1. Amanda Maycock & Piers Forster
Impact of forcing agents on the width of the tropical belt in PDRMIP models
Tom Wood
Supervised by:
Amanda Maycock & Piers Forster

2. Introduction Tropical Expansion Expansion of the Hadley Cells
ARID ZONE
Expansion of the Hadley Cells
into sub-tropics
Intensification & poleward shift of hydrological cycle
‘LUSH’ ZONE

3. Magnitude and Rate of Expansion
Observations:
0.25° – 3.00° decade-1 since 1979 (Davis & Rosenlof 2012)
Satellite observations
Radiosonde data
Reanalysis
Total expansion estimates over last years range between 2° - 5°
Could be up to 8°
Reproduction in models tends to be slower than observations
Models tend to estimate rate of expansion at around 0.1 – 0.2 ° decade-1 (eg. Hu & Fu, 2007)
Closest statistically significant estimates: 1.0 – 1.5 ° decade-1 between 1979 – 2009 (Davis & Rosenlof 2012)
Sign of change consistently in same direction i.e. +ve in NH and –ve in SH
Hemispherically asymmetric
Stronger in SH than NH (Davis et al, 2016)
Varies with longitude
Highly dependent on metric used to define tropical edge

4. Metrics used to define tropical edge
Mean meridional streamfunction.
Latitudes of the subtropical jet streams.
Latitudes of the subtropical tropopause breaks / tropopause height
Precipitation – evaporation (P - E) (especially poleward zero crossing point)
Outgoing longwave radiation (OLR)
Extratropical storm tracks
Subtropical dry zones
Total column ozone
Subtropical ridges in sea level pressure
Latitude of maximum downwelling in the Hadley cell.
Width varies by more than 10° latitude depending on the metric used

5. Metrics used to define tropical edge
Latitude of edge of Hadley Cells (∫Ψdp)
(vertically averaged mean meridional streamfunction)
Latitude of tropopause breaks (Δθ)
Latitude of subtropical jet (Umax)
Latitude of maximum downwelling in the Hadley Cell (∂y∫Ψdp)
500-hPa mean meridional streamfunction (Ψ500)
(Grey contour indicates Ψ = 0)
Latitude of zero zonal-mean surface wind (Usfc)
Precipitation – Evaporation (P-E)
Figure from Davis & Birner, JOC (2017)

6. Possible Drivers of Tropical Expansion
Tropospheric warming in the Hadley cells
Increases in GHG concentrations
Stratospheric cooling
Ozone
stratospheric water vapour
Mean SST and meridional temperature gradients
Black carbon, aerosols & tropospheric ozone
Especially in NH (Allen et al., 2012)
Natural cycles e.g.:
Pacific decadal oscillation (PDO)
El Nino Southern Oscillation (ENSO)
Southern annular mode (SAM)
Recent studies (eg. Mantsis et al, 2017) have emphasised natural cycles as playing significant role
Drivers likely to depend on the hemisphere in question

7. Current Work
Change in poleward location of zonal mean P-E = 0 in Southern Hemisphere
Responding to PDRMIP coupled idealised perturbation experiments:
abrupt2xCO2
abrupt3xCH4
abrupt5xSul
abrupt10xBC
abruptSol
Why edge of subtropical dry zone (P-E=0) metric?
Location could have significant impacts on population living in and around dry zones
Important metric in PDRMIP
Relatively straightforward calculation
Why Southern Hemisphere?
Circulation more zonally symmetric than NH
Less land than NH makes SH analysis more simple
Response tends to be stronger in the SH than NH (Davis et al., 2016)

8. Current work 2xCO2 3xCH4 5xSul 10xBC 2%Sol P-E=0 MMM
P-E=0 MMM 10-year running mean
P-E=0 MMM
GMST MMM 10-year running mean
GMST MMM
5xSul
10xBC
2%Sol

9. Current Work CCCma tends to respond most strongly to forcings
2xCO2 and 2%Sol result in largest movement of P-E=0

10. Current Work CCCma has largest response ratio in 4/5 experiments
CCCma appears to be highly sensitive to BC
Response to CH4 most sensitive
Response to BC has large inter-model spread
CAM5 produces slight equatorward movement
Sulphate perturbation has lowest sensitivity
System more sensitive to expansion than contraction?
Large inter-model spread result of model differences?

11. Some Open Questions
Why does 5xSul have a lower sensitivity than others
Contraction rather than expansion?
Some characteristic of the forcing agent?
What accounts for the large inter-model spreads?
And why are the spreads different for different forcings?
Why does BC perturbation result in contraction in CESM1-CAM5?
What makes certain models so sensitive to certain perturbations?
What is the regional variability?

12. Future Work
Examine further aspects of the zonal-mean Hadley cell circulation
Examine other metrics for tropical width including:
Mean meridional streamfunction = 0 at 500 hPa (Ψ500)
Latitude of mid-latitude eddy driven jet (u850)
Examine responses of upper tropospheric metrics vs lower tropospheric metrics to different forcings
Examine inter-model error
Analyse precipitation and evaporation responses separately
Analyse fixed SST experiments
Fast adjustments
Analyse the vertical atmospheric stability
i.e. upper tropospheric temperature changes vs lower tropospheric temperature changes.
These will respond differently depending on the forcing used
Analyse results alongside maps of surface temperature response
Some forcings produce more localised results than others
Temperature response will vary by hemisphere
Closely examine initial response to forcings i.e. first ~20 years after abrupt perturbation
Inter-model differences between rate of response
Differences between rate of response to different forcings
Analyse signal to noise issues
Examine seasonal responses