Quantifying storm tide risk in Fiji due to climate variability and change
Introduction
Extreme sea levels brought about by tropical cyclone storm surges pose a major threat for many South Pacific islands. For example, in Fiji, Cyclone Gavin in 1997 caused 18 deaths and a damage bill estimated at F$33.4 million 1997 dollars (FMS, 1998) (around $20.9 million US 1997 dollars) and caused extensive flooding of Labasa town, on the north coast of Vanua Levu island, after the storm surge breached sea walls (Terry, 2007).
Storm surges are caused by the inverse barometer effect (IBE) together with surface wind stresses acting over coastal seas, which produces wind setup (see for example, Pugh, 2004). The severity of the extreme sea levels arising from storm surges is also influenced by other variations in sea level that operate on time scales that vary from hours to years. These include wave breaking processes that lead to wave setup and run-up (Walsh et al., 2012) and sea level variations due to modes of climate variability (e.g. Church et al., 2006). Bathymetric depths over the adjacent coastal shelves also influence the severity of storm surges (Kennedy et al., 2012, Hoeke et al., 2013).
The extreme sea levels arising from storm surges are further modulated by astronomical tides, the combination of the two often referred to as a storm tide. A storm surge generally has significantly greater impact if it coincides with high astronomical tide. Astronomical tides vary not only on semi-diurnal or diurnal time-scales, but large variations in maximum tide height also can occur monthly (due to the spring-neap and lunar declination cycles) as well as annually due to semi-annual modulations in tide-producing forces (e.g. Merrifield et al., 2007, Merrifield et al., 2013). These variations are location-specific, but generally locations where diurnal tidal constituents are larger relative to semi-diurnal constituents, i.e. mixed semidiurnal (as at Fiji) and diurnal tidal locations, the highest tides are experienced around the solstices, while locations with more strictly semidiurnal tides tend to produce the highest spring tides around the equinoxes (Chowdhury et al., 2007, Merrifield et al., 2013). These semi-annual tidal variations may be on the order of 10 cm. Clearly, for locations where large semi-annual variations in tides occur, there is potential for tides to make larger contributions to extreme sea levels at certain times of the year. Longer term tidal variations (e.g. lunar nodal and perigee cycles; see for example Haigh et al., 2011) are an order of magnitude smaller and not considered in the context of this study.
On longer time scales, the most dominant mode of climate variability in the Pacific is the El Niño Southern Oscillation (ENSO) phenomenon. The phases of ENSO may be characterised by the Southern Oscillation Index (SOI), which is a measure of MSL pressure at Tahiti minus that at Darwin. During an El Niño event (SOI negative) sea levels are low in the western tropical Pacific and higher to the east while the opposite situation occurs in La Niña conditions (SOI positive) (Collins et al., 2010). Finally, global sea level rise, driven primarily by the thermal expansion of the oceans and a net decrease in terrestrial ice storage, both associated with anthropogenic global warming, will worsen the impacts of severe storm tide events in the future. While there is high confidence that globally averaged sea levels will increase, there is less certainty regarding the regional patterns of change. Global warming-induced changes to other factors such as the frequency and intensity of tropical cyclones and ENSO may also occur, although precisely how these factors will change is less certain than the prediction of future sea level rise (Walsh et al., 2012). Another less explored issue is the effect of climate variability as well as projected changes in tropical cyclone behaviour on storm surge risk. It is likely that the relatively short and sparsely located network of tide gauge records along many tropical cyclone-prone coastlines, which limits the direct observation of historical cyclone storm surges, has contributed to the limited analysis of tropical cyclone storm surge risk (e.g.Walsh et al., 2012, Hoeke et al., 2013).
In regions such as the Pacific, where a statistically robust analysis of storm tide risk from observations is not possible due to the limited spatial coverage and length of tide gauge observations, modelling approaches that simulate the storm tide response from a large population of plausible tropical cyclones over a coastline of interest provide an alternative method for estimating storm tide risk (e.g. Haigh et al., 2013). Such approaches have also been used to investigate scenarios of future tropical cyclone change (e.g. McInnes et al., 2003, Harper et al., 2009, Lin et al., 2012). For example, along the tropical east coast of Australia, Harper et al. (2009) showed that a 10% increase in tropical cyclone intensity, assumed to represent tropical cyclone intensity change in 2050, led to an increase in the 1-in-100 year storm tide level that was considerably smaller than the assumed 0.3 m sea level rise scenario considered. A recent study on hurricane storm surge change for New York City finds that changes in cyclones from two out of four global climate models (GCMs), scaled-up to realistic cyclone numbers via a synthetic cyclone approach, yields storm surge changes that are comparable to the projected change in sea level rise (Lin et al., 2012), while the remaining two show storm surge changes that are considerably smaller. To our knowledge, application of these and similar techniques in the South Pacific region has not been undertaken to date.
The purpose of this study is to use a combination of dynamical and statistical modelling to evaluate the storm tide risk in Fiji and investigate how climate variability and change influence storm surges. Preliminary results of this study were presented in McInnes et al. (2011). Over 90% of Fiji's population are coastal dwellers (Govt. of the Fiji Islands, 2005). In addition, there is considerable sugar cane industry centred around Lautoka and coastal tourism around Nadi on the northwest coast of Viti Levu island, all of which are potentially vulnerable to the coastal hazards arising from tropical cyclones.
The remainder of the paper is organised as follows. Section 2 describes the statistical representation of tropical cyclones under average present climate conditions as well as those associated with La Niña and El Niño. Modification of these relationships to represent future climate conditions is also discussed. Section 3 introduces the models and methods used to evaluate storm tide risk and discusses model performance. Section 4 assesses the variation in sea level in Fiji due to ENSO and summarises future projections for sea level rise in the Fiji region. Section 5 presents the main results of the study, followed in Section 6 by a discussion and conclusions.
Section snippets
Tropical cyclones
This section analyses observed tropical cyclone data over Fiji to provide the basis for stochastic sampling to generate a synthetic cyclone record. The basis for the perturbations to tropical cyclone intensity and frequency used here to represent future climate conditions is also discussed.
Model setup and performance
In this section the hydrodynamic model setup is described. Model performance with tidal forcing is assessed and a comparison is undertaken of storm tides modelled using cyclone winds and pressure derived from the Holland analytical cyclone model with those from gridded winds and pressure from available high resolution reanalyses products.
Sea level variability and change
In addition to variations in cyclone forcing arising from ENSO variability and future climate change, regional sea levels are also expected to vary and this will contribute to the total sea levels during storm tides. This section describes the characteristics of sea level variability on annual and interannual time-scales in Fiji and also provides recent projections of sea level rise for the region. The results presented in this section will be combined with results from the synthetic cyclone
Current climate
Fig. 9 presents the 98th percentile modelled storm tide height estimated from the baseline set of simulations. This level corresponds approximately with a 1-in-200 year event taking into account the frequency of cyclones occurring within the 1.5° radius of the main islands that was imposed for the stochastic cyclones. The highest storm tide values are seen on the northwest coast of Viti Levu and the southern northwest coast of Vanua Levu. Higher sea levels are also evident around the Yasawa
Discussion and conclusions
This study has employed hydrodynamic modelling forced by stochastically-generated cyclones to estimate storm tide risk around the coastline of Viti Levu and Vanua Levu in Fiji. This novel approach allows for statistically meaningful investigation of spatially discrete storm tide risk. It also provides insight into how storm tide risk is affected by changes in cyclone tracks associated with La Niña and El Niño, and on how climate change may further modify storm tide levels.
Under baseline
Acknowledgements
The Pacific Climate Change Science Program of the Australian Department of Climate Change and Energy Efficiency is acknowledged for funding this research.
References (44)
- et al.
Sea-level rise at tropical Pacific and Indian Ocean islands
Glob. Planet. Chang.
(2006) - et al.
Widespread inundation of Pacific islands triggered by distant-source wind-waves
Glob. Planet. Chang.
(2013) Tropical cyclone inundation potential on the Hawaiian Islands of Oahu and Kaua
Ocean Model.
(2012)- et al.
Climate change impacts on tropical cyclones and extreme sea levels in the South Pacific—a regional assessment
Glob. Planet. Chang.
(2012) Climate Change in the Pacific: Scientific Assessment and New Research
Volume 1: Regional Overview. Volume 2: Country Reports
(2011)- et al.
Tropical cyclone activity in the Fiji region: spatial patterns and relationship to large-scale circulation
J. Clim.
(2009) - et al.
Seasonal sea-level forecasts by canonical correlation analysis—an operational scheme for the U.S.-affiliated Pacific Islands
Int. J. Climatol.
(2007) Geodetic and direct mass-balance measurements: comparison and joint analysis
Ann. Glaciol.
(2009)An Introduction to Statistical Modeling of Extreme Values
(2001)The impact of global warming on the tropical Pacific Ocean and El Nino
Nat. Geosci.
(2010)