It is important to understand the role of oceans on different temporal scales of (intraseasonal, interannual and decadal) monsoon variability. This requires clear understanding of the following points
(i) role of the Indian Ocean warming on the monsoon climate variability, (ii) MOC variability and its impact on monsoon climate, (iii) the Indo-Pacific exchanges both through atmospheric and oceanic pathways in different time scales.
There is a lack of consensus on the estimation of meridional heat transport in the ocean models. This is leading to uncertainty in the variability estimates of heat transport. This needs to be understood especially in the context of its relationship with monsoon variability via air sea interaction. Secondly it is important and essential for benchmarking of existing coupled general circulation models (CGCMs) with respect to ISM variability. Do models represent mean monsoon and predict interannual variability of the monsoon rainfall well enough? Identifying the common problems in models might help in resolving the unresolved processes. So we propose to carry out model (atmosphere, ocean and coupled) experiments to understand and isolate the role of different climate modes on monsoon variability. Further we propose to examine the impact of data assimilation (both ocean and atmosphere) on monsoon variability in models.
- Identification and benchmarking of models on capturing the natural modes of climate variability and their relationship with Indian southwest monsoon variability.
- Emergence of new climate drivers and predictors.
- Indian summer monsoon rainfall variability in response to differences in the decay phase of El Niño
In general the Indian summer monsoon (ISM) rainfall is near normal or excess during the El Niño decay phase. Nevertheless the impact of large variations in decaying El Niño on the ISM rainfall and circulation is not systematically examined. Based on the timing of El Niño decay with respect to boreal summer season, El Niño decay phases are classified into three types in this study using 142 years of Sea Surface Temperature (SST) data, which are as follows: (1) early-decay (ED; decay during spring), (2) mid-summer decay (MD; decay by mid-summer) and (3) no-decay (ND; no decay in summer). It is observed that ISM rainfall is above normal/excess during ED years, normal during MD years and below normal/deficit in ND years (Fig. 1), suggesting that the differences in El Niño decay phase display profound impact on the ISM rainfall. Tropical Indian Ocean (TIO) SST warming, induced by El Niño, decays rapidly before the second half of the monsoon season (August and September) in ED years, but persists up to the end of the season in MD years, whereas TIO warming maintained up to winter in ND case. Analysis reveals the existence of strong sub-seasonal ISM rainfall variations in the summer following El Niño years. During ED years, strong negative SST anomalies develop over the equatorial central-eastern Pacific by June and are apparent throughout the summer season accompanied by anomalous moisture divergence and high sea level pressure (SLP). The associated moisture convergence and low SLP over ISM region favour excess rainfall (mainly from July onwards). This circulation and rainfall anomalies are highly influenced by warm TIO SST and Pacific La Niña conditions in ED years. Convergence of southwesterlies from Arabian Sea and northeasterlies from Bay of Bengal leads to positive rainfall over most part of the Indian subcontinent from August onwards in MD years. ND years are characterized by negative rainfall anomaly spatial pattern and weaker circulation over India throughout the summer season, which are mainly due to persisting El Niño related warm SST anomalies over the Pacific. (Chowdary, J.S., Harsha, H.S., Gnanaseelan, C., Srinivas, G., Parekh, A., Pillai, P. and Naidu, C.V., (2016) Climate Dynamics. doi:10.1007/s00382-016-3233-1)
Figure 1. Composite of monthly and seasonal rainfall anomalies (IMD, mm/day) averaged over the Indian Subcontinent for El Niño Early Decay (ED), Mid-Summer Decay (MD) and No Decay (ND) years.
- Combined influence of remote and local SST forcing on Indian Summer Monsoon Rainfall variability
The combined influence of tropical Indian Ocean (TIO) and Pacific Ocean (TPO) sea surface temperature (SST) anomalies on Indian summer monsoon rainfall (ISMR) variability is studied in the context of mid-1970s regime shift. The rainfall pattern on the various stages of monsoon during the developing and decaying summer of El Niño is emphasized. Analysis reveals that ISMR anomalies during El Niño developing summer in epoch-1 (1950–1979) are mainly driven by El Niño forcing throughout the season, whereas TIO SST exhibits only a passive influence (Fig. 2). On the other hand in epoch-2 (1980–2009) ISMR does not show any significant relation with Pacific during the onset phase of monsoon whereas withdrawal phase is strongly influenced by El Niño. Again the eastern Indian Ocean cooling and westward shift in northwest Pacific (NWP) cyclonic circulation during epoch-2 have strong positive influence on the rainfall over the central and eastern India during the matured phase of monsoon. ISMR in the El Niño decaying summer does not show any significant anomalies in epoch-1 as both Pacific and Indian Ocean warming dissipate by the summer. On the other hand in epoch-2 ISMR anomalies are significant and display strong variability throughout the season. In the onset phase of monsoon, central and east India experience strong negative precipitation anomalies due to westward extension of persistent NWP anticyclone (forced by persisting Indian Ocean warming). The persistent TIO warming induces positive precipitation anomalies in the withdrawal phase of monsoon by changing the atmospheric circulation and modulating the water vapour flux. Moisture budget analysis unravels the dominant processes responsible for the differences between the two epochs. The moisture convergence and moisture advection are very weak (strong) over Indian land mass during epoch-1 (epoch-2) in El Niño decaying summer. The changing moisture availability and convergence play important role in explaining the weakening of ENSO monsoon relation in the recent years. The local TIO SST forcing and NWP circulation are prominent forcing factors for the interannual variability of ISMR during epoch-2. (Chakravorty, S., Gnanaseelan, C. and Pillai, P.A. (2016) Clim Dyn 47: 2817. doi:10.1007/s00382-016-2999-5)
Figure 2: Lead lag correlation of NDJ(0/1) Niño 3.4 index with JJAS(0) rainfall anomalies (a, d), JJAS(1) rainfall anomalies (b, e) and simultaneous correlation of NIO SST anomalies with JJAS rainfall anomalies (c, f) for epoch-1 and epoch-2. The boxes in (f) are named (A) Southern box (74°E–78.5°E, 14°N–21°N), (B) Northern box (79°E–88.5°E, 20°N–25°N). The shaded region are significant above 95 % level
- Arabian Sea SST evolution during spring to summer transition period and the associated processes in coupled climate models.
Many climate models have problems in simulating the sea surface temperature (SST) in the tropical Indian Ocean (TIO). The Coupled Model Inter-comparison Project Phase 5 (CMIP5) models, in general, underestimate SST over the entire TIO region. This study examines the SST evolution during spring to summer transition months (May and June) over the Arabian Sea (AS) region in the historical simulations of 13 CMIP5 models and the Climate Forecasting System coupled models CFSv1 and CFSv2. The annual cycle of SST shows that the summer monsoon cooling is not adequately captured by many models. Based on the state of June SST tendency, models have been divided in to three groups, the first group (G1) consists of models having stronger than observed cooling, second group (G2) considers models having closer to observed cooling and the third group (G3) includes models having lesser than observed cooling. Mixed layer heat budget analysis revealed that atmospheric flux is mainly responsible for unrealistic SST warming in most of the G3 models during June. The vertical mixing and horizontal advection contribute considerably to the SST cooling in summer (June) especially for G1 and G2 models. On the other hand, spring warming in all the models is consistently forced by the surface heat flux. It is also found that the monsoon low-level jet (LLJ) is not accurately represented in most of the models. The misrepresentation of LLJ causes bias in the oceanic processes leading to unrealistic SST evolution in many models. One way of LLJ affecting the oceanic processes is by modulating mixed layer depth (MLD). It is observed in general that the models with deeper MLD display strong SST cooling. The model deficiency in representing AS SST is speculated to be a major limiting factor in capturing the monsoon rainfall in the current coupled models. The proper simulation of AS SST is therefore very crucial for the accurate representation of Indian summer monsoon precipitation. (Ojha S., Gnanaseelan C., Chowdary J.S., Parekh A., Rahul S. (2016) International Journal of Climatology, 36, DOI:10.1002/joc.4511, 2541-2554)
Data Assimilation Research:
Impact of temperature profile assimilation on simulation of Monsson circulation
Four dimensional Data Assimilation (FDDA) for simulation
- The FDDA is a continuous dynamic data assimilation method that relaxes the model state toward observed state.
- In the analysis, Newtonian relaxation term is added to the prognostic equation
- T is a prognostic variable (i.e., temperature), M represents model which includes the physical processes, x represents the independent variables, and t is time. y◦is the observation vector, H is the observation operator that transforms or interpolates the model forecast variable to the observation variable and location, G is the nudging magnitude matrix, and Ws, Wtare the spatial and temporal nudging (or weighting) coefficients.
- The nudging strength is specified to be 3 × 10−4 s−1 and ε denotes observation quality factor.
- Each observation is ingested into the model at its observed time and location with proper space-time weights and the model spreads the information in time and space according to the model dynamics.
- Tempeature Profile assimilation eliminate the asymmetric SLP bias resulted better monsoon circulation, reduced temperature bias in lower & mid-troposphere and simulated monsoon Indices better.
Fig. 1. Time series of monsoon indices (a) WYI (ms−1) (b) MHI (ms−1) (c) EIMRI (mmday−1) (d) WNPSMI (ms−1) and (e) EASMI (ms−1). WRFAIRS is based on assimilation experiments.
Raju, A., Parekh, A., Sreenivas, P., Chowdary, J.S. and Gnanaseelan, C., ‘Estimation of improvement in Indian summer monsoon circulation by assimilation of satellite retrieved temperature profiles in WRF model’, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2015, 10.1109/JSTARS.2015.2410338, 1591-1600.
Raju A., Parekh A., Chowdary J.S., Gnanaseelan C., ‘Assessment of the Indian summer monsoon in the WRF regional climate model’, Climate Dynamics, 44, 2015, DOI:10.1007/s00382-014-2295-1, 3077-3100
Impact of temperature and moisture profile assimilation on Seasonal prediction of Indian summer monsoon
- The WRF 3DVar data assimilation system [Skamarock et al., 2008] is used in this study. Two assimilation experiments are performed, one with conventional data and the other with both conventional and AIRS profiles. These experiments are in addition to the control run where no assimilation is performed.
- Six-hourly assimilation cycles is performed with ±3 h time window for entire month of May. Model is initialized at 00:00 UTC, 1 June 2010, and gives the forecast for 1 June to 30 September 2010.
- Conventional observations are assimilated, such as surface synoptic observations; Meteorological Aerodrome Report (METAR); buoy, ships, aircraft and SSMI wind speed and total perceptible water and satellite-observed cloud motion vectors from GTS during May 2010. It includes surface station reports as well as upper air observations.
- In the third experiment, AIRS-retrieved T and Q profiles (more than 43% of observations) are assimilated along with the conventional data. level 2, version 6 AIRS atmospheric profiles are used. The T (Q) profiles are available at 28 (14) standard pressure levels between 1000 and 0.1 hPa (1000hPa to 50hPa).
Fig 2. Temporal variation of RMSE of the predicted (a–d) WVMR (g kg-1) and (e–h) precipitation (mm d-1) against ERAI/GPCP during monsoon 2010.
- Assimilation of AIRS profiles has significant impact on predicting the seasonal mean monsoon characteristics such as tropospheric temperature, low-level moisture distribution, easterly wind shear, and precipitation.
- The vertical structure of the RMSE is substantially affected by the assimilation of AIRS profiles, with smaller errors in temperature, humidity, and wind.
- The consequent improved representation of moisture convergence in the boundary layer (deep convection as well) causes an increase in precipitation forecast skill.
- This finding has large implications to the operational seasonal forecasting capabilities over the Indian subcontinent.
Raju A., Parekh A., Kumar Prashant, Gnanaseelan C., ‘Evaluation of the impact of AIRS profiles on prediction of Indian summer monsoon using WRF variational data assimilation system’, Journal of Geophysical Research, 2015, DOI:10.1002/2014JD023024, 1-20
Impact of temperature/moisture profile assimilation on predictability of MISO
- Two separate simulations are carried out for 2003 to 2011. First simulation is forced by NCEP (CTRL), is forced apart from NCEP forcing, AIRS T & Q profiles are assimilated (ASSIM). Ten active and break cases are identified from thses simulations.
- Three dimensional Temperature states are perturbed using twin perturbation method for active and break cases and carried out predictability tests.
- Models is integrated for the 30 days from the peak of active/break in forecast mode for each of these perturb initial conditions of CTRL and ASSIM respectively. Hence, there are total 160 model runs of 30 days period are carried out.
- Signal is defined as the variance within a sliding window of width (2 L + 1) in the experiments CTRL and ASSIM, where L is taken as 31 days to encompass a complete ISO event. Equation form of signal is
σ represents variance, S stands for the signal, τ is time window (i.e., -L to +L days); X represents the parameter from simulation, i stands number of active/break cases.
Noise is estimated as the variance among perturb cases and is determined by averaging over all ensembles and events (Equation 2).
Where N stands for the noise, τ is ranging 1 to L days of ISO events and j is the number of perturbation cases, Xp is geophysical parameter from predictability experiments.
Fig 3. Time evolution of signal & noise of rainfall (mmd−1) over MCR for active (a-b) and break (c-d) phases from CTRL (upper panel) and ASSIM (bottom panel). Vertical red lines are indicating predictability limit where signal and noise intersects each other.
- Analysis reveals that the limit of predictability of low level u wind is improved by four(three) days during active(break) phase. Similarly, the predictability of upper level u wind(precipitation) is enhanced by four(two) and two(four) days respectively during active and break phases.
- More realistic baroclinic response and better representation of vertical state of atmosphere associated with monsoon enhance the predictability of circulation and rainfall.
Parekh A., Raju A., Chowdary J.S., Gnanaseelan C., ‘Impact of satellite data assimilation on the predictability of monsoon intraseasonal oscillations in a regional model’, Remote Sensing Letters, 8, 2017, DOI:10.1080/2150704X.2017.1312614, 686-695.
Project: Climate Variability and Data Assimilation Research
Project Director: Dr. C. Gnanaseelan, Scientist-F
Dr. C. Gnanaseelan
Phone No - +91-(0)20-25904271
Dr. Anant Parekh
Air-sea interaction, IO variability
Phone No - +91-(0)20-25904264
Dr. J.S. Chowdary
Air-sea interactions, Monsoon Variability and Predictability
Phone No - +91-(0)20-25904273
Shri. Prem Singh
Ocean Modelling and Simulation Studies
Phone No - +91-(0)20-25904279
Smt. J. S. Deepa
Phone No - +91-(0)20-25904230
Ku. Rashmi Arun Kakatkar
Phone No - +91-(0)20-25904239
Dr. Sreenivas Pentakota
Phone No - +91-(0)20-25904303