This study investigates two aspects of the variability of the Indian climate, using in situ or satellite data and the four-active-thermodynamic-layer model of the Indian Ocean developed by Han et al. (1999).
One aspect is the role of the high-frequency fluctuations of winds and rainfall in the thermodynamics of this ocean. We use daily variations of winds and rainfalls observed by satellites (QuikSCAT and TRMM) and the monthly variations of wind estimated by FSU (Legler et al, 1981) and of rainfall estimated by Xie and Arkin (1996) over 1980-2001. The satellite winds can be stronger than 0.20Pa, while the monthly averaged winds are always weaker than 0.06 Pa (Fig. 1a). Most of the wind energy over the Indian Ocean is found in the 3day-to-60day range, corresponding to Westerly Wind Bursts (WWB) or Madden and Jullian Oscillations (MJO). Two model experiments are performed to examine the impact of the WWBs and MJOs on the ocean. Experiment FSU is forced by the monthly winds since 1980. Experiment QuikSCAT is forced by the same monthly winds to which is added the daily fluctuations of QuikSCAT in year 2000. Results are validated with the sea level provided by the satellite TOPEX/Poseidon every 3 day. It is striking that the daily fluctuations of the wind drastically change and improve the simulations of the sea level (Fig. 1b). In addition, it is very striking that the ocean energy is shifted towards lower-frequencies compared to the wind input: the sea level energy is in the 30-to-90 day range. Even the 90 day fluctuations are considerably improved by the 3 day-to-60 day fluctuations of the wind. These results described thoroughly in (Zhang and Perigaud, 2002) demonstrate the importance of the WWBs and MJOs in the variability of the Indian Ocean circulation.
Similarly, the satellite rainfall can be more than 25mm/day, while the monthly rainfall averages are always less than 10 mm/day (Fig. 2a). Most of the rainfall energy is found correspond to very sudden events in the 2day-to-20day range. In addition to the wind forcing of the two model experiments described above, Experiment Arkin and Experiment TRMM were forced by the rainfall estimated with the monthly averaged values, or with the daily satellite values respectively . Adding rainfall has a negligible impact on sea level. But, for reasons explained in Perigaud et al. (2002a), it has a drastic impact on the salinity and temperatures in subsurface and at the surface where it can be larger than 1ºC (Fig. 2b). This impact is large, even though the air-sea heat fluxes are imposed to climatological values in both experiments. These results are very meaningful in terms of the role of the Indian Ocean surface and subsurface circulation in the coupling with the atmosphere: coupled with a tropical atmosphere model, they demonstrate that the Indian ocean SST anomalies play a role in driving the winds and rainfall over the Indian ocean (Perigaud et al., 2002b).
The monthly observations of subsurface temperatures from XBT analyzed by Smith (1995) are used to estimate the surface dynamic topography relative to 400m (hdyn) over the Indian and Pacific between January 1980 and December 2000. Between October 1992 and December 2000, the hdyn estimates can be compared with the TOPEX/Poseidon satellite measurements. Indeed, the difference is largely explained by the salinity changes. After correcting the XBT-measured topography for the lack of salinity measurement, and after correcting the TOPEX-measured topography for the lack of measurements prior to 1992, a continuous time series of monthly variations of sea level is estimated over 1980-2000. Then, the basin wide averages are taken over the Indian Ocean North of 30S up to the equator, or up to the Indian Continent. Multiplied by the surface of the area where the average is taken, it is mostly the South of the Indian Ocean which has significant volume changes, these changes being only twice smaller than the changes in the North or South subtropical Pacific (Fig. 3), even though the Indian area is smaller than one third of the Pacific area.
The monthly averages of the FSU winds since 1980 have been used to compute the sea level in Sverdrup balance. It is very striking that the South subtropical shows a decadal trend between 1982 and 1997 which is in quasi-Sverdrup balance with its winds, whereas the Indian Ocean is not (Fig. 4). Indeed, to explain the difference between the observed Indian sea level and the Sverdrup balance with its observed winds, one needs to take into account the corresponding difference in the North subtropical sea level, the transport variations of the Indonesian Throughflow and those of the inflow/outflow across 30S. These results are consistent with the fact that the Indian Ocean plays a key role in balancing the pressure difference between the Pacific, the Southern Ocean and Atlantic Ocean, while involving its own active coupling locally with its atmosphere.
It was found that it is the combination of WWBs and decadal ocean/atmosphere trends that matter for ENSO forecasts (Perigaud and Cassou, 2000). Relatively to the energy in the seasonal-to-interannual range, the daily and decadal changes of the Indian Ocean have even more energy than in the Pacific which is dominated by El Niño events. These results illustrate that it is the whole spectrum of variability from WWBs to decadal variations of the Ocean/Atmosphere which plays a role in the variability of the Indian Climate.