The influence of Indian Ocean SST on tropical East African rainfall: Insights from observations and

1
The influence of Indian Ocean SST on tropical
East African rainfall Insights from
observations and GCMs

• Emily Black
• Centre for Global Atmospheric Modelling,
  University of Reading
• with help from Julia Slingo, Hilary Spencer and
  Ken Sperber
• emily_at_met.rdg.ac.uk
• www.met.rdg.ac.uk/emily

2
Questions

• How well can the relationship between East
  African rainfall and the Indian Ocean Dipole be
  simulated by an atmosphere only GCM?

3
Questions

• How well can the relationship between East
  African rainfall and the Indian Ocean Dipole be
  simulated by an atmosphere only GCM?
• To what extent is the simulation affected by the
  lack of ocean-atmosphere coupling in an AGCM?

4
Talk outline

• Background
• The observed relationship between the Indian
  Ocean SST and East African rainfall
• The simulation of the relationship between the
  Indian Ocean SST and East African rainfall by
  atmosphere only models
• - Integrations forced with GISST data
• - Integrations forced with idealised anomalies
• The impact of the lack of coupling on the
  simulation
• - Comparison between CGCM and AGCM forced
  with CGCM SST

5
Talk outline

• Background
• The observed relationship between the Indian
  Ocean SST and East African rainfall
• The simulation of the relationship between the
  Indian Ocean SST and East African rainfall by
  atmosphere only models
• - Integrations forced with GISST data
• - Integrations forced with idealised anomalies
• The impact of the lack of coupling on the
  simulation
• - Comparison between CGCM and AGCM forced
  with CGCM SST

6
Background I the Seasonal Cycle
East African rainfall
Rainfall
Month
Short rains
Rainfall
Month
Short rains
7
Background II the Indian Ocean dipole and East
African rainfall
8
Observed relationship between rainfall and Indian
Ocean SST

• Non-dipole years with warm SST in the region of
  interest
• Non-dipole years with cold SST in the region of
  interest

Nino 3
WIO
Rainfall (mm)
EIO
SST anomaly
9
Observed relationship between rainfall and Indian
Ocean SST

• Non-dipole years with warm SST in the region of
  interest
• Non-dipole years with cold SST in the region of
  interest
• Dipole years with warm SST in the region of
  interest
• Dipole years with cold SST in the region of
  interest

Nino 3
WIO
Rainfall (mm)
EIO
SST anomaly
10
Talk outline

• Background
• The observed relationship between the Indian
  Ocean SST and East African rainfall
• The simulation of the relationship between the
  Indian Ocean SST and East African rainfall by
  atmosphere only models
• - Integrations forced with GISST data
• - Integrations forced with idealised anomalies
• The impact of the lack of coupling on the
  simulation
• - Comparison between CGCM and AGCM forced
  with CGCM SST

11
HadAM3 forced with observed SST
Observed rainfall
Modelled rainfall

• the model captures some aspects of the rainfall
  variability
• BUT the extremely high rainfall seen in strong
  dipole years is not replicated

12
Dipole composite winds observed and modelled
Modelled
Observed
Long term mean

• Compare modelled and observed winds
• observed reduction in the flow of wind away from
  the African coast is not simulated by the model
• observed along shore Sumatran coast anomalies are
  not simulated

Dipole composite anomaly
13
Dipole composite winds observed and modelled
Modelled
Observed
Long term mean

• Compare modelled and observed winds
• observed reduction in the flow of wind away from
  the African coast is not simulated by the model
• observed along shore Sumatran coast anomalies are
  not simulated

Dipole composite mean
14
Talk outline

• Background
• The observed relationship between the Indian
  Ocean SST and East African rainfall
• The simulation of the relationship between the
  Indian Ocean SST and East African rainfall by
  atmosphere only models
• - Integrations forced with GISST data
• - Integrations forced with idealised anomalies
• The impact of the lack of coupling on the
  simulation
• - Comparison between CGCM and AGCM forced
  with CGCM SST

15
Forcing HadAM3 with idealised SST anomalies

• Model
• HadAM3 (3.75 x 2.5 lateral resolution and 19
  vertical levels)
• Run in perpetual October mode for 150 days

• Experiments
• Western Indian Ocean forcing (anomalies -2, -1.5,
  -1, -0.5, -0.25, -0.1, 0.1, 0.25, 0.5, 1,
  1.5, 2)
• Eastern Indian Ocean forcing (anomalies 2, -1.5,
  -0.5, 0.5, 1, 1.5, 2)
• Symmetrical dipole forcing (DMI 4, -2, -1, 1,
  2, 4)

16
The effects of Western anomalies, Eastern
anomalies and the SST gradient on rainfall on land
DMI vs rainfall for all runs

• Dominant effect is from the Western Indian ocean
  (r0.91!)
• Eastern Indian Ocean has no discernable effect
• The only impact of the SST gradient (dipole
  forcing) is weakening of the impact WIO negative
  SST anomalies

WIO SSTA vs rainfall for all runs
EIO SSTA vs rainfall for all runs
17
Observed relationship between rainfall and Indian
Ocean SST

• Non-dipole years with warm SST in the region of
  interest
• Non-dipole years with cold SST in the region of
  interest
• Dipole years with warm SST in the region of
  interest
• Dipole years with cold SST in the region of
  interest

Nino 3
WIO
Rainfall (mm)
EIO
SST anomaly
18
Atmosphere only runs

• Atmosphere only model can simulate some aspects
  of the variability of wind and rain within the
  Indian Ocean basin
• BUT
• Atmosphere only model cannot simulate the
  observed relationship between the Indian Ocean
  dipole, low-level wind within the Indian Ocean
  basin and extreme East African rainfall

19
Atmosphere only runs

• Atmosphere only model can simulate some aspects
  of the variability of wind and rain within the
  Indian Ocean basin
• BUT
• Atmosphere only model cannot simulate the
  observed relationship between the Indian Ocean
  dipole, low-level wind within the Indian Ocean
  basin and extreme East African rainfall
• Question
• Is the lack of coupling between ocean and
  atmosphere a significant factor?
• - investigated by comparing coupled model data
  with an atmosphere only run forced with coupled
  model SST.

20
HadAM3 forced with HadCM3 SST
HadAM3
Rainfall (mm/day)
HadCM3
4
Difference
-5
0 10 20
Model Year

• Comparison
• in most years the difference is small (and
  climatology is very similar)
• BUT in some years there is a big difference eg
  year 4

21
HadAM3 forced with HadCM3 SST Year 4 (strongest
dipole)
Atmosphere only
Coupled
Total wind field

• Differences in the wind fields
• CGCM has stronger westerly anomalies over the
  whole basin
• AGCM has a more localised response to the
  strongest gradients

Wind anomaly
Precipitation anomaly
22
Conclusions

• The observed relationship between the Indian
  Ocean dipole and East African rainfall cannot be
  represented by an atmosphere only GCM
• A comparison between AGCM and CGCM simulations
  suggests that this may in part be due to the
  AGCMs lack of ocean-atmosphere coupling