Sud and Mocko, 1997, personal communication). This additionally minimizes the effect modulation of the Asian monsoon. For this reason, and the relative uniformity of snow of snow mass in contributing to Eurasian scale soil moisture variability, since much of the water from snow runs off. Barnett et al. (1989) and Vernekar et al. (1995) found that hydrologic effects of snow mass were most important in the mass removal, we can ignore snow mass effects on the monsoon in these experiments.

For the soil moisture sensitivity experiments WISM and ZISM, only the initial soil moisture state is changed. Otherwise, the experiment ensembles use the same initial land surface, SST, and atmospheric states as the FILS and FISM ensembles. Any differences in ensemble statistics are thus directly attributable to the reduction or removal of Eurasian soil moisture in the WISM and ZISM experiments, respectively, and the consequent effects on vegetation control over evaporation. For the WISM ensemble (Fig. 6.3(b)), we note that one ensemble member (year 3) reaches and even briefly exceeds the control mean ensemble root zone soil moisture level by the end of the monsoon season in September. However, the ensemble mean soil moisture level is less than in the control at the end of the simulated summer monsoon.

Changes in the mean Asian monsoon and East Asian summer monsoon states for WISM and ZISM compared to the controls (FILS, FISM) are discussed in section 6.2. Mechanisms for the changes in simulated Asian monsoon and East Asian summer monsoon are analyzed in section 6.3. Discussion and conclusions are presented in section 6.4.



6.2 Asian monsoon and East Asian summer monsoon mean climates with dry initial Eurasian soils

The analysis of changes in ensemble mean climates will begin with the monthly mean land surface state variables for the period May through July (MJJ) period for the WISM simulation. This covers the monsoon or Mei-Yu onset periods, and, as we will see, is the period of the summer monsoon most sensitive to drying of Eurasian soils.



6.2.A Hydrologic cycle

The hydrologic response to the initial soil moisture perturbation can be examined in a number of ways. Time series of area-averaged soil moisture values indicate the response of the land hydrology to the initial soil moisture perturbation, and gives a time scale for return to climatological soil moisture values. Figure 6.3 shows the evolution of the East Asian monsoon region soil moisture for the WISM and ZISM ensembles, compared to the FILS.

For the WISM ensemble (Fig. 6.3(b)), we note that one ensemble member (year 3) reaches and even briefly exceeds the control ensemble mean root zone soil moisture (RZSM) by the end of September. The WISM ensemble mean RZSM, however, is less than that of the control at the end of the monsoon season.

Figure 6.3: Time evolution of area-averaged East Asian root zone soil moisture capacity for (a) FILS control, (b) WISM (wilted initial soil moisture ensemble), and (c) ZISM (zero initial soil moisture ensemble). The initial land surface states were identical otherwise for each ensemble member (identified by the legend at the top of the figure). The ensemble mean is represented by the bold black line. Units are percentage of volumetric saturation.



For the ZISM ensemble (
Fig. 6.3(c)), no ensemble member started from the same initial state (other than soil moisture) as the control reaches the control ensemble RZSM. This indicates that hydrologically, the East Asian monsoon region does not recover to its normal soil moisture state within one monsoon season.

We note that the ensemble member starting from the same initial atmospheric state has the highest soil moisture level in each ensemble. This would seem to suggest that the initial atmospheric state may play a role in the subsequent evolution of the regional hydrology in the East Asian monsoon region, and in the character of the seasonal hydrologic forcing provided to the land surface by the atmosphere (i.e., P-E). For example, some of the WISM simulations return, on an area averaged basis, to a state where soil moisture does not play an important role in the continental scale climate of East Asia (See Fig. 4.20 on the land interaction with the global scale circulation when soil are wet.).

Since the evaporability of the land surface is the key variable in determining the land-atmosphere hydrologic coupling, and is non-linear around the initial and total wilting levels, average soil moisture values may not be an accurate measure of the soil moisture effect on this coupling. Another way of examining the soil moisture role in land-atmosphere coupling is to calculate the percentage area within a region where moisture levels at root zone and surface soil moisture stores are at the start-of-wilting levels (y1 in formula B.1 in Appendix B). The daily time series of percent of East Asian monsoon region with soil moistures above the level where wilting begins are shown for the control integrations (Fig. 6.4(a)), initially wilted ensemble (Fig. 6.4(b)), and the initially dry soil moisture ensemble (Fig. 6.4(c)). For the control integrations, the aerial coverage is close to unity for all ensemble members, as we expect with the hydrologic uncoupling of the land found in these integrations. We note that the ZISM ensemble shows the most variability in the aerial coverage of uncoupled land. WISM, surprisingly, shows similar percentages of uncoupled land to the control integrations for three of the four years by mid-June.


Figure 6.4: Fraction of East Asian monsoon region land area with soil moisture at or above field capacity for (a) control, (b) wilted initial soil moisture ensemble experiment, and (c) zero initial soil moisture ensemble experiment. Abscissa is days after commencement of monsoon experiments (May 1). Ordinate is fraction of total East Asian monsoon region with soil moisture in excess of field capacity. Identical shades of gray indicate ensemble members started from the same initial land surface state, other than soil moisture. The ensemble average is indicated by the black, bold line.

 

Figures 6.5 and 6.6 shows the mean ensemble precipitation for June over the Eurasian continent, northern Africa, and adjacent oceans, and the change in those variables between experiment and control for ZISM and WISM, respectively. The impact of initially dry soils is to reduce precipitation in two regions: the Eurasian extratropics and the East Asian monsoon region from the Southeast Asian peninsula to Korea and Japan. A vast area of the Eurasian extratropics have precipitation reductions from 1-3 mm(day)-1, while the East Asian monsoon region and northern India show decreases of as much as 5 mm(day)-1 in ZISM. These represent ensemble mean precipitation reductions in excess of 50%. The ensemble mean precipitation reductions show a similar spatial structure in WISM, but are reduced in magnitude by about a factor of 2 in the East Asian monsoon region and by 10-20% in the Eurasian extratropics.

In both experiments, the oceans adjacent to the East Asian monsoon region and the southern part of the Indian peninsula show increased precipitation, of similar amounts to the East Asian monsoon region and northern India precipitation decrease. The spatial structure of the ensemble mean precipitation reductions for experiment minus control, take on a distinctly zonal character over the East Asian monsoon region in June, when the Mei-Yu monsoon jump climatologically takes place.

 

Figure 6.5: Average June precipitation for control ensemble (top panel) and zero initial soil moisture ensemble (ZISM). The top panel is contoured at 2 mm day-1. The bottom panel is contoured at 0.5 mm day-1. In the bottom panel, positive contours are solid, negative contours are dashed, and shaded regions indicate positive differences. A nine-point smoothing function was applied to the data before plotting, using the Cressman weighting scheme.





Figure 6.6: Same as Fig. 6.5, for the control ensemble and the wilted initial soil moisture minus the control.



Control and experiment minus control changes in evaporation are shown in
Figures 6.7 and 6.8 for ZISM and WISM ensembles, respectively. The evaporation reductions are less than those for precipitation in the East Asian monsoon region (1-2 mm(day)-1), so total column moisture convergence is reduced. In the extratropics, the reduction in evaporation is almost as much as the reduction in precipitation (1-2 mm(day)-1), indicating that local moisture is very important to the hydrologic cycle in the summertime extratropics. From the hydrologic cycle data, it is clear that the hot and dry mode of the EAPD is quite active in these experiments, especially ZISM. Monsoon onset is especially affected in June, when the Mei-Yu rains normally become established over central China (See Chapter 2.2.).

Figure 6.7: Same as Fig. 6.5 for evaporation. Contour intervals are 1 mm day-1 for top panel and 0.5 mm day-1 for the bottom panel.


Figure 6.8: Same as Fig. 6.7 for evaporation. Contour intervals are 1 mm day-1 for top panel and 0.5 mm day-1 for the bottom panel.



Table 6.1 shows MJJ precipitation in the East Asian monsoon region for each ensemble member, and mean and sample variances of precipitation for each ensemble. Most apparent is the dramatic effect on ensemble members from drying the Eurasian soil. Two of four years in the WISM ensemble are more than two standard deviations less from the control mean. All ZISM ensemble members fall in this category, with three members at more than three standard deviations below the control mean. Clearly, the mean monsoon is affected by the degree of initial Eurasian drought.

For dry initial soils, MJJ precipitation variability increases as well. The variance is increased by a factor of a bit more than five over the control ensemble in the WISM, and by eight in the ZISM ensemble. This reflects the increasing importance land surface control has in altering the monsoon under conditions of initial drought (See Fig. 6.4.).

 

 

Table 6.1: Mean MJJ total precipitation P for the East Asian Monsoon Region (100°-130°E, 18°-34°N, land points only), for the control (FILS) and two sensitivity experiments (WISM, ZISM). Units are mm(day)-1 for yearly and mean values, mm2(day)-2 for variance. Bold type indicates 99.9% significance level for difference from control ensemble mean, and italics indicates 95% significance level for difference, using Student t test for small samples.

Ensemble

year 1

year 2

year 3

year 4

ensemble mean

ensemble variance

FILS

7.586

7.842

7.296

8.474

7.800

0.252

WISM

7.609

6.434

5.067

7.356

6.617

1.322

ZISM

6.487

5.197

3.348

3.818

4.712

2.015

 

Table 6.2 shows the same data for evaporation. The same general results obtain, with less evaporation for the experiment ensembles. Effects are larger for initially dry versus wilted soils for both ensemble means and ensemble variances. Three of four MJJ periods for WISM are more than three standard deviations below the control mean. All ZISM ensemble members are more than three standard deviations below the control mean.

 

Table 6.2: Mean MJJ total evaporation E for the East Asian Monsoon Region, for the control (FILS) and two sensitivity experiments (WISM, ZISM). Units are mm(day)-1 for yearly and mean values, mm2(day)-2 for variance. Same type face indications for statistical significance as in Table 6.1 are used.

Ensemble

year 1

year 2

year 3

year 4

ensemble mean

ensemble variance

FILS

4.556

4.686

4.582

5.110

4.734

0.066

WISM

4.391

3.858

3.024

3.693

3.742

0.318

ZISM

2.893

2.495

1.592

1.866

2.212

0.349

 

The mean and variation of area-averaged total P-E for MJJ over the East Asian monsoon region are shown in Table 6.3. The ensemble mean total column moisture convergence decreases as the soil dries, but not nearly to the extent that the precipitation and evaporation do. Interannual variation of P-E for the sensitivity experiments is once again somewhat larger, and larger for ZISM than for WISM. Two ensemble members exceed the control mean P-E in WISM, and one exceeds the control mean in ZISM. These ensemble members illustrate that the total column moisture convergence in some cases can compensate for the lack of local soil moisture, so that precipitation is not as deficient.

 

Table 6.3: Mean MJJ total atmospheric column moisture convergence (P-E) for the East Asian Monsoon Region, for the control (FILS) and two sensitivity experiments (WISM, ZISM). Units are mm(day)-1 for yearly and mean values, mm2(day)-2 for variance. Same type face indications for statistical significance as in Table 6.1 are used.

Ensemble

year 1

year 2

year 3

year 4

ensemble mean

ensemble variance

FILS

3.028

3.156

2.715

3.364

3.066

0.073

WISM

3.218

2.576

2.043

3.663

2.875

0.507

ZISM

3.594

2.701

1.756

1.952

2.501

0.697

 

In fact, for the two WISM ensemble members where P-E exceeds the control mean, the seasonal precipitation falls within two standard deviations of (but is still below) the control sample mean. However, when compensation from increased monsoonal circulation does not take place, statistically significant (95%)drought prevails. It should be noted that while two years have what can be considered 'normal' precipitation in WISM, there are no ensemble members in ZISM where total column moisture convergence is sufficient to bring MJJ precipitation to the 'normal' range (i.e., less than two sample standard deviations away from the control mean). So in terms of the hydrologic budget, it appears that both total atmospheric column moisture convergence, and the local land surface moisture sources, are important in maintaining the monsoon in the East Asian monsoon region.


6.2.B Land surface energy cycle

As discussed in Chapter 2, the hydrologic and energy cycles over land are coupled through the latent heat flux. In the soil moisture sensitivity experiments, and consistent with the reduced evaporation, we find increased sensible heat flux SHF (Figures 6.9 and 6.10 for ZISM and WISM, respectively) and surface temperature Ts (Figures 6.11 and 6.12 for ZISM and WISM, respectively) over Eurasia for June. The largest increases are found in the same regions where the hydrologic cycle is most affected (see. Figs. 6.5-8). Lack of available water for evaporation over land increases the amount of available absorbed short wave radiation (ASWS) for sensible heating. As a result, SHF, Ts, and the upward long wave radiation (ULWS, not shown) increase. The increased fluxes act as a negative feedback to the increased temperature, through increasing vertical turbulent and radiative heat transport from the land surface. However, the amount of ASWS also increases because of reduced cloudiness in the regions most affected by the reduction in soil



Figure 6.9: Same as Fig. 6.5 for sensible heat flux. Contour interval is 20 W m-2 for the top panel and 10 Wm-2 for the bottom panel.



 

Figure 6.10: Same as Fig. 6.6 for sensible heat flux. Contour interval is 20 W m-2 for the top panel and 10 W m-2 for the bottom panel.



Figure 6.11: Same as Fig. 6.5 for land surface temperature. Contour interval is 5K for top panel and 2K for bottom panel. Units are Kelvins.



Figure 6.12: Same as Fig. 6.6 for land surface temperature. Contour interval is 5K for top panel and 2K for bottom panel. Units are Kelvins.


moisture, by 20-50 Wm
-2 in the East Asian monsoon region (Figures 6.13 and 6.14 for ZISM and WISM, respectively). This represents a positive feedback to the surface temperature. Both WISM and ZISM show considerable persistence in drought over the Eurasian extratropics; in the East Asian monsoon region, ZISM shows more persistence than WISM, as the monsoon in the WISM simulation is able to reestablish the control mean climate more effectively in two ensembles through increased total column moisture convergence.

 

Figure 6.13: Same as Fig. 6.5 for the absorbed short wave flux at the surface. Units are W m-2, and contour interval is 20 W m-2 for the top panel, and 10 W m-2 for the bottom panel.



Figure 6.14: Same as Fig. 6.6 for the absorbed short wave flux at the surface. Units are W m-2, and contour interval is 20 W m-2 for the top panel, and 10 W m-2 for the bottom panel.

 

6.2.C Changes in ensemble mean general circulation

It is clear from the above discussion that the reduction in Eurasian soil moisture favors the hot, dry mode of East Asian monsoon region climate. This is reflected in the mean energy and hydrologic values for the East Asian monsoon region. How does the general circulation contribute to the generally reduced moisture convergence over the East Asian monsoon region?

There are significant differences in the ensemble mean tropospheric circulations between the soil moisture sensitivity experiments and controls. The June circulation differences are representative. Figures 6.15 and 6.16 show the ensemble mean, upper (200 hPa) tropospheric vector winds for the control and ZISM and WISM experiment minus control, respectively, for Eurasia. We note a Eurasian wide anticyclone-cyclone meridional dipole in the extratropics with anticyclone difference centers over Europe and central Asia at about 50°N, and cyclonic circulation difference centers to the east and west of the Tibetan Plateau, at about 25°-30°N. The extratropical features are consistent with anomalously early development of the July circulation in the mean ZISM and WISM ensembles, associated with an early jump in the Eurasian upper tropospheric jet to its highest latitude position. Meanwhile, the Tibetan Plateau cyclone features are associated with reduced convective heating because the monsoon advance is delayed. This difference dipole in the upper troposphere strengthens through the boreal summer, and is associated with the increased sensible and upward long wave heat fluxes over the Eurasian extratropics. Additionally, the difference dipole strengthens with height, indicative of warming through the depth of the troposphere (not shown).

Figures 6.17 and 6.18 show the June lower (700 hPa) tropospheric ensemble mean circulation for control (panel (a) in each figure) and ZISM and WISM minus control (panel (b) in each figure), respectively. In the lower troposphere, the response to soil moisture reduction or removal reflects the shifting of mass convergence from land toward the oceans, and the delay in monsoon onset over the Asian Monsoon region. The lower tropospheric summer monsoon westerlies are weakened by as much as 5 ms-1 (2 ms-1 in WISM) over regions where, climatologically, lower tropospheric monsoon westerlies prevail, and the monsoon trough over and to the east of the Bay of Bengal, over southeast Asia, and into the South China Sea and western Pacific Ocean is shifted southeast from its position in the control ensemble.

 

6.3 Mechanisms for delayed or missed East Asian monsoon region onset

6.3.A Moisture flux convergence

The moisture flux convergence (MFC) was calculated for the mean ensembles as the sum of the time tendency of the precipitable water W, less the difference between precipitation P and evaporation E (after Peixoto and Oort 1992):

(6.1)

where Q is the vertically integrated moisture flux vector; h is the operator i /x + j /y, where is the partial differential and i and j are unit vectors in the zonal and meridional directions, respectively; and represents the dot product between two vectors.

Figure 6.15: June, ensemble average wind vectors at 200 hPa for the FILS control (top panel) and the zero initial soil moisture experiment minus control(bottom panel). Units are m s-1. Grey shading and length of arrow compared to grayscale bar to right, and arrow at lower right, respectively, indicate magnitude of the wind and difference in wind vectors in each panel.

 

 

Figure 6.16: Same as Fig. 6.15 for the WISM (wilted initial soil moisture) ensemble experiment.


 

Figure 6.17: Same as Fig. 6.15 for the 700 hPa wind vectors.


Figure 6.18: Same as Fig. 6.16 for the 850 hPa wind vectors.



Control ensemble mean MFC and the difference between the ZISM and WISM ensembles minus the control, are shown in
Figures 6.19 and 6.20 respectively. The ensemble mean for the control integrations indicates maximum moisture flux convergence over southern India, Southeast Asia, the adjacent South China Sea, and the equatorial central Indian Ocean, with MFC in excess of 4 mm(day)-1. The entire southern hemisphere oceans exhibit negative MFC, and act as a moisture source for the northern hemisphere monsoon regions.

For the drought experiment minus control ensemble mean MFC, we note increased MFC over the northern extent of the MFC maximum over the South China Sea (about 1-2.5 mm day-1) as well as a strong increase in MFC over the southern half of India (up to 4 mm day-1), in a region where control MFC was already 8 to 10 mm per day. These increases in maximum MFC are at the expense of MFC over the land and ocean regions surrounding them. Toward the landward fringes of the Asian monsoon region, MFC decreases by 1-2 mmday-1, while over the adjacent oceans decreases are from 1-3 mm day-1 with maximum decrease over the southern part of the South China Sea. Control ensemble MFC variability (not shown) is largest over the South China Sea, the Chinese mainland, and the Bay of Bengal near Burma. Individual ensemble members for years 3 and 4 tend to exhibit a meridional

Figure 6.19: Same as Fig. 6.15 for moisture flux convergence. Units are mm day-1. Contour interval is 2 mm day -1 in the top panel and 1 mm day -1 in the bottom panel. In the bottom panel, positive changes in moisture flux convergence are shaded and have solid contours, while negative regions are dashed.



Figure 6.20: Same as Fig. 6.16 for moisture flux convergence. Units are mm day-1. Contour interval is 2 mm day -1 in the top panel and 1 mm day -1 in the bottom panel. In the bottom panel, positive changes in moisture flux convergence are shaded and have solid contours, while negative regions are dashed.

Figure 6.21: FILS control ensemble (a, c, e, g)and zero initial soil moisture (ZISM) minus control (b, d, f, h) moisture flux convergence for individual ensemble members. These experiments initially differ only by their soil moisture states. Contour intervals are as in Fig. 6.19.



anomaly structure similar to that of the EAPD, while years 1 and 2 seem to show zonally asymmetric anomalies of opposite sign between each other.

The ZISM experiment minus control, mean June MFC differences for matched years from each ensemble (Figure 6.21a-h) are more zonal in structure than the within ensemble anomalies from the ensemble mean for the control (not shown). MFC differences over the East Asian monsoon region and adjacent oceans show a negative/positive/negative structure, with positive anomalies over the south China coast for two simulations and over the northern South China Sea for two others. These are flanked by negative anomalies to the north over mainland China and to the south over the central and southern South China Sea.



6.3.B Moist static energy

The effect of the local land surface, and of the reduction in moisture convergence, on the local static stability is significant. The vertical structure of static stability is usually measured by calculating the moist static energy (MSE), which is given by:

(6.2)

= (IE + PE + LE)

 

where cp is the specific heat at constant pressure (1004 J kg-1 K-1), T is the temperature (K), g is the gravitational constant (ms-2), z is the height of a constant pressure surface (m), Lv is the latent heat of vaporization (2.5 x 106 J kg-1), and q is the specific humidity (kg kg-1).

Term 1 in equation 6.2 represents the internal energy (IE) of an air parcel, term 2 represents the potential energy (PE), and term 3 is the moist or latent energy (LE). Figure 6.22a-h shows the difference in each component and total mean MSE from 1000 to 300 hPa for the ZISM experiment ensemble versus the FILS control ensemble, zonally averaged from 100°-130°E, and from 10°S to 60°N, for the month of June. Differences for the WISM ensemble are of similar structure but smaller magnitude (not shown).

Fig. 6.22(b) shows that IE in general is increased in the experiment versus the control (Fig. 6.22(a)), particularly at lower levels. This is consistent with the increase in sensible heat which results from reducing or removing the initial soil moisture. The effect becomes larger with increasing latitude, and is noticeable at increasing heights north of the tropics. This may be the result of better efficiency in vertical transport of sensible heat in the middle latitudes. Over the East Asian monsoon region (22°-34°N), we note an interesting feature: in the PBL, the decrease of IE with height is less in the experiment than in the control, indicating that there is increased stability in the temperature structure in the dry soil moisture ensemble.

Fig. 6.22(d), indicates the change in LE for the experiment minus control (Fig. 6.22(c)). The decrease in LE in the lower troposphere over land dominated latitudes is striking. The decrease in PBL LE is as much as 7,000 J kg-1 over land between 20°-40°N. Overlying the relatively drier air in the PBL is a moister layer in the lower to mid-troposphere from 5°-40°N, with increased latent energy of up to 1,000 J kg-1. Thus zonally averaged, land based convective instability below 700 hPa is markedly decreased in the experiment versus the control. The mechanism for the drying of the PBL is the lack of evaporation from the local land surface and reduced MFC (see section 6.3.A). Moistening above 700 hPa with dry initial soils may be a function of the convective parameterization; with less convection, there is less subsidence between parameterize convective towers, and thus less drying of the middle troposphere. The control potential energy (PE) and difference in PE between experiment and control are shown in Figs. 6.22 (e) and (f), respectively. Note that the differences are about two orders of magnitude less for PE than for either IE or LE, with largest differences in the far northern extratropics. Thus, there is little involvement of the PE in determining the static stability in the East Asian monsoon region.

Finally, Figs. 6.22 (g) and (h) shows the control, and experiment minus control difference, in total MSE. The 100°-130°E zonally averaged vertical structure of MSE is dominated by variations in latent energy, which take place mainly over land. Everywhere poleward of 5°N, MSE is decreased in the PBL and increased in the layer above it. Thus the ZISM ensemble mean static stability is increased, in spite of higher temperatures in the PBL, because of the reduced moisture in the PBL.



 

Figure 6.22: Ensemble mean latent, internal, potential, and total moist static energy for (a, c, e, g) FILS control and (b, d, f, h) the zero initial soil moisture minus control. Contour interval is 104 J kg-1 in (e), 5000 J kg-1 in (a), (c), and (h); 1000 J kg-1 in (b), (d) and (h), and 100 J kg-1 in (f). For differenced fields, positive differences are shaded and use solid contours, and negative contours are dashed.



6.3.C Convective heating, cloudiness, moistening changes

The change in distribution of MSE noted in the previous section would be expected to result in reduced convection over the East Asian monsoon region. In this section, the June ensemble mean convective heating and moistening will be analyzed for the dry initial land surface ensemble versus the control ensembles. We note here that the results for the wilted initial soil moisture experiment, while of lesser magnitude, are again similar in structure (not shown).

Figure 6.23 shows the June ensemble mean convective heating for the control simulation, zonally averaged from 100°-130°E, from 10°S to 62°N, and for the ZISM ensemble minus the control. We find maximum convective heating at two vertical levels, the PBL and middle troposphere, in the same latitudes as the convective cloudiness maximum. Heating rates of 2-4K day-1 are found. For the experiment minus the control, we find reduced heating rates of 0.5-2K day-1 in the areas of maximum heating in the control, with the largest reductions found in the PBL (where the reduction in latent energy is largest).

Finally, Figure 6.24 depicts the June ensemble mean convective moistening for the control simulation, again zonally averaged from 100°-130°E, from 10°S to 50°N, and for the ZISM ensemble minus the control. The control pattern shows moistening minima of -1 to -6 mm day-1 in the PBL. The magnitudes of these minima are reduced by 1 to 2 mm day-1 in the ZISM experiment.



Figure 6.23: Ensemble mean, zonally averaged convective heating over the region from 100°-130°E and from 10°S-62°N for (a) the FILS control, with areas with over 3K day-1 heating shaded, and (b) the difference between the ZISM experiment and the control, with positive differences shaded and with solid contours, negative regions with dashed contours, and the zero contour highlighted. Contour interval is 1K day-1 for (a) and 0.5 K day-1 for (b). Units throughout are K day-1.





Figure 6.24: Same as Fig. 6.23 for the convective moistening. Areas with less that -3 g kg-1 day-1 in (a) are shaded. In (b), positive differences between the ZISM experiment ensemble minus the FILS control are shaded, the negative contours are dashed, and the zero change contour highlighted. Units are g kg-1day-1.


6.3.D Changes in dynamics associated with East Asian monsoon advance

The zonal wind jumps associated with the East Asian monsoon sudden progressions to the north (See Chapter 2.2.A.) are changed by the respective reduction or total removal of Eurasian soil moisture in the WISM and ZISM experiments. The zonal jet maximum moves poleward more quickly, and spends less residence time in the Mei-Yu position (the first jump) in the experiments than in the control. This can be seen in the ensemble mean, 90°-140°E zonal winds at 200 hPa from FILS, and the difference between ZISM and control (Figure 6.25). Figure 6.26 shows the same information for the control versus the WISM ensemble. The positions for the zonal jet over the East Asian monsoon are marked with heavy solid lines in the control in both figures.

The experiment minus control difference in zonally averaged zonal wind for the two different experiments are remarkably similar, with the ZISM differences larger than those for WISM. However, the latitudinal positions and timing of the ensemble mean differences are almost identical. Note in the control ensemble mean that the two monsoon jumps during the summer monsoon, while too far north, represent a model rendition of an observed dynamical process, discussed in Chapter 2.2. In the WISM ensemble, the first jump is weaker and further north, while in the ZISM ensemble mean, the first jump is skipped almost entirely.

 

Figure 6.25: Zonal average of ensemble mean 200 hPa zonal wind from 90°-140°E and from 10°S-74°N for (a) FILS control and for (b) ZISM ensemble minus control, from simulated May 1 to August 31 dates. Contour interval is 5 m s-1 in (a) and 2 m s-1 in (b). In (a) and (b), the negative (easterly) regions dashed, and the zero contour highlighted. In (a), regions with greater than 20 m s-1 westerlies are shaded, and the latitude of the maximum zonal wind is marked with a solid line. In (b), positive values are shaded.



Figure 6.26: Same as Fig. 6.25 for the control ensemble mean, and wilted initial soil moisture ensemble experiment ensemble mean minus control.


The change in the circulation may be considered a large scale remote forcing similar to that provided by ENSO SST anomalies, which resulted in anomalous conditions over the East Asian monsoon region, repeated for several years. In these experiments, however, the anomalous boundary forcing is provided by the lack of soil moisture over the Eurasian continent, and the subsequent large increase in land surface temperature in both the East Asian monsoon region and the Eurasian extratropics. Additionally, the local dry soil moisture condition may feed back on the large scale forcing provided, resulting in the shortening or skipping of phases of the East Asian monsoon region onset, which contributes to relative drought. The relative contributions of the East Asian monsoon region, compared to that of the Eurasian, soil moisture removal cannot be assessed from the current experiments, since the simulated initial drought condition is continent-wide. In Chapter 7, an experiment designed to test local versus remote soil moisture sensitivities will be described.



6.4 Discussion and conclusions

In this chapter, we examine the effect of Eurasian scale drought on the Asian monsoon and East Asian monsoon. Two four-member ensemble experiments were performed: one in which the initial soil moisture condition was set to the total wilting level for the parameterized vegetation within each Eurasian grid square (Figs. B.1 and B.2, Appendix B), and another in which initial Eurasian soil moisture was removed entirely.

We find that generally, the mean hydrologic state in both experiments is significantly changed over both the Eurasian extratropics, and the East Asian monsoon region. In the experiments, mean root zone soil moisture does not recover to mean control values over the course of a monsoon season. The risk and degree of persistence of drought in individual ensemble members are dependent on the degree of initial Eurasian drought. We note also that the reduced East Asian monsoon is in spite of the increased land-sea temperature contrast. The Eurasian land surface temperatures are increased by as much as 10K in the extratropics and 6K in the East Asian monsoon region and northern India.

Ensemble mean precipitation and evaporation decrease with initial Eurasian drought in the East Asian monsoon region and the Eurasian extratropics. Ensemble mean moisture flux convergence (MFC) is reduced over the East Asian monsoon region in the ensemble experiments as well. The more extreme the initial drought, the larger the drying, especially in the East Asian monsoon region. Precipitation variability is increased in the East Asian monsoon region with initial drought conditions. Two of four ensemble members from the wilted initial soil moisture experiment have monsoon precipitation which, while drier than the control mean, are not statistically distinguishable from the control. In these cases, total column moisture convergence (P-E) is larger than the control mean (Table 6.3) and is the compensating mechanism for the initially dry hydrologic state of the East Asian monsoon region.

The mechanism for the weakened monsoon is the reduction of MSE in the lower troposphere, and the increase in MSE in the middle troposphere. Moist convective processes are inhibited, and the mid to upper tropospheric convective heating required to set up the summer monsoon circulation is delayed and weakened. This is clearly evident in the June zonal mean profiles of convective heating, and moistening, which show reduced convective activity over land-dominated latitudes (20°-34°N) between 100°-130°E. A similar mechanism was found by Eltahir et al. (1996) to operate over the African monsoon region to regulate intraseasonal precipitation variability.

Change in regional dynamics of the East Asian monsoon also plays a role. The Mei-Yu monsoon transition, which represents monsoon onset in central China, is evident in the control ensemble mean, but is reduced in residence time or eliminated entirely in the experiment ensemble means and ensemble members (Figs. 6.25 and 6.26). The relative roles of the East Asian monsoon region scale versus the Eurasian scale dryness in the change in monsoon climatology, however, cannot be assessed. However, it would be consistent to assume a non-linear interaction between the land surface forcing in the East Asian monsoon region and forcing by the large scale general circulation induced by Eurasian initial drought. This forcing would be analogous in scale to that found from anomalous SSTs, but extratropical to tropical in nature, rather than tropical to subtropical/extratropical.