Page 1© Crown copyright 2004 Simulated Future Changes in Extreme Water Levels Jason Lowe 1, Katja Woth 2, Kathy McInnes 3 June The Hadley Centre, Met Office, UK. 2 GKSS, Geesthacht, Germany. 3 CSIRO, Aspendale, Australia.

Page 2© Crown copyright 2004  The problem and the tools  Surge case studies:  1 – Europe  2 – Australia  Conclusions and key recommendations  Don’t forget about waves

Page 3© Crown copyright 2004 We require credible predictions of future changes in extreme water levels caused by storm surges. 1.Are we able to simulate real surge events with existing surge models driven by numerical weather predictions or climate model simulations of the present day? 2.Can we estimate the future (century scale) time average local sea level? 3.Can we estimate future (century scale) Meteorology?

Page 4© Crown copyright 2004 Two main approaches for surges and waves Dynamic approach  Physically-based models used to simulate storm surge levels and waves in past/present day and future periods. Driven by tidal and meteorological (wind stress and air pressure) forcings across the model domain.  Driving winds and pressure are taken directly from atmospheric climate models for both past/present and future periods or large-scale climate model predictions used to perturb reanalysis winds and pressure.  Do not rely on the past or present relationship between Meteorological drivers and surges being the same in the future.  May be bias even in present day. Statistical approach 1.Relationships between large-scale driving meteorology and extreme water levels are developed for present day. 2.Projection made of future large- scale meteorology using a climate model. 3.Future extreme water levels estimated from 1 and 2.  Don’t need to run dynamic storm surge or wave model for future.  Assumes that the relationship between the large-scale variables and the extreme sea level remain unchanged in a future perturbed climate.

Page 5© Crown copyright 2004 Surge Model h: Surface elevationH: The total depth q: Depth mean currentTs: Wind stress on sea-surface t b : Stress on the sea bottomA: Coefficient of horizontal diffusion  : Water densityC d : Drag coefficients  air : Air density  : Friction coefficient Barotropic shelf seas models: e.g. CSX/CS3, TRIMGEO, barotropic version of POM, GCOM2D b

Page 6© Crown copyright 2004 Are surge models adequate? Comparison of 40+ years hindcast with observations from Woth et al. studies at Cuxhaven. RMS errors on storm surge forecasts from 5 operational European surge models along North sea coasts. Results courtesy of Martin Verlann RIKZ and Martin Holt, NCOF +simulations of Bernier and Thompson for Canadian region (see poster)

Page 7© Crown copyright 2004  The problem and the tools  Surge case studies:  1 – Europe  2 – Australia  Conclusions and key recommendations  Don’t forget about waves

Page 8© Crown copyright 2004 Surge results are region specific  European Von Storch and Reichardt (1997) Langenberg et al. (1999) Flather and Smith (1998) WASA and STOWASUS Lowe et al. (2001) Debernard et al. (2002) Lowe and Gregory (2005) Woth et al. (2005) and Woth (2005)  Australia (North and South) McInnes et al. (2003) McInnes and Hubbert (2003) McInnes et al. (2005)  Bay of Bengal Flather and Khandker (1993) Flather (1994) As-Selek and Yasuda (1995) Unnikrishnan et al. (2006) Mitchell, Lowe, Wood and Vellinga (2006) + CLASIC (ongoing) Note: surges do occur in other regions. The regions highlighted in the position paper are only a sample. They do include NH and SH plus tropical and mid- latitude regimes.

Page 9© Crown copyright 2004 Overview of modelling system Global coupled model Higher resolution atmospheric model Regional climate model Barotropic storm surge model Historic scenario SRES future scenario Tide only Surge plus tide Results & Statistics Estimate of mean SLR Generate 2x30 year regional time slices

Page 10© Crown copyright 2004 Changes in 50-year storm surge height (m) due to changes in storminess. A2 ScenarioB2 Scenario 2080s minus present day.

Page 11© Crown copyright 2004 Changes in 50-year water level (m) due to changes in storminess, mean sea-level rise and vertical land movement. A2 ScenarioB2 Scenario 2080s minus present day.

Page 12© Crown copyright 2004 Simulated extreme water levels (m) for Immingham. 2080s and present day. SRES A2 scenario for 2080s 2080s includes changes in storminess, mean sea-level rise and vertical land movements.

Page 13© Crown copyright 2004 All models all SRES Include uncertainty in ice parameters IPCC TAR range of global sea-level rise

Page 14© Crown copyright m Sea level rise regional variations due to thermal expansion and ocean circulation changes only Source:IPCC

Page 15© Crown copyright 2004 Comparison of storm surge predictions (50-year surge height [m]). Changes are due to future changes in storminess. ECHAM4 2*CO2 GEV HadCM2/HadRM2 IS92a Gumbel HadCM3/HadAM3H/HadRM3 SRES A2 GEV Lowe, Gregory and Flather, 2001 STOWASUS (from R Flather, POL) Lowe and Gregory, 2005

Page 16© Crown copyright 2004 An alternative examination of uncertainty by Woth et al. Domain Spread due to choice of downscaling RCM was less important Spread due to driving GCM and scenario See poster by Katja Woth 99.5th

Page 17© Crown copyright 2004  The problem and the tools  Surge case studies:  1 – Europe  2 – Australia  Conclusions and key recommendations  Don’t forget about waves

Page 18© Crown copyright 2004 Methodology + see poster by Kathy McInnes  Historic  Identify population of sea level events in historical record  Model under current conditions using reanalysis plus surge model  Future 2070  Since storm surges are driven by mid latitude westerlies, analyse changes in surface winds in climate models  Range of change in wind speed determined from analysis of 13 climate models using pattern scaling technique which regresses wind against model’s global warming signal, then scales to temperature uncertainty range  Changes applied as a perturbation to current climate winds and surge model was rerun

Page 19© Crown copyright 2004 Future (2070) extreme levels (m)

Page 20© Crown copyright 2004 Combining storm surge and tide using Monte-Carlo sampling. 100 yr event. storm surge + astronomical tide = storm tide Land subsidence could add a further 1 m AtSurge level (m)Storm tide level (m) Storm tide level (m) with wind speed (High) increase Storm tide level (m) with wind speed increase and SLR (49cm) Lakes Entrance (adds 0.09)1.56 (adds 0.49) Further downscaled with nested higher resolution model 100 year events

Page 21© Crown copyright 2004 Modelling conclusions and recommendations  European examples show importance of both mean sea level change and changes in storminess for projections of future extreme water levels. In present studies both uncertainties are probably underestimated.  Australian example shows dominance of mean sea level uncertainty in projections of future extreme water levels. This is probably underestimated – e.g. no MSL pattern information.  The current studies do not provide information on the shape of the uncertainty distribution. This would be useful for risk calculations.  The results need to be linked to credible inundation models with knowledge of defences (where appropriate) at more sites.

Page 22© Crown copyright 2004 Waves  Damage coastal defences plus lead to additional overtopping  WASA and STOWASUS → future increases in high waves were found in the north eastern part of the North Atlantic but decreases occurred further southwest.  Correlation with the NAO (e.g. Woolf et al., 2002). Wang et al. (2004) assumed the relationship will continue to hold for predictive purposes.  Caires et al. (2006); Wang and Swail (2006) → significant changes. Most significant changes under the more severe emission scenarios.  Wolf and Woolf (2006) used a dynamic wave model approach to show how different climate change effects (e.g. increase in wind speed or change in wind direction) are likely to alter wave conditions around the United Kingdom.