The Ocean's Role, Coupled Models and Monsoon Links
CSIR NET Part C meteorology: ocean heat capacity and transport, ocean GCMs, coupled model design and biases, ENSO–IOD–monsoon links and monsoon prediction.
- 20 min
On this page · 13 sections
Prerequisites
- Dynamical cores, parameterisation and the model hierarchy, from Grid-Point and Spectral GCMs.
- ENSO mechanisms and the MJO and BSISO, from Climate Variability: MJO, ENSO, QBO and Sunspot Cycles.
- Ocean circulation, Ekman transport, the thermohaline circulation and the IOD, from Ocean Circulation for Part C: Ekman Spiral, Upwelling, ENSO, IOD and Thermohaline Flow.
- Monsoon systems and Indian rainfall categories, from Monsoon, Cyclones, Western Disturbances and ENSO.
The Hook: Why an Atmosphere Model Alone Cannot Forecast the Monsoon
What the Ocean Does for Climate
Heat capacity and memory
The column heat capacity of the whole atmosphere is
$$ C_{atm} = \frac{c_p\,p_s}{g} = \frac{1004\times 1.013\times 10^{5}}{9.81} \approx 1.0\times 10^{7}\ \text{J m}^{-2}\,\text{K}^{-1} $$The same heat capacity is held by a layer of sea water only about 2.5 m deep (\( \rho c_w \approx 1025\times 3990 \) J m\(^{-3}\) K\(^{-1}\)). A 50 m mixed layer stores about 20 times as much heat per kelvin as the atmosphere above it.
| Consequence | Explanation |
|---|---|
| Seasonal lag and moderation | maritime climates have smaller annual ranges and later temperature maxima than continental ones |
| Slow response to forcing | the mixed layer responds over years, the deep ocean over centuries to millennia; the transient climate response is smaller than the equilibrium climate sensitivity |
| Climate commitment | warming continues after forcing stops increasing, because the ocean has not caught up |
| Predictability | ocean anomalies persist for months, which is the main source of seasonal forecast skill (ENSO above all) |
| Uptake of excess energy | the ocean has absorbed roughly 90% of the extra energy accumulated by the climate system in recent decades |
A simple mixed-layer (slab) model makes the time scale explicit:
$$ C\frac{dT'}{dt} = F - \lambda T' \quad\Rightarrow\quad \tau = \frac{C}{\lambda} $$where \( F \) is the forcing and \( \lambda \) the climate feedback parameter (W m\(^{-2}\) K\(^{-1}\)). For a 50 m mixed layer and \( \lambda = 1.5 \) W m\(^{-2}\) K\(^{-1}\), \( \tau \approx 4 \) years.
Heat transport
The ocean carries a large part of the poleward energy transport in the tropics and subtropics, mainly through:
- Wind-driven gyres and western boundary currents (Gulf Stream, Kuroshio, Agulhas), and
- the meridional overturning circulation, especially the Atlantic Meridional Overturning Circulation (AMOC), in which warm upper water flows north and cold North Atlantic Deep Water returns south. The AMOC transports heat northward at all latitudes in the Atlantic, even in the South Atlantic.
A simple estimate of overturning heat transport is
$$ H \approx \rho\,c_w\,\Psi\,\Delta T $$with \( \Psi \) the overturning volume transport (1 Sv \( = 10^{6} \) m\(^3\) s\(^{-1}\)) and \( \Delta T \) the temperature difference between the northward and southward flows.
In the Indian Ocean the monsoon winds reverse the upper-ocean circulation seasonally (the Somali Current flows north in summer, with strong coastal upwelling off Somalia and Oman), and the cross-equatorial cell carries heat southward. The seasonal heat transport changes sign with the monsoon.
Carbon and other roles
The ocean also absorbs a substantial fraction of the CO\(_2\) emitted by human activities, roughly a quarter, slowing atmospheric accumulation but acidifying the surface ocean. It is the main source of atmospheric water vapour, and therefore of the latent heat that drives tropical convection.
Ocean General Circulation Models
| Element | Choice in ocean GCMs |
|---|---|
| Equations | primitive equations with Boussinesq (density variations only in buoyancy) and hydrostatic approximations; equation of state of sea water for \( T \), \( S \), \( p \) |
| Pioneers | Bryan’s ocean model at GFDL (late 1960s), the basis of the Bryan–Cox family |
| Vertical coordinate | \( z \)-levels (with partial cells), terrain-following (sigma, for coastal models), isopycnal (density layers), or hybrid |
| Resolution issue | the oceanic Rossby radius is only tens of kilometres, so mesoscale eddies need grid spacings of about 1/10° to be resolved |
| Eddy parameterisation | at coarser resolution, eddy effects are represented by along-isopycnal mixing and the Gent–McWilliams scheme, which flattens isopycnals as baroclinic eddies do |
| Convection | static instability removed by convective adjustment or enhanced vertical mixing |
| Spin-up | the deep ocean takes thousands of years to reach equilibrium, so long spin-up runs precede climate experiments |
Coupled Models
Architecture
graph TD
A[Atmosphere GCM] -->|wind stress, heat flux, freshwater flux| C[Coupler]
C -->|SST, sea-ice cover, surface currents| A
C --> O[Ocean GCM]
O --> C
C --> I[Sea-ice model]
I --> C
C --> L[Land surface and rivers]
L --> C
O --> R[River runoff and ice-sheet melt enter as freshwater]The coupler passes fluxes and states between components that usually run on different grids and time steps, interpolating conservatively so that energy and water are not created or lost at the interface.
The modelling hierarchy for ocean–atmosphere work
| Configuration | Ocean | Typical use |
|---|---|---|
| AMIP-type AGCM | observed SST prescribed | atmospheric response to known SST; teleconnections |
| Slab ocean | mixed layer of fixed depth with prescribed ocean heat transport (“Q-flux”) | equilibrium climate sensitivity at low cost |
| Fully coupled AOGCM | dynamic ocean, sea ice | ENSO, IOD, MJO–ocean interaction, transient climate change, seasonal prediction |
| Earth system model | plus carbon cycle, biogeochemistry, chemistry | emission-driven projections |
Flux adjustment and drift
Early coupled models drifted to unrealistic climates because small flux errors in each component accumulated once they were coupled. Flux adjustment (flux correction) added fixed, non-physical corrections to the heat and freshwater fluxes passed to the ocean to keep the coupled climate near observations. It hides errors rather than fixing them, and as models improved most centres abandoned it; most current-generation models in international intercomparisons run without it.
Experiments and intercomparison
- Control run: fixed pre-industrial forcing, used to measure drift and internal variability.
- Historical run: observed natural and anthropogenic forcings.
- Scenario runs: future forcing pathways (the Shared Socioeconomic Pathways, SSPs, in the sixth phase of the Coupled Model Intercomparison Project, CMIP6).
- Idealised runs: abrupt quadrupling of CO\(_2\), or 1% per year increase, used to diagnose equilibrium climate sensitivity (ECS) and transient climate response (TCR).
The IPCC Sixth Assessment Report (2021) assessed ECS as likely 2.5–4 °C, with a best estimate of 3 °C, combining models with observational and paleoclimate evidence.
Common coupled-model biases
| Bias | Description | Likely contributors |
|---|---|---|
| Double ITCZ | a spurious second rain band south of the equator in the eastern Pacific | convection and cloud parameterisation, SST errors |
| Cold tongue too far west | equatorial Pacific cold tongue extends too far into the warm pool | too-strong trades, ocean mixing |
| Warm eastern-boundary SST | too warm off Peru, Namibia and California | too little stratocumulus, weak coastal upwelling |
| ENSO errors | wrong amplitude, period, spatial pattern or seasonal phase locking in many models | feedback balance (Bjerknes vs heat-flux damping) |
| Weak or non-propagating MJO | in many models | convection parameterisation, moisture sensitivity |
| Dry Indian monsoon | rainfall too low over the Indian land region in many models, often with a cold Arabian Sea SST bias | SST and convection errors; Himalayan orography at coarse resolution |
Monsoon Links
The ocean as part of the monsoon
| Ocean feature | Link to the monsoon |
|---|---|
| Arabian Sea | moisture source for the west coast and much of central India; pre-onset warming in the southeastern Arabian Sea; summer cooling by Somali and Oman upwelling and by wind mixing |
| Bay of Bengal | low surface salinity from river runoff and rain creates a shallow, stable layer and barrier layers, keeping SST high and favouring convection and monsoon depressions |
| Intraseasonal SST | warmer SST tends to lead active convection and cooler SST follows it: the ocean responds to and then feeds back on the BSISO |
| Cross-equatorial transport | monsoon winds drive southward heat transport across the equator in summer |
ENSO and the monsoon
- El Niño tends to bring a deficient monsoon, La Niña a good one, through the shifted Walker circulation and anomalous subsidence over South Asia (see Tropical East–West Circulations).
- The relationship is not one-to-one: some El Niño years have normal rainfall, and some droughts occur without El Niño.
- In the late twentieth century the inverse ENSO–monsoon correlation weakened (Kumar, Rajagopalan and Cane, 1999). Proposed reasons include Eurasian warming, changes in the Walker circulation and the type of El Niño; whether the weakening persists is debated.
- The type matters: central-Pacific warming is associated with a more reliable monsoon deficit than warming confined to the far east.
The Indian Ocean Dipole and EQUINOO
The Indian Ocean Dipole (IOD), described by Saji and colleagues and by Webster and colleagues in 1999, is a coupled mode with SST anomalies of opposite sign in the western and southeastern tropical Indian Ocean. Its index, the Dipole Mode Index (DMI), is the western box SST anomaly minus the eastern box anomaly.
| Phase | SST | Winds and convection | Monsoon tendency |
|---|---|---|---|
| Positive IOD | warm west, cool southeast (near Sumatra–Java) | anomalous easterlies along the equator; convection shifts west | tends to favour Indian summer monsoon rainfall; can offset El Niño |
| Negative IOD | cool west, warm southeast | anomalous westerlies | tends to be unfavourable |
The IOD develops in boreal summer and peaks in September–November. Gadgil and colleagues (2004) identified an atmospheric counterpart, the Equatorial Indian Ocean Oscillation (EQUINOO): a see-saw of convection between the western and eastern equatorial Indian Ocean, which together with ENSO explains much of the year-to-year variation in Indian monsoon rainfall.
Other drivers
| Driver | Proposed link |
|---|---|
| Eurasian and Himalayan snow cover | heavy winter–spring snow delays land heating and tends to weaken the monsoon |
| Atlantic multidecadal variability | warm North Atlantic phases tend to favour a stronger monsoon on decadal scales |
| Pacific decadal variability | modulates ENSO teleconnections to India |
| Tropospheric biennial oscillation | a tendency for a strong monsoon year to be followed by a weak one through coupled Indo-Pacific feedbacks |
| Aerosols | absorbing and scattering aerosols over South Asia alter surface heating and the land–sea contrast; implicated in part of the observed late-twentieth-century decline in monsoon rainfall |
Monsoon indices
- Webster–Yang index: the vertical shear of zonal wind, \( U_{850} - U_{200} \), averaged over 0–20° N, 40–110° E. A strong monsoon has strong low-level westerlies and strong upper-level easterlies, so a large positive value.
- All-India summer monsoon rainfall as a percentage departure from the long period average (deficient below −10%, excess above +10%).
- Tropospheric temperature gradient between a box over the Asian land region and a box over the equatorial Indian Ocean: the monsoon season is when the land box is warmer.
Prediction
| Approach | Basis | Limitation |
|---|---|---|
| Statistical | regressions on predictors such as ENSO indices, Eurasian snow, SST patterns, pressure | relationships change over time; risk of overfitting |
| Dynamical (coupled models) | initialised coupled ocean–atmosphere GCMs run for months | biases; the monsoon’s large internal (chaotic) variability limits the predictable fraction |
| Multi-model ensembles | combine several coupled models | better calibrated, still limited skill |
India’s Ministry of Earth Sciences launched a Monsoon Mission to improve dynamical prediction, building on a coupled forecast system adapted from the US National Centers for Environmental Prediction; India Meteorological Department seasonal forecasts now draw on this coupled system together with multi-model ensembles. The Indian Institute of Tropical Meteorology also developed an Earth system model (IITM-ESM) that took part in CMIP6.
The monsoon in a warming climate
Two effects compete:
- Thermodynamic: warmer air holds more moisture (about 7% per kelvin), which tends to increase monsoon rainfall and especially heavy-rain extremes.
- Dynamical: a weakening tropical overturning circulation and changes in land–sea contrast tend to weaken monsoon winds.
Observations show an increase in widespread extreme rain events over central India since about the 1950s, alongside a decline in mean summer rainfall over parts of the Indian region during the second half of the twentieth century. Climate models project an increase in South Asian monsoon precipitation in the long term under continued warming, with large uncertainty in regional detail.
Memory Tricks
Don’t Confuse
| These | Difference |
|---|---|
| ECS vs TCR | equilibrium warming for doubled CO\(_2\) vs warming at the time of doubling with CO\(_2\) rising 1% per year; TCR is smaller because of ocean heat uptake |
| AMIP run vs coupled run | SST prescribed (no ocean feedback) vs SST computed by an interactive ocean |
| Flux adjustment vs coupler | an artificial correction to fluxes vs the software that exchanges physical fluxes between components |
| IOD vs EQUINOO | oceanic SST dipole (DMI) vs atmospheric see-saw of equatorial convection and winds |
| Webster–Yang index vs all-India rainfall | circulation (wind shear) index vs rainfall amount |
| Mixed-layer vs deep-ocean response time | years vs centuries to millennia |
Common Mistakes
Practice
All questions below are original practice questions written for this page. They are not from any CSIR NET paper. Take \( \rho = 1025 \) kg m\(^{-3}\) and \( c_w = 3990 \) J kg\(^{-1}\) K\(^{-1}\) for sea water, \( c_p = 1004 \) J kg\(^{-1}\) K\(^{-1}\) for air, \( g = 9.81 \) m s\(^{-2}\) and surface pressure 1013 hPa.
Level 1: Foundation
- Write the e-folding response time of a slab mixed-layer climate model.
- Define the Dipole Mode Index.
- What is flux adjustment?
Answers
- \( \tau = C/\lambda \): heat capacity per unit area divided by the climate feedback parameter.
- Western tropical Indian Ocean SST anomaly minus southeastern tropical Indian Ocean SST anomaly.
- A fixed, non-physical correction added to the heat and freshwater (sometimes momentum) fluxes passed between atmosphere and ocean to prevent coupled-model drift.
Level 2: Part B Style
Practice Q1. The depth of sea water whose heat capacity equals that of the entire atmospheric column above it is closest to:
(a) 0.25 m (b) 2.5 m (c) 25 m (d) 250 m
Answer: (b) about 2.5 m
\( C_{atm} = c_p p_s/g = 1004\times 1.013\times 10^{5}/9.81 = 1.04\times 10^{7} \) J m\(^{-2}\) K\(^{-1}\). Depth \( = 1.04\times 10^{7}/(1025\times 3990) = 2.5 \) m.
Practice Q2. What net surface heat flux is needed to warm a 50 m ocean mixed layer by 1 K in 30 days?
(a) about 8 W m\(^{-2}\) (b) about 79 W m\(^{-2}\) (c) about 790 W m\(^{-2}\) (d) about 2,000 W m\(^{-2}\)
Answer: (b)
\( Q = \rho c_w h\,\Delta T/\Delta t = 1025\times 3990\times 50\times 1/(30\times 86{,}400) = 2.04\times 10^{8}/2.59\times 10^{6} \approx 79 \) W m\(^{-2}\). Intraseasonal flux anomalies of tens of W m\(^{-2}\) can therefore change SST by a fraction of a degree to a degree over an MJO or BSISO cycle.
Practice Q3. In a slab-ocean model with a 50 m mixed layer and climate feedback parameter 1.5 W m\(^{-2}\) K\(^{-1}\), the e-folding response time is closest to:
(a) 4 months (b) 1 year (c) 4 years (d) 40 years
Answer: (c)
\( C = 1025\times 3990\times 50 = 2.04\times 10^{8} \) J m\(^{-2}\) K\(^{-1}\); \( \tau = C/\lambda = 1.36\times 10^{8} \) s \( \approx 4.3 \) years. Real transient response is slower still, because heat is mixed into the deep ocean.
Practice Q4. In a particular season the 850 hPa zonal wind averaged over 0–20° N, 40–110° E is +9 m s\(^{-1}\) and the 200 hPa zonal wind over the same box is −14 m s\(^{-1}\). The Webster–Yang index is:
(a) −23 m s\(^{-1}\) (b) −5 m s\(^{-1}\) (c) +5 m s\(^{-1}\) (d) +23 m s\(^{-1}\)
Answer: (d)
\( U_{850} - U_{200} = 9 - (-14) = +23 \) m s\(^{-1}\). Strong low-level westerlies with strong upper easterlies: a vigorous monsoon circulation. Adding the two winds (b or c) is the usual slip.
Level 3: Part C Style
Practice Q5. At a section in the subtropical North Atlantic the overturning circulation is 17 Sv, and the northward-flowing upper branch is on average 18 K warmer than the southward-flowing deep return flow. Which statements are correct?
- The overturning heat transport is about 1.25 PW.
- A 20% weakening of the overturning with unchanged temperatures would reduce the transport by about 0.25 PW.
- The estimate ignores the gyre (horizontal) component of heat transport.
- Like the other oceans, the Atlantic carries heat poleward (southward) in the Southern Hemisphere.
(a) 1, 2 and 3 only (b) 1 and 4 only (c) 2 and 3 only (d) 1, 2, 3 and 4
Answer: (a) 1, 2 and 3 only
- 1 true: \( H = \rho c_w\Psi\Delta T = 1025\times 3990\times 1.7\times 10^{7}\times 18 = 1.25\times 10^{15} \) W.
- 2 true: \( H \propto \Psi \), so 20% less is about 0.25 PW less.
- 3 true: horizontal gyre circulation (warm western boundary current north, cooler interior flow south) adds a separate component.
- 4 false: the AMOC carries heat northward throughout the Atlantic, so in the South Atlantic the ocean heat transport is northward, towards the equator. This equatorward transport in a southern ocean is a well-known peculiarity of the Atlantic overturning.
Practice Q6. Which statements about coupled ocean–atmosphere models are correct?
- An AGCM forced with observed SST cannot represent the cooling of the sea surface by monsoon rain and cloud, which limits its simulation of monsoon variability.
- Flux adjustment corrects the underlying physical errors that cause drift.
- A slab ocean with prescribed ocean heat transport is adequate for estimating equilibrium climate sensitivity but cannot simulate ENSO dynamics.
- A double ITCZ in the eastern Pacific is a common coupled-model bias.
(a) 1, 3 and 4 only (b) 1 and 2 only (c) 2 and 4 only (d) 1, 2, 3 and 4
Answer: (a) 1, 3 and 4 only
- 1 true: prescribed SST cannot respond to the atmosphere, so the atmosphere-to-ocean forcing in the monsoon region is missing.
- 2 false: flux adjustment compensates for errors without fixing them, which is why it was abandoned as models improved.
- 3 true: a slab has no ocean dynamics (no thermocline, no equatorial waves), so no Bjerknes–wave oscillation; it does reach equilibrium quickly, which suits ECS estimates.
- 4 true: the double ITCZ is a long-standing bias.
Practice Q7. The table gives illustrative June–September values for six years: Niño 3.4 SST anomaly (°C), the Dipole Mode Index (°C) and all-India monsoon rainfall departure from the LPA (%).
| Year | Niño 3.4 | DMI | Rainfall departure |
|---|---|---|---|
| A | +1.5 | +0.9 | +2 |
| B | −1.0 | −0.3 | +9 |
| C | +0.2 | +0.1 | −1 |
| D | +1.1 | −0.4 | −14 |
| E | −0.6 | 0.0 | +6 |
| F | +0.9 | +0.5 | −3 |
Which statements are correct?
- The correlation coefficient between Niño 3.4 and rainfall departure is about −0.69.
- Only one of the six years is a deficient monsoon by the ±10% rule.
- Year A is consistent with a strong positive IOD offsetting El Niño.
- With six years the correlation in statement 1 is significant at the 5% level (two-tailed).
(a) 1, 2 and 3 only (b) 1, 2, 3 and 4 (c) 1 and 4 only (d) 2 and 3 only
Answer: (a) 1, 2 and 3 only
- 1 true: computing Pearson’s \( r \) for the six pairs gives \( r \approx -0.686 \).
- 2 true: only year D (−14%) is below −10%; the others lie within ±10%.
- 3 true: Niño 3.4 +1.5 °C (El Niño) yet near-normal rain, with DMI +0.9 °C: the classic offsetting pattern. Compare year D, El Niño with a negative DMI: deficient.
- 4 false: with \( n = 6 \) (4 degrees of freedom) the two-tailed 5% critical value of \( \lvert r\rvert \) is about 0.81 (from \( t_{0.025,4} = 2.776 \)). Here \( t = r\sqrt{n-2}/\sqrt{1-r^2} \approx -1.89 \), not significant. The sign agrees with theory, but six years are too few to establish it.
Quick Revision Table
| Idea | Fact or formula |
|---|---|
| Atmospheric heat capacity | \( c_pp_s/g \approx 1.0\times 10^{7} \) J m\(^{-2}\) K\(^{-1}\) = about 2.5 m of sea water |
| 50 m mixed layer | about \( 2.0\times 10^{8} \) J m\(^{-2}\) K\(^{-1}\), about 20 atmospheres |
| Slab model | \( C\,dT'/dt = F - \lambda T' \); \( \tau = C/\lambda \) (a few years) |
| Ocean heat uptake | roughly 90% of the excess energy in the climate system |
| Overturning heat transport | \( H = \rho c_w\Psi\Delta T \); AMOC carries heat north throughout the Atlantic |
| Ocean GCMs | Boussinesq, hydrostatic; \( z \), sigma, isopycnal or hybrid; Gent–McWilliams for eddies; long spin-up |
| Coupled model | AGCM + OGCM + sea ice + land via a coupler; flux adjustment largely abandoned |
| Hierarchy | AMIP → slab → AOGCM → ESM |
| ECS (AR6) | likely 2.5–4 °C, best estimate 3 °C |
| Biases | double ITCZ, westward cold tongue, warm upwelling zones, weak MJO, dry Indian monsoon |
| ENSO–monsoon | El Niño → tendency to deficit; relation weakened late 20th century; CP El Niño more effective |
| IOD | DMI = west − east; positive IOD favours the monsoon; peaks Sep–Nov |
| EQUINOO | atmospheric see-saw of equatorial Indian Ocean convection |
| Webster–Yang | \( U_{850} - U_{200} \), 0–20° N, 40–110° E |
| Prediction | statistical, dynamical, multi-model; limited by intraseasonal “noise” |
Key Takeaways
What’s Next
- The Quick Revision sheet for the whole chapter.
- Long-term change, forcing and feedbacks in Climate Change.
- The Indian regional picture in Indian Climatology and Climate Zones.
- Back to the chapter overview.