Dynamics of the quasi-biweekly variability of the Asian monsoon anticyclone
概要
The Asian monsoon anticyclone (AMA; also known as the South Asian high or the Tibetan high) is a planetary-scale anticyclone in the upper troposphere and lower stratosphere (UTLS) over Asian region which appears in the boreal summer. The AMA is mainly driven by persistent convective heating over South and Southeast Asia associated with the Asian summer monsoon. While the existence of the AMA is persistent throughout the boreal summer, its intensity, location, and structure exhibit significant variability with a sub-seasonal timescale. The AMA variability is one of the important factors affecting the circulation and convection in the troposphere.
The AMA having such variability is also important in the tracer transport and mixing processes between the troposphere and stratosphere. The air originating from the boundary layer is uplifted by monsoonal deep convection to the upper troposphere and trapped in the UTLS by the strong anticyclonic flow associated with the AMA. Previous observational and modeling studies on the role of the AMA in tracer transport showed the importance of the AMA variability. Particularly, the quasi-isentropic turbulent mixing between the tropospheric air inside the AMA and the lower stratospheric air outside the AMA is enhanced by mechanical stirring with the subseasonal variability.
The AMA variability has been described in various ways. The well-known spatial characteristic is the frequent east-west movement of the center location of the anticyclone. The longitude of the center location seen as the maximum of geopotential at a specific pressure level exhibits a clear bimodal distribution. From a viewpoint of the potential vorticity (PV) on an isentropic surface, the AMA variability can be seen as frequent movements, deformations and even splittings of an anticyclonic vortex. Although various processes with different timescales are involved in the AMA variability, a well-known dominant timescale is between 10 and 20 days and called quasi-biweekly
Despite its importance, the mechanism of subseasonal variability of the AMA is not well understood. The AMA variability can be forced by external processes, such as convective heating variability in South and Southeast Asia or Rossby wave propagation along the subtropical jet. On the other hand, dynamical instability may also play a role. A previous study using an idealized shallow water model with a steady forcing demonstrated the possibility of spontaneous generation of variability by dynamical instability. The characteristic spatial feature of such variability is quasi-periodic westward eddy shedding from the main anticyclone. The occurrence of rapid westward movement of anticyclonic vortices during the AMA variability is clearly observed in PV and attributed to eddy shedding. However, there is an important difference between the spatial structure of PV variability in the shallow water model and that in the real atmosphere. The anticyclonic vortices after shedding are rarely departed away from the main anticyclone, but are trapped within a finite longitudinal range. In addition, the relative importance of eddy shedding in the whole AMA variability is uncertain, because the spatial characteristics of PV associated with the AMA variability, including eddy shedding, have not been systematically investigated. Thus, the importance of the dynamical instability for the AMA variability is still not clear.
The purpose of this study is to improve the understanding of the spatial and temporal characteristics of the AMA and their underlying dynamics. For this, the possible role of dynamical instability is investigated through data analysis and shallow water model experiments. The statistical analysis on the spatial structure of the AMA variability is performed, focusing on the temporal evolution using reanalysis data. Low PV area is used to describe the anticyclone. The dominant component of the AMA variability with quasi-biweekly timescale is extracted and its typical life cycle in PV distribution and other variables are described. Thereafter, the shallow water model is used to investigate the behavior of the anticyclone variability caused by dynamical instability. The effect of background latitudinal structure on the change of spatial characteristic of reproduced eddy shedding state is focused on.
First, the spatial structure of the AMA variability is examined using ERA-Interim reanalysis and Outgoing Longwave Radiation (OLR) data from 1979 to 2016. The zonal distribution of thicknessweighted low PV area and its longitudinal flux is calculated using ERA-Interim reanalysis data. The dominance of variability with substantial east-west movement of the anticyclone is implied by the large temporal fluctuation of longitudinal flux on 370 K level. The budget analysis of thickness-weighted low PV area over the western part of the AMA shows that the variability there is predominantly controlled by the longitudinal advection of low PV air. Thus the variability can be characterized by event-like rapid westward movements of the air inside the anticyclone by nearly adiabatic and inviscid process, consistent with the existing description of ’eddy shedding’. However, the budget analysis also suggests that there are significant temporal eastward movements of low PV air as well as westward movements. Thus, it is found that the subseasonal variations of the air inside the AMA in the real atmosphere is oscillatory rather than dissipative.
To extract the time evolution of dominant pattern of the quasi-biweekly AMA variability, the Empirical Orthogonal Function (EOF) decomposition is conducted for 5-20 day filtered and normalized geopotential height anomaly field for 0◦E–150◦E and 10◦N–50◦N on the 100 hPa level. It is found that the combination of the first two significant EOF modes shows nearly periodic and spatially propagating pattern. By defining 8 phases based on a two-dimensional phase space by the two EOFs, spatial structures of geopotential, PV, and OLR anomalies at each phase are examined. The spatial distribution of the occurrence of low PV on the 370 K level for each phase indicates the oscillatory longitudinal movement of the air inside the AMA. The oscillatory cycle is also indicated by a significant relation between phases and longitudinal fluxes of low PV air. A composite of geopotential anomaly shows westward-propagating large-scale pattern over both middle and low latitudes. The vertical structure is nearly barotropic and trapped around the tropopause. However, there are deeper downward influences to the middle troposphere at midlatitudes than at low latitudes. The composite of OLR anomalies over Tibetan Plateau shows statistically significant signals which are in phase with the geopotential anomalies with the same sign, indicating the effect on local convection variability.
The extracted AMA variability pattern in this study corresponds well with the known pattern of the AMA center location variability. The phase corresponding to the timing of westward eddy shedding shows a significant preference of western locations of the AMA center. The opposite phase corresponds to a preference of eastern locations. Thus, the life cycle of the pattern described by 8 phases based on the EOF decomposition can explain considerable part of the transition between two dominant locations of the anticyclone center in the western and eastern part.
Next, the possibility of the subseasonal variability of the AMA emerging from dynamical instability is examined through numerical experiments using the shallow water model. The effect of the subtropical jet and associated latitudinal thermal structure around the tropopause is included in the shallow water model in the form of latitudinally varying background parameters. Specifically, an equivalent depth and a background zonal jet are estimated as functions of latitude, using threedimensional reanalysis data on isentropic coordinates. It is found that the estimated equivalent depth to the north of approximately 30◦N had a large positive latitudinal gradient.
Numerical experiments are performed using the modified shallow water model including latitudinal variation of background parameters, to examine the impact on the spatial structure of resultant states. In experiments using the shallow water model that featured a latitudinally varying mean depth, new types of solutions with different spatial structures from those of the conventional shallow water model are found. When the large latitudinal gradient of the mean depth is imposed sufficiently close to the forcing region, a quasi-periodic eddy shedding state but with a longitudinally bounded structure is reproduced. The behavior of such unsteady state, in which the low PV area is shed westward and thereafter migrates clockwise inside the anticyclone toward the east, bears similarity to the observation. It is also found that the parameter values which enables such state is within the realistic range.
The change in spatial structure of the anticyclone by including latitudinally varying mean depth is explained through the change in background PV structure. A positive gradient of mean depth corresponds to a locally weakened, and possibly even negative effective beta. Such background PV structure leads to the anticyclone centered around the latitude of maximum background PV, which is close to an axisymmetric f-plane solution. The PV structure is able to have smaller scale than the macroscopic structure of the whole anticyclone. Thus it become possible for anticyclonic eddies shed from the main anticyclone to be trapped within the finite longitudinal range.
Another set of experiments using the shallow water model with a background jet is performed. Similar to the case of imposing a varying mean depth, new types of solutions with a bounded structure, including eddy shedding, are found in the realistic range of parameters. The explanation using background PV can be also applied for this case, since imposing a westerly jet produces a localized negative background PV gradient around the latitude of the jet axis.
The spatial structure of the bounded anticyclone reproduced by the modified shallow water model has important resemblance to the reality. First, the longitudinal width of the reproduced anticyclone is comparable to the AMA, which is about 10000 km. Second, in the case of experiments with a bounded structure, the western part of the anticyclone is slightly shifted northward. Third, the anticyclonic vortices with low PV is advected clockwise within the western part of the anticyclone.
The periods of eddy shedding reproduced by the shallow water model in this study are 5 and 10 days for most cases. Considering the uncertainty in parameter values, specifically for the thermal relaxation timescale τ and a constant equivalent depth in tropics H0, these values may explain the dominant quasi-biweekly timescale of the AMA in the real atmosphere. Additionally, the timescale of the entire anticyclone variability may be determined by multiple cycles of eddy shedding. That possibility is supported by the comparison between the spatial structure of low PV area of the shallow water model and of the observed quasi-biweekly AMA variability pattern.
There are several important factors which is ignored in the shallow water model in this study. For example, convective variability, Rossby waves propagating along the subtropical jet, the effect of Tibetan Plateau and Himalayan mountains, and baroclinic disturbances in the midlatitudes may affect the variability of the AMA. Their relative roles and effect on the view of dynamical instability in this study should be investigated in detail in future studies. It is also interesting to examine the potential use of the phase description of the AMA variability pattern in this study for the diagnose of tracer transport and mixing around the AMA. The relative importance of the AMA variability pattern among other patterns for the short to middle range weather prediction in the troposphere is also interesting topic for future study.