1. Jennifer Catto Supervisors: Len Shaffrey and Kevin Hodges Extra-tropical cyclones and Storm Tracks

2. Introduction Higher resolution atmosphere models should be able to represent structures of storms better – e.g. fronts. Will compare HiGEM, HadGEM and ERA-40 would like to look at ECMWF interim reanalysis in the future. Using Kevin Hodges’ tracking program and some case studies to do this.

3. Feature tracking Example of storm tracks – pressure tracks generated within 75W to 65W and 30N to 40N. Using Kevin Hodges Feature tracking method described in Hoskins and Hodges (2002) Fields filtered to T42 and background field - (wavenumber ≤ 5) removed Once tracks are found they are referenced back to full resolution for further analysis

4. ERA-40HiGEM1.2HadGEM1.2 Vorticity preferred for tracking as it picks up more small scale features and is not an extrapolated field. Both models capture main features of storm tracks including the Mediterranean track. Both too strong in some places especially HiGEM in the Pacific (Cyclones per month per 5° spherical cap) Vorticity (850hPa) Tracking Statistics I - Track Density

5. Vorticity (850hPa) Tracking Statistics I - Track Density Errors Quite different error patterns between models in Atlantic Track density too strong near Greenland Big differences in Mid-East Pacific - Tracking of trailing cold fronts? HiGEM1.2 – ERA-40HadGEM1.2 – ERA-40HadGEM1.2 – HiGEM1.2 (Cyclones per month per 5° spherical cap)

6. Vorticity (850hPa) Tracking Statistics II - Genesis Density Both models capture main features of genesis including Mediterranean cyclogenesis. Overestimated genesis in mid-Pacific in HiGEM Underestimated genesis in mid-Atlantic in both models. ERA-40HiGEM1.2HadGEM1.2 (Cyclones per month per 5° spherical cap)

7. Summary so far… Tracking statistics show that both models do a reasonable job representing the storm tracks. Large differences between the models – HiGEM generally produces more storms. Have looked at atmosphere only runs, SSTs, baroclinicity. Want to investigate differences further by:  looking at individual cyclones in the models – case studies  investigating the mechanisms producing these differences – PDFs of cyclone variables

8. Case Studies - MSLP Mean Sea Level Pressure (hPa) HiGEM1.2HadGEM1.2

9. Case Studies - vorticity HiGEM1.2HadGEM1.2 850hPa Vorticity (x10 -5 s -1 )

10. Maximum intensity – vorticity on vorticity tracks HiGEM storms a lot more intense than HadGEM. More high intensity storms in the Atlantic than Pacific

11. Maximum intensity – pressure on vorticity tracks Smaller difference between the two models – what we expect as vorticity is more sensitive to scale. Indicates that storms may be of smaller scale in HiGEM.

12. Growth Rates Some cyclones of larger deepening rates in HiGEM than HadGEM Difference more obvious in vorticity tendency – many more rapidly intensifying cyclones Pressure tendencyVorticity tendency

13. Conclusions so far… HiGEM produces more storms generally than HadGEM – shown by track density and genesis density. There are more extreme high intensity storms in HiGEM than HadGEM. Storms in HiGEM are smaller scale and more coherent as can be seen in the case studies. It is possible that there is more secondary cyclogenesis in HiGEM than HadGEM.

14. Future Work Investigate mechanisms further: Spatial analysis Latent heating - vertical velocity and precipitation distributions Deformation strain Greenland – tip-jets. Atmosphere only model runs. ECMWF interim reanalysis Teleconnections. Role of coupling to the ocean. Climate change run (1% CO 2 increase per year) of HiGEM1.2

15. Different kinds of secondary cyclones – from Parker (1998). An early model of a secondary cyclone forming from a wave on the cold front of a primary cyclone – from Parker (1998) after Bjerknes and Solberg (1922)

16. Surface analyses for 12 UTC 25- 28 April 1992 – IOP 3 from FASTEX – from Parker (1998) after Hewson (1993).

17. Vorticity Tracking Statistics II Genesis Density Errors HiGEM1.2 – ERA-40HadGEM1.2 – ERA-40HadGEM1.2 – HiGEM1.2

18. Pressure Tracking Statistics I Track Density Less cyclones overall as pressure tracking only picks up larger scales – noticeable in Mediterranean especially. ERA-40 HiGEM1.2HadGEM1.2 (Cyclones per month per 5° spherical cap)

19. Pressure Tracking Statistics I Track Density Error Similar error patterns to vorticity tracking Large difference in Mid-East Pacific – so probably not tracking trailing cold fronts. HiGEM1.2 – ERA-40 HadGEM1.2 – ERA-40HadGEM1.2 – HiGEM1.2 (Cyclones per month per 5° spherical cap)

20. Pressure Tracking Statistics II Genesis Density ERA-40HiGEM1.2HadGEM1.2 (Cyclones per month per 5° spherical cap) Differences around Japan – sea ice? More secondary cyclogenesis in HiGEM than HadGEM?

21. Pressure Tracking statistics II Genesis Density Error HiGEM1.2 – ERA-40HadGEM1.2 – ERA-40HadGEM1.2 – HiGEM1.2 (Cyclones per month per 5° spherical cap)

22. Case studies – vertical velocity 700hPa vertical velocity (Pas -1 )

23. Eady Growth Rate Maximum- A measure of the baroclinicity where Mean growth and decay rate from ERA-40 from 850hPa vorticity Eady Growth Rate for ERA-40 calculated at 700hPa

24. Eady Growth Rate 850hPa ERA-40HiGEM1.2HadGEM1.2 HadGEM1.2 -HiGEM1.2

25. Sea Surface Temperature ERA-40 HiGEM1.2

26. Sea Surface Temperature HadGEM1.2 HiGEM1.2

27. Regions of interest Atlantic Gulf StreamKuroshio Current Mid-PacificMid-Atlantic Pacific

28. Maximum intensity – vertical velocity Values found taking the minimum within a 10° area HiGEM upward velocities stronger than HadGEM Values slightly higher in Pacific than Atlantic

29. Vertical Velocities - Sensitivity to sampling method

30. Growth Rates – Gulf Stream & Kuroshio Current Pressure tendencyVorticity tendency Expect to see larger differences in smaller regions. Vorticity tendency much higher in HiGEM in both Gulf-Stream and Kuroshio current regions. Unusual differences in pressure tendency for Gulf-Stream. Slightly higher number of rapidly deepening cyclones in HiGEM in Kuroshio current region.

31. Secondary cyclones Develop in smaller local baroclinic regions (e.g. trailing cold fronts) Horizontal scales < 1000km Can develop explosively Other factors influencing genesis and development of frontal cyclones Latent Heat Release Deformation strain and stress

32. Surface pressure analysis for 18 GMT 21 st Jan 1995. Showing parent cyclone L A and a wave on its cold front, L U.

33. Regions of interest Atlantic Gulf StreamKuroshio Current Mid-PacificMid-Atlantic Pacific

34. Growth Rates – Mid-Atlantic & Mid- Pacific Pressure tendencyVorticity tendency Much greater number of storms in the Mid-Pacific region in HiGEM than HadGEM – seen previously in track density Same number of Mid-Atlantic storms in the models – much higher vorticity tendency in HiGEM.