5.1 Experiment design
The role of convection in development and maintenance of the monsoon has long been recognized. It has also been shown that the diurnal cycle is intimately connected with convection, both over land and over the oceans (e.g., Gray and Jacobson 1977). In this chapter, we examine the role of the diurnal cycle of energy input at the land surface on the East Asian summer monsoon, to determine if the diurnally varying energy input affects its onset, maintenance, and variability.
A pair of four member ensembles, started from the same set of initial conditions as the VISM and A ensembles in Chapter 4, were run with mean daily, rather than diurnally varying, solar radiation. Thus while the diurnal variability in energy input was removed, the total energy input was unchanged. Comparison to the diurnally forced experiments from Chapter 4 will help determine whether diurnal variation of energy input over land affects East Asian summer monsoon onset, maintenance, and variability. We will also be able to determine whether effects of diurnal energy variations are changed through removal of land-atmosphere hydrologic coupling. The mean monsoon, East Asian monsoon advance, and the behavior of the EAPD with mean solar forcing, and the comparison to ensembles with diurnal forcing, will be discussed in section 5.2. Further discussion and conclusions can be found in section 5.3.
5.2 Diurnal cycle of solar radiation and the East Asian summer monsoon
5.2.A Asian monsoon dynamics, energy, and hydrological cycles without diurnal cycle
We now perform an analysis of large scale Asian monsoon and East Asian summer monsoon variability as in Sections 4.3 and 4.4. The M1 index described in Chapter 4 is calculated for each non-diurnal ensemble. There is a hint of longer lasting excursions from the ensemble mean M1 when the diurnal cycle is removed. Otherwise, removing the diurnal cycle of incoming solar radiation seems to make little difference in the mean large scale Asian monsoon (not shown).
We next check the evolution and variability of the East Asian monsoon region hydrologic and energy balance without diurnal energy input. The time series of ensemble member and ensemble mean, area-averaged root zone soil moisture evolution over the East Asian monsoon region for VISMND and VISMD are shown in Figure 5.1(a) and (b), respectively. VISMND shows a tendency for larger excursions from the ensemble mean without diurnal forcing, early in the monsoon season. The VISMND simulation also shows a tendency for more extreme anomalies than VISMD, and a stronger tendency to July drying of the root zone soil moisture. There are too few ensemble members, however, to make more than a qualitative statement regarding variability in this sense.
Figure 5.1: Time evolution of area averaged East Asian root zone soil moisture for (a) varying initial soil moisture, identical initial state otherwise with no diurnal cycle; (b) same initial soil moistures, varying initial state otherwise, including diurnal cycle. Ensemble mean is bold line, and ensemble members are indicated in the legend at the top of the figure. The time series runs from May 1 to September 30. Solid vertical lines mark the beginning of each month, with dashed vertical lines indicated the mid-month point. Units for soil moisture are percentage of volumetric saturation.
Figure 5.2 (a) and (b) shows the evolution of area-averaged East Asian monsoon region Ts. Again there is a tendency for earlier excursions from the ensemble mean in the non-diurnal ensemble, and for a warmer ensemble mean in July, consistent with the drier soils found then. Anomalies in Ts thus seem to be driven at least in part, as would be expected, by soil moisture anomalies; we note in Fig. 5.2(a) that year 1 Ts is warmest and driest. However, the change in soil moisture also seems to be a function of Ts; year 3 is warmest during the first half of July for the non-diurnal ensemble, though it is the wettest. The decrease in soil moisture for year 3 during this warm period during first half of July, however, is dramatic, consistent with the development of warm, dry conditions.
Figure 5.2: Same as Fig. 5.1 for the East Asian land surface temperature.
An examination of the EAPD under mean daily solar forcing indicates that there is no significant change in that mode of precipitation variability, either in amplitude, spatial scale, or the associated land surface and general circulation anomalies (not shown).
5.2.B Indian and East Asian summer monsoon precipitation advance
In this final section, we examine the effect of removal of the diurnal solar cycle on monsoon advance. In general there does not appear to be much impact on the intraseasonal variability of the established monsoon when the diurnal cycle of incoming solar radiation is removed. However, there appears to be a reduction in the broad scale Asian monsoon during the May-June onset period (MJ). The five-day centered average ensemble mean M1 for VISM ND is reduced by as much as 3.5 m s-1 in late May, and by 1 m s-1 in mid-to-late June. During the 31-day period from May 15 to June 15, the ensemble mean M1 for VISM ND (13.30 m s-1) is almost 1.5 m s-1 less than the grand mean for the four diurnal ensembles, which have a sample standard deviation of 0.28 m s-1 (Table 5.1). Reduced VISM ND mean M1 recovers to the VISM D control value by July 1. In fact, July is marked by stronger mean M1 in the non-diurnal ensemble.
5.1: May 15-June 15 M1 for each ensemble member, and total
ensemble means (in bold print). Acronyms are as defined in
4.1, or in this chapter.
Units are m s-1. Statistics for all diurnal ensembles are
indicated in the bottom row of the table for determination of
statistical significance of difference in non-diurnal experiment
ensemble mean compared to diurnally forced experiments.
A FILS FISM VISMD VISMND #1 14.03 15.91 13.28 13.28 12.05 #2 14.95 15.25 16.36 15.65 13.6 #3 14.87 13.93 14.49 15.18 13.74 #4 15.84 14.55 13.35 13.99 13.83 Ens.
mean 14.92 14.91 14.37 14.52 13.3 Data for diurnal
ensemble means: Grand
ensemble mean= 14.77 Grand
ensemble std. dev.= 0.279
Data for diurnal ensemble means:
Grand ensemble mean= 14.77
Grand ensemble std. dev.= 0.279
The MJ period marks the time of rapid monsoon advance. The following discussion will highlight the May 15-June 15 period. Figure 5.3 (a) and (b) shows the precipitation fields for VISM D , and the difference fields for VISM ND minus VISM D . The control experiment (with diurnally varying solar radiation) shows heaviest rainfall over southern India and adjacent Arabian Sea, the Bay of Bengal, and the East Asian land mass. The effect of removing diurnal variations of solar radiation is generally to shift the ensemble mean precipitation oceanward, i.e., slow the advance of the Asian monsoon in its subregions, as well as at the large scale.
Figure 5.3: Ensemble mean for VISMD (control) and VISMND (experiment) minus control for May 15 to June 15 precipitation (monsoon onset). Contours are 2 mm day-1 for (a) and 1 mm day-1 for (b). Positive differences are shaded with solid contour lines, negative contours are dashed, and the zero contour is highlighted in bold.
Figure 5.4 (a) and (b) shows the 200 hPa winds for VISM D and the difference between experiment and control for the May 15-June 15 period. In the control, the circulation is dominated by an east-west oriented anticyclone with its axis at 15°-20°N latitudes. The extratropical westerly jet is to its north, while the upper tropospheric monsoon easterlies are found to its south. There are ridges in the extratropics at 65°E and 120°E in both control ensembles, with a trough at about 90°-100°E. The difference maps for the coupled and uncoupled ensembles show cyclonic circulations in the ridge positions and anticyclonic circulations in the trough positions, indicating that the upper tropospheric waves are more amplified in the control (diurnally varying) ensembles.
Figure 5.4: Same as Fig. 5.3, for 200 hPa winds. Contours indicate wind velocity, as do relative size of vectors to the arrows below each panel. Units are m s-1. Contour interval in (a) is 10, 15, 20, 25, and 30 m s-1. Contour interval in (b) is 5.0, 7.5, 10.0, 12.5, and 15.0 m s-1.
Figure 5.5 (a) and (b) shows the 850 hPa winds for the control ensemble and difference between experiment and control ensembles over the same period as Fig. 5.4. The control shows the development of the Somali Jet over the Arabian Sea to the west and southwest of India. The monsoon westerlies are in evidence over the southern half of the Indian peninsula, the Bay of Bengal, and northeastward through southeast Asia and China. The difference maps for the experiment ensemble less the control indicate a distinct weakening of monsoon westerlies, and a general reduction in cross equatorial flow from south to north.
Figure 5.5: Same as Fig. 5.4 for 850 hPa. Contour interval in (a) is 10.0, 12.5, and 15.0 m s-1. Contour interval in (b) is 2.5, 5.0, and 7.5 m s-1.
Finally, Figure 5.6 (a) and (b) shows the mean vertical profile of moist static energy (MSE) from May 15 to June 15, including all land points zonally averaged from 100°-130°E, from 10°S-50°N, for the mean daily solar radiation for the control ensemble and the experiment minus control. The top panel shows the zonally averaged MSE over land for VISMD, while the bottom panel shows the difference in zonally averaged MSE over land for VISMD minus VISMND. We note that mean daily MSE for the May 15-June 15 period is actually larger at pressures higher than 800 hPa when solar radiation is prescribed to its mean daily value. This increase in MSE results from both increased latent energy (Lvq) and internal energy (cpT), and indicates that the mean daily moist static stability is greater when the diurnal daily solar radiation, rather than its mean diurnal cycle, is used for forcing. This apparent paradox can be explained by considering the diurnal cycle of MSE and the non-linear relationship between static stability and convection, akin to an 'on/off switch'.
Figure 5.6: Zonally averaged mean May 15-June 15 moist static energy (MSE) difference field for land points between non-diurnal and diurnal forcing in (a) coupled and (b) uncoupled ensembles. Units are in joules. Positive regions are shaded with solid contours, negative regions are indicated by dashed contours.
Since the precipitation advance and the monsoon onset during this period, have been shown to be delayed when the diurnal cycle is disabled, this would suggest reduced convection during the onset period when there is no diurnal solar forcing. We hypothesize that, while the mean land surface temperature is warmer and the mean boundary layer somewhat moister with mean daily solar forcing, the threshold for convection is reached less frequently because the diurnal variability is removed. Diurnal variation in heating and evaporation from the land surface, results in greater instability during the solar insulation maximum than when there is mean solar forcing. Convective instability thresholds may be reached less frequently without the diurnal pulse of solar radiation.
That the May 15-June 15 difference in ensemble means indicates a delay in onset, rather than a weakening of the total monsoon, is not yet established. To do so, we calculate area-average time series for precipitation for two critical south Asian monsoon regions; the East Asian monsoon region and the IMR. Figure 5.7 shows the pentad smoothed time series of ensemble mean precipitation for five to 90 days after the start of the integrations (May 1) over the East Asian monsoon region (top panel), defined as all land points from 22°-34°N and 100°-130°E. In the bottom panel we show the same data for the IMR (10°-22°N, 70°-90°E).
Figure 5.7: Ensemble mean precipitation time series for area averaged land precipitation. In (a) we show the East Asian Monsoon Region (100°-130°E, 22°-34°N) and in (b) the Indian Monsoon Region (70°-90°E, 10°-22°N). Solid line is diurnally forced ensemble (VISMD) and the broken line is the mean daily solar radiation-forced ensemble (VISMND). Units on the ordinate are mm day-1. Units on the abscissa are days after the May 1 simulation starting date.
In both subregions, there is evidence that removal of diurnal solar forcing results in a delay in the commencement of monsoon rains. In the East Asian monsoon region, the late May and early June period are particularly affected. The peak rains (over 12 mm day-1. ) in the control occur around late May. This peak is delayed to mid-June without diurnal solar forcing, and reduced by over 2 mm day-1. However, subsequent precipitation is not appreciably different in the experiment versus the control. In the IMR, the delay in ensemble mean onset is striking, and of long duration. Indian monsoon onset in the control appears to be in late May, with a further jump in precipitation in late June. In the non-diurnal experiment, these jumps are delayed by 10-20 days throughout, with the secondary maximum in June somewhat reduced from that found in the control. It is not until mid-July that IMR precipitation in the experiment reaches the levels found in the control.
These results are consistent with the oceanward shift of ensemble mean precipitation for May 15 to June 15 (Fig. 5.3), the cyclonic circulations in the wind vector difference fields at 200 hPa straddling the Tibetan Plateau (Fig. 5.4), and the weakened monsoon westerlies at 850 hPa (Fig. 5.5). Clearly these circulation changes are the result of reduced convective heating when the diurnal solar forcing is removed. A similar result during the early part of the monsoon season is found for diurnal versus non-diurnal experiments when the land and atmosphere are uncoupled (not shown).
5.3 Discussion and conclusions
Convection is an important driver for the large scale Asian monsoon and its regional components. Since convection is strongly influenced by the diurnal cycle, as evidenced by its strong diurnal component in the tropics and summertime extratropics, we have examined the effect of two aspects of the diurnal cycle on the East Asian monsoon.
Then, a coupled and uncoupled ensemble set were run with the diurnal cycle of short wave radiation removed. These were compared to ensembles with diurnal solar forcing started from the same initial states. The main results from comparing these ensembles were as follows:
1. The ensemble mean broad scale Indian monsoon, as measured by the WY horizontal shear index, is delayed.
2. The onset of Indian monsoon rainfall is delayed by one to two weeks when the diurnal cycle of solar radiation is removed. The same peak monsoon strength is still attained, but at a somewhat later date.
3. The Mei-Yu rainfall for the East Asian monsoon is reduced, but subsequent rainfall is unchanged.
The mechanism for the delay and/or weakening in summer monsoon precipitation is delay in reaching of the threshold for commencement of convection, because of the removal of the diurnal cycle of solar radiation. This is illustrated in Figure 5.8, and discussed below.
The commencement of convection over land is vital to establishing the monsoon. The diurnal variability in energy input from the sun allows the planetary boundary layer to exceed the threshold of convection more frequently than would take place if a steady, seasonally varying forcing is applied. The convective parameterization uses the change in moist static stability to trigger convection, a reasonable parameterization of the actual physical process. So the delay in convection in the model seems to be reasonable, and thus the monsoon onset, which is controlled by establishment of moist convective heating over land, is also delayed.
In the case of the East Asian Mei-Yu phase of the monsoon, the model results suggest that the diurnal cycle is important in the amount of model precipitation which takes place over East Asia during June. Since 80-90% of the precipitation over this region during this time period is generated by the convective parameterization, this is not surprising. Given the additional involvement of extratropical forcing in this phase of the monsoon advance, however, the applicability of the results to the East Asian monsoon is somewhat less clear.
Figure 5.8: Schematic of the mechanism responsible for delay in monsoon onset when the diurnal cycle of incoming solar energy is removed. The thick broken line marks the amount of moist static energy in the planetary boundary layer required to start convection, given the vertical structure of the temperature and moisture in the atmosphere. The thin straight line marks the moist static energy in the planetary boundary layer with mean daily solar forcing, while the thin jagged line represents the moist static energy in the boundary layer with diurnal solar forcing. Convection will take place when the planetary boundary layer moist static energy exceeds the convective threshold denoted by the thick broken line. Note that for diurnal forcing, the threshold is exceeded each day, even though the mean moist static energy. For mean daily solar forcing, the threshold is exceeded only on two of four days, in spite of the mean moist static energy being about the same.