AMIP II Diagnostic Subproject 26
Monsoon precipitation
Background
Under the AMIP subproject 26 we analyzed the monthly precipitation in AMIP runs of 31 GCMs (Gadgil and Sajani 1998a,b). We find that while most of the models simulate the seasonal migration of the primary rainbelt over the African region and the rainfall patterns associated with the African monsoon in austral and boreal summers, the seasonal variation over the Asia-West Pacific region and the seasonal mean pattern over the Indian monsoon zone is realistically simulated only by some models.
In the AMIP decade there were large fluctuations of the Indian summer monsoon with the droughts of 1982 and 1987 being associated with a warm event (El Nino) in the Pacific and the excess monsoon season of 1988 with a cold event. We expect best results for the simulation of these three events in the AMIP runs because they were associated with large SST anomalies which were used as boundary conditions in the AMIP runs. Yet only five models were able to simulate anomalies of significant amplitude and the correct sign, while another six simulated the correct signs but not the amplitudes of the anomalies of these three events.
We find that, on the whole, models which simulate the seasonal migration of the primary rainbelt over the Asia-West Pacific region realistically have a greater propensity for the simulation of the mean rainfall pattern over the Indian monsoon zone as well as the simulation of the interannual variation associated with ENSO. If analysis of the AMIP II runs support this hypothesis, then in the strategy for improvement of the simulation of the Indian monsoon rainfall and its interannual variation the focus should be on better simulation of this large-scale feature.
The additional years included in AMIP II are of particular interest in understanding the interannual variation of the Indian monsoon and its teleconnections with events over the Pacific. Whereas each El Nino event in the first AMIP decade was associated with a deficient monsoon (although the converse was not true!), despite the occurrence of El Nino warmings of 1992, 93 and 94 no deficient monsoon seasons occurred after 1987. It is possible that the dominating influence of the Pacific SSTs found for the earlier period (Palmer et al. 1992) may have become less significant in the last few years. Whether SSTs over other oceans or atmospheric factors have become more important need to be investigated. AMIP II with improved versions of the models, ensemble runs etc. provides a unique opportunity to understand this and hence assess the potential for prediction of the interannual variation of the Indian monsoon.
The interannual variation of the Indian monsoon between good and poor monsoons is linked to the intraseasonal variation between active and weak spells or breaks over the monsoon zone (Krishnamurti and Bhalme 1976, Gadgil 1988). The intraseasonal variation over monsoonal regions comprises such fluctuations between active/weak spells and, over the Indian longitudes poleward propagations of the rainbelt (Sikka and Gadgil 1980, Gadgil and Srinivasan 1990, Gadgil and Asha 1992 , Srinivasan and Smith 1996). Mechanisms proposed so far involve variation in the stability of the atmosphere and surface hydrological feedbacks (Krishnamurti and Bhalme 1976, Webster 1983, Nanjundiah et al. 1992, Srinivasan et al. 1993). Analysis of the simulated variation of the relevent variables should provide insight into the underlying mechanisms.
Objectives
- Analysis of the simulation of monthly fields of precipitation and circulation and observations to assess the skill of simulation of the mean seasonal rainfall pattern and the interannual variation during AMIP II period in monsoonal regions
- to test the hypothesis suggested by our analysis of AMIP I runs (Gadgil et al. 1997) viz that the propensity of a realistic simulation of the mean seasonal rainfall pattern over the Indian monsoon zone and its interannual variation is greater for the class of models which simulate the seasonal variation over the Asia-West Pacific sector (70-140E) realistically
- further diagnostic studies to understand the observed change of the teleconnection between the interannual variation of the Indian monsoon and ENSO during the AMIP I decade (1979-88) and the period 1989-95 in AMIP II.
The dataset comprising monthly precipitation at all meteorological observatories in India during 1901-90, obtained from the India Meteorological Department analysed earlier (Gadgil et al. 1993) will be extended to cover the AMIP II period. For analysis of satellite derived convection an objective method developed for delineation of TCZ from the daily OLR-albedo data (Gadgil and Guruprasad 1990) will be used. These indices of convection along with HRC (for 1979-87), OLR from NOAA (for 1979-86), ERBE (available for the period 1985-89) and MSU (for 1979-93) will be compared to model simulated precipitation as in the AMIP I subproject 26.
Since, besides the subproject proposed here, there are two monsoon subprojects (No. 6, 25), we will ensure that duplication of effort is avoided by collaboration and exchange of results.
- Monthly totals of precipitation and circulation data at 1000, 850, 700 and 200 hPa and vertical profiles of air temperature and specific humidity, and vertical motion in all the AMIP II runs (from Table 1 of AMIP II guidelines)
- High frequency outputs of precipitation, OLR and circulation at 850 and 200 hPa (Table 3) for all the models.
- Single-level low frequency (monthly-mean) outputs and fixed geographical fields (Tables 2, 5)
For further information, contact the AMIP Project Office (amip@pcmdi.llnl.gov).
Last update: 2 March 1998. This page is maintained by mccravy@pcmdi.llnl.gov
UCRL-MI-127350