1. 11 TC Activity in WNP, murphree@nps.edu, Oct08 33 rd Annual Climate Diagnostics and Prediction Workshop 21 October 2008 Long Term Changes in Tropical Cyclone Activity in the Western North Pacific Tom Murphree and David Meyer Naval Postgraduate School (NPS) murphree@nps.edumurphree@nps.edu and dwmeyer@nps.edudwmeyer@nps.edu Typhoon Saomai & Tropical Storm Bopha, Western North Pacific, 08Aug06

2. 22 Overview Assessment of changes in tropical cyclone (TC) formations and intensities in western North Pacific during 1970-2006: 1.Intraseasonal to multidecadal changes 2.Long term net changes 3.Relationships between TC changes and known large scale environmental factors (LSEFs) 4.Relationships between TC changes and global warming (GW) Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08 Hypothesis: Assessments of GW impacts on TCs need to account for all LSEFs that affect TCs, and do so at time and space scales at which LSEFs affect TCs.

3. 33 Prior Studies 1.LSEFs strongly affect TC formations (e.g., McBride 1995): a.SST b.Vertical shear c.Vertical velocity d.Absolute vorticity e.Relative humidity 2.ENLN impacts on TC activity (e.g., Chan 2000, Ford 2000): a.Impacts on formation sites, intensities, tracks 3.GW impacts on TCs so far (e.g., Chan and Liu 2004, Emanuel 2005, Webster and Holland 2005, Klotzbach 2006, Landsea 2007): a.Little or no increase in TC numbers b.Increase in TC intensities via SST increases c.Impacts from interannual-multidecadal variations dominate d.TC data inadequate to determine GW impacts e.Limitations of prior studies: 1.Coarse resolution: basin and seasonal averages 2.Focus on one LSEF: SST 3.Uncertainties in TC data 4.Distinction and attribution: GW impacts vs. other impacts Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08

4. 44 Data 1.TC activity: JTWC best track data (no adjustments) 2.LSEFs: NCEP reanalysis 1 and 2 3.Period: 1970-2006 4.Main TC season: May-December 5.Region: western North Pacific (WNP): 0-40° N, 115–185°E 6.Main development region (MDR): 0-20° N, 115–165°E 7.Temporal resolution: one week 8.Spatial resolution: 5°x5° 9.Number of TCs: 989 Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08

5. 5 Long Term Changes in WNP Tropical Cyclone Activity TC Numbers, WNP 1950-2006 1970 Global Surface T Anomaly 1850-2007 1970 From CRU/UEA TC Activity in WNP, murphree@nps.edu, Oct08 Selection of analysis period

6. 66 Methods 1.Identify major temporal and spatial patterns of changes in: a.TCs: formations and accumulated cyclone energy (ACE) b.LSEFs 2.Develop regression models that relate TC formation probability and ACE to local LSEFs at weekly and 5x5 resolutions 3.Validate models using independent TC and LSEF data 4.Use models to assess sensitivity of TC formations and ACE to LSEF variations 5.Relate GW to LSEF changes, and, thus, to TC changes 6.By-product: Models for sub-seasonal forecasting of TC formations and general tropical convection Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08

7. 77 Number of TC Formations, 1970-2006 Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08 Main development region (MDR): 0-20°N, 115-165°E 0° N No. of TCs

8. 88 Long Term Changes in WNP Tropical Cyclone Activity Annual TC Numbers Annual ACE Year TC Number ACE (kts 2 ) Year Changes in TC Numbers and ACE, 1970-2006 TC Activity in WNP, murphree@nps.edu, Oct08 1.Large interannual-multidecadal variations (linked to ENLN) 2.Long term net increases in TC formations and ACE Can changes in LSEFs explain these long term changes in TC activity?

9. 99 LSEF Changes – SST in MDR Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08 Year SST (C)

10. 10 Long Term Changes in WNP Tropical Cyclone Activity LSEF Changes – SST 1.Overall SST increase in MDR (+0.4 °C) is favorable for increases in TC formations and intensities, and consistent with GW. 2.Some SST decreases in TC active areas of WNP. SST increases consistent with long term increases in TC numbers and intensities. TC Activity in WNP, murphree@nps.edu, Oct08 SST Change, JASO, 1970-2006 (°C) TC Unfavorable TC Favorable

11. 11 LSEF Changes – Vertical Velocity in MDR Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08 Vertical Velocity (-Pa/s) Year

12. 12 LSEF Changes – Vertical Velocity Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08 Vertical Velocity Change, JASO, 1980-2006 (-Pa/s) TC unfavorable TC favorable Have vertical velocity decreases countered impacts of SST increases on TC formations? 1.Overall decrease in MDR is unfavorable for TC formation and ACE increases. 2.Other LSEFs (shear, vorticity, RH) show no pronounced long term changes.

13. 13 Regression Modeling of TC – LSEF Relationships 1. TC formation probability = log(p/(1-p)) = a + b·SST + c·VertVel - d·Shear – e·Shear 2 + f·AbsVort + g·Equat 2. ACE = A + B·SST + C·VertVel + D·Shear - E·Shear 2 + F·AbsVort + G·RH Sensitivity to LSEF Changes: Increase in each LSEF by one standard deviation leads to following changes in formation probability and ACE: Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08 LSEF% Increase in Formation Probability Vertical velocity+73.68 Shear-40.51 SST+25.67 Absolute vorticity+1.68 LSEF% Increase in ACE Absolute vorticity+21.20 SST+14.83 Relative humidity+10.31 Shear -8.63 Vertical velocity +3.38 1.TC formation relatively sensitive (insensitive) to vertical velocity (SST). 2.ACE moderately sensitive to SST and relatively insensitive to vertical velocity.

14. 14 Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08 ACE (kts 2 ) Year ACE, 1970-2006: Best Track (Blue) and Modeled (Red) Validation of TC Formation Probability Model 1.Week-block LSEF values used to calculate ACE, given a TC had formed or was present in that week-block. 2.If model ACE within 25% of analyzed scored as a “hit”, then model hit over 90% of the time, including the data from the 1970s and 2006 which were not used in model formation.

15. 15 Validation of TC Formation Probability Model TC Activity in WNP, murphree@nps.edu, Oct08 1.Modeled TC formation probabilities for individual week at 5x5 degree resolution (calculated using independent LSEF data) 2.Developing sub-seasonal TC formation forecasts based on model Modeled TC formation probabilities, 01-07 Oct 2006 10% 25% 40% 55% 70% Verifying observations of TC formations from best track data Long Term Changes in WNP Tropical Cyclone Activity

16. 16 Summary 1.Shortcomings of TC and LSEF data sets are problematic. 2.TC formations and intensities have undergone long term increases since 1970. 3.SST and vertical velocity are the only LSEFs that experienced net long term changes in WNP during 1970-2006. 4.SST (vertical velocity) changes were favorable (unfavorable) for increases in TC formation and intensity. 5.Impacts of SSTs on TC formations may have been minimized by vertical velocity decreases. 6.GW may have contributed to increases in TC formations and ACE, by way of SST increases but mitigated by vertical velocity decreases. 7.But attribution is problematic, given: a.Shortcomings of data sets b.Large interannual to multidecadal variations c.Uncertainties about LSEF-TC relationships and GW signatures 8.LSEFs affect both TC formations and intensities, and need to be considered at realistic time and space scales in assessing GW impacts. Long Term Changes in WNP Tropical Cyclone Activity TC Activity in WNP, murphree@nps.edu, Oct08