The Influence of Future Changes in Tidal Range, Storm Surge, and Mean Sea Level on the Emergence of Chronic Flooding

Sea‐level rise is leading to increasingly frequent coastal floods globally. Recent research shows that changes in tidal properties and storm surge magnitudes can further exacerbate sea‐level rise‐related increases in flood frequencies. However, such non‐stationarity in tide and storm surge statistics are largely neglected in existing coastal flood projection methodologies. Here we develop a framework to explore the effect that different realizations of various sources of uncertainty have on projections of coastal flood frequencies, including changes in tidal range and storminess. Our projection methodology captures how observed flood rates depend on how storm surges coincide with tidal extremes. We show that higher flood rates and earlier emergence of chronic flooding are associated with larger sea‐level rise rates, lower flood thresholds, and increases in tidal range and skew surge magnitudes. Smaller sea‐level rise rates, higher flood thresholds and decreases in sea level variability lead to commensurately lower flood rates. Percentagewise, changes in tidal amplitudes generally have a much larger impact on flood frequencies than equivalent percentagewise changes in storm surge magnitudes. We explore several implications of these findings. Firstly, understanding future local changes in storm surges and tides is required to fully quantify future flood hazards. Secondly, existing hazard assessments may underestimate future flood rates as changes in tides are not considered. Finally, identifying the flood frequencies and severities relevant to local coastal managers is imperative to develop useable and policy‐relevant projections for decisionmakers.


Introduction
Coastal floods in harbors, bays and estuaries are caused by high still water levels (SWLs) (Gold et al., 2023;Li et al., 2022;Woodworth et al., 2019).The occurrence of high water levels depends on how various phenomena coincide, including tides, storm surges, and seasonal variations in mean sea levels.Flooding occurs if the resultant water level exceeds local flood thresholds (Baranes et al., 2020;Enriquez et al., 2022;Kopp et al., 2019;Li et al., 2022;Sweet et al., 2016;Thompson et al., 2021).Variations in tides often leads to coastal flood events being clustered around high astronomical tides, even if such periods do not coincide with large storm surges (Hague, Grayson, et al., 2023;Haigh et al., 2011;Ray & Merrifield, 2019;Talke et al., 2018;Thompson et al., 2021).Long-term tidal modulations such as the 18.6 years nodal cycle also lead to variations in the heights and the likelihood of extreme flooding occurring in a given year (Baranes et al., 2020;Enriquez et al., 2022;Talke et al., 2018).
The frequencies of coastal flooding around the world have increased in response to sea-level rise (SLR) (Hague et al., 2022;Moftakhari et al., 2015;Sweet et al., 2018;Sweet & Park, 2014).SLR raises the heights of typical high tides, reducing the distance between high tide marks and flood thresholds (Dusek et al., 2022;Sweet & Park, 2014).Hence, increases in tidal range reduce the non-tidal contribution required for flooding to occur (Devlin, Jay, Talke, et al., 2017;Devlin, Jay, Zaron, et al., 2017).Tidal range also plays a key role in modulating how flood hazards change under sea-level rise.For example, locations with small tidal ranges are more susceptible to rapid changes in flood rates under SLR (Koroglu et al., 2019;Thieler & Hammar-Klose, 2000).This follows from the ratio of SLR to tidal range being higher at such locations (Ritman et al., 2022;Rueda et al., 2017).For example, for a location with 1 m tidal range, 1 m of SLR causes future low tides to be as high as today's high tides.In contrast, for a location with 2 m tidal range, the same 1 m of SLR means the future mean tide level is the height of present-day high tides.
In some locations, SLR now causes coastal flood thresholds to be exceeded due to tides alone under average weather and oceanic conditions (Dusek et al., 2022;Hague et al., 2022;Li et al., 2022;Ray & Foster, 2016;Ritman et al., 2022).A key impact of further SLR is that infrequent or rare events will become much more frequent.Many studies suggest that floods that have typically occurred once per century will likely occur every decade by mid-century, and annually before 2100, in many locations worldwide (Buchanan et al., 2017;Hunter, 2012;Taherkhani et al., 2020;Tebaldi et al., 2021).Similarly, present-day minor and nuisance flood levels may be exceeded hundreds of days per year by mid-century and daily before 2100 for current levels of exposure (Ezer, 2022;Hague et al., 2020;Ritman et al., 2022;Sweet et al., 2018;Sweet & Park, 2014;Thompson et al., 2019Thompson et al., , 2021)).These minor floods can cause disruption to transportation networks, vegetation damage, salination, and impact infrastructure (Cantelon et al., 2022;Gold et al., 2022;Hino et al., 2019;Kasmalkar et al., 2020).
The emergence and severity of chronic flooding poses a challenge to coastal management (Ghanbari et al., 2020;Gold et al., 2023;Le Cozannet et al., 2021;Moftakhari et al., 2018;Thompson et al., 2019).In many regions, the emergence of chronic flooding is expected to have a greater impact on communities than extreme floods becoming more extreme (Buchanan et al., 2019;Ghanbari et al., 2019;Moftakhari, Aghakouchak, et al., 2017;Paulik et al., 2021;Stephens et al., 2021).Unlike episodic or occasional floods, which are characterized by the severity of their impacts (Hague et al., 2022;Moore & Obradovich, 2020), chronic floods are defined by their frequency (Buchanan et al., 2019;De Leo et al., 2022;Gold et al., 2023;Sweet & Park, 2014;Thompson et al., 2019).Hence the concept of chronic flooding applies equally to minor, major or extreme floods.As they occur at lower sea levels, minor flooding will become chronic first, but eventually more extreme events will as well with continued SLR.Given the known limitations of some coastal protection measures (Nunn et al., 2021;Trace-Kleeberg et al., 2023), and absent adaptation measures, many coastal communities will eventually transition to experiencing chronic major flooding, with future timing depending on local SLR (Ghanbari et al., 2019;Hague et al., 2020;Sweet et al., 2018).This approach is embedded in coastal management frameworks in Australia, with guidelines for the State of Victoria indicating that major engineering works should only be considered if managed retreat is not a viable adaptation option (Department of Energy Environment and Climate Action, 2023).
Despite the potential significance of sea-level variability (tides, storm surge) for flood hazards, most coastal flood hazard frameworks assume that water level variations around the average remain statistically stationary.Future water level distributions are obtained by shifting the historical distribution toward higher sea level, without any Earth's Future 10.1029/2023EF003993 change in the shape of the distribution (Buchanan et al., 2016(Buchanan et al., , 2017;;Fox-Kemper et al., 2021;Ghanbari et al., 2019;Hague et al., 2020;Hunter, 2012;Kirezci et al., 2020;McInnes et al., 2013;Sweet et al., 2018;Taherkhani et al., 2020;Tebaldi et al., 2012Tebaldi et al., , 2021;;Thompson et al., 2019Thompson et al., , 2021)).Several studies have explored how future changes in storminess may impact sea level extremes regionally (Cannaby et al., 2016;Howard et al., 2019;McInnes et al., 2009).Similarly, some studies show that rising sea levels can influence tidal and storm surge magnitudes, and therefore flood hazards, in shallow seas (e.g., Arns et al., 2017).A recent study concluded that global-scale changes in tides, primarily due to SLR, were insignificant compared to other factors (Vousdoukas, Mentaschi, et al., 2018) suggesting that tidal non-stationarity is an insignificant factor in altered flood hazard.However, this study only modeled global-scale changes in tides, and not local or regional scale effects which have been the drivers of the largest historical tidal changes (Haigh et al., 2020;Li et al., 2021;Talke & Jay, 2020).While Ray and Foster (2016) recommended the inclusion of trends in tides be included in projections of flood frequencies, little progress has been made in projection frameworks to the present time.
This study develops a new framework to estimate the possible effect of future changes in mean sea levels and sealevel variability on flood frequencies at local and national scales.We apply our framework to assess future changes in Australian flood frequencies under various combinations of changes in mean sea level, tidal range, and storminess.This provides the first assessment of the sensitivity of chronic flood hazards to trends in tidal range and skew surges continuing, or emerging, in the future.From this assessment we identify the characteristics of locations most at-risk of large changes in chronic flood hazards in response to changes in variability.

Methods and Data
Our coastal flood projection framework produces timeseries of projected daily maximum SWLs under some specified climate pathway at a tide gauge location.We define a climatological baseline period of 1995-2014, and a future projection period of 2020-2150 to align with the baseline and projection periods used in the Intergovernmental Panel on Climate Change's 6th Assessment report (IPCC AR6) (Fox-Kemper et al., 2021).
The coastal flood frequency projection methodology is shown schematically in Figure 1, with further details provided in Sections 2.2-2.9.Our method is a variant of joint probability methods (Baranes et al., 2020;Batstone et al., 2013;Pugh & Vassie, 1978).It is also conceptually similar to methods used to attribute flood events to specific physical processes (Hague et al., 2020;Hague, Grayson, et al., 2023;Li et al., 2022;Ritman et al., 2022).The underpinning assumption is that a tide gauge record can be decomposed into several independent components, which are modeled separately and then recombined (e.g., Kirezci et al., 2020;Li et al., 2022;Lowe et al., 2021).Here we consider four main components: astronomical tides, mean sea level, within-year non-tidal variability and between-year (or interannual) non-tidal variability.The demarcation of these components is influenced by choices made in harmonic analysis and how long-term changes in mean sea level are represented (Hague & Taylor, 2021); thus, slightly different attributions to specific forcing may emerge, depending on specific methodology.The data produced by this study is provided in Hague (2024).These data are described in Supporting Information S1.

Historical Still Water Levels
Still water level (SWL) data are obtained from the Australian National Collection of Homogenised Observations of Relative Sea Level (ANCHORS) (Hague et al., 2021).Data over the baseline period of 1995-2014 were used to evaluate tides, storm surges, and annual mean sea level anomalies.This evaluation formed the basis for projections into the future.The locations included in the ANCHORS data set are shown in Figure 2.

Future Mean Sea Levels
Future mean sea levels are obtained from the Intergovernmental Panel on Climate Change's 6th Assessment report (IPCC AR6) (Fox-Kemper et al., 2021;Garner et al., 2021).The SLR projection closest to the published latitude and longitude for each gauge (Hague et al., 2021) was used for each respective site.We primarily focus on the SSP3-7.0emissions scenario, following the IPCC AR6 Sea-Level Projection Tool (Garner et al., 2021).As in the tool, annual mean sea levels over the projection period of 2020-2150 were obtained through linear interpolation between the provided decadal values.In addition, we consider SSP5-8.5 to evaluate how other SLR scenarios impact the influence of increased tidal ranges and skew surges.The SLR estimates of the SSP3-7.0 and SSP5-8.5 scenarios span the sea level planning guidelines and scenarios used by Australian coastal planners (Norman & Earth's Future 10.1029/2023EF003993 HAGUE AND TALKE Gurran, 2018), and likely bound future SLR trajectories based on current emission commitments (Bamber et al., 2019;Jackson et al., 2018;Rogelj et al., 2023).

Future Astronomical Tides
Astronomical tides are the SWLs expected under average weather conditions.In this study, we considered 114 harmonic constituents, using the TideHarmonics tidal harmonic analysis package (Stephenson, 2017).We fitted the tidal model to hourly observed SWLs over the climatological baseline period (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014).The corresponding annual mean sea level was first subtracted from the hourly values.This ensures that annual mean sea level contributions are not double counted when components are added together in Section 2.6.As 20 years is long Figure 1.Schematic of the coastal flood projection method developed here (a, upper, Section 2.1-2.8) and the climatological projection methodology (b, lower, Section 2.9) used for comparisons in this study.Gray boxes are determined by the analyst, based on the specifics of the future climate storyline of interest.White boxes indicate timeseries of water levels produced during the application of the method.Green boxes denote methodological steps; solid borders indicate a step that is always performed in method.Dotted borders indicate a step that is only performed if variations in tidal ranges and skew surge magnitudes are specified in the future climate in the our framework (i.e., Section 2.10).
Earth's Future 10.1029/2023EF003993 HAGUE AND TALKE enough to consider multi-annual tidal cycles (Haigh et al., 2011;Ray & Merrifield, 2019), no nodal corrections are applied.This model of 114 constituents is used to produce a tidal hindcast over 1995-2014 baseline period, which is used to compute skew surges (Section 2.4).Additionally, a tidal forecast is produced for the 2020-2150 projection period.Each tidal prediction is made at hourly frequency, with constituent amplitudes held constant.This approach is consistent with previous coastal flood projections that have explicitly included future astronomical tides with SLR scenarios (Dusek et al., 2022;Ghanbari et al., 2019;Hague et al., 2020;Ray & Foster, 2016;Thompson et al., 2021).The maximum of (24) hourly tide heights is reported as the tidal contribution to daily maximum water levels (Section 2.6).

Future Within-Year Sea Level Variability
We represent within-year sea level variability due to short-lived meteorological, oceanographic, and (for river influenced locations) hydrologic processes using the skew surge metric.The (daily) skew surge is defined here as the difference between daily maximum observed SWL (i.e., Section 2.1) and daily maximum astronomical tide level (i.e., Section 2.3), following Hague et al. (2022).In most coastal locations, skew surges are found to be statistically independent of the tides (Baranes et al., 2020;Batstone et al., 2013;Santamaria-Aguilar & Vafeidis, 2018;Williams et al., 2016).Where surges and tides are not independent, assuming independence generally results in an over-estimation of extreme water levels and flood risk (Arns et al., 2020;Haigh, Wijeratne, et al., 2014).
Estimates of future skew surges are obtained by sampling an empirical distribution comprised of skew surges over the historical period.We compute the skew surge quantity for every day with complete observations.We first subtract the annual means from these historical SWL so that between-year sea level variability is not included in the skew surge.We generate R (here, R = 1,000) realizations of skew surges for each day in the projection period from this distribution.R is chosen as an arbitrary number that likely encompasses typical variability and enables statistics like the 5% and 95% percentiles to be assessed robustly.
The skew surge term encapsulates the water level response to all sources of short-lived non-tidal sea level forcing, including synoptic storm events, coastal circulation, local and far-field wind forcing, seasonal water temperature shifts, the inverse barometer effect, and river flow effects, both at the coast and in estuaries (Piecuch et al., 2018;Talke et al., 2021).This study adopts the common approach of not differentiating between the drivers of these non-tidal water level contributions, especially since some effects are not completely independent of each other (Hague, McGregor, et al., 2023;Li et al., 2022;Thompson et al., 2021).Instead, possible shifts in non-tidal forcing are assessed through a bulk trend in skew surge (see Section 2.9).Future efforts could investigate the influence of hydrological, meteorological, and oceanic factors on skew surge separately, especially if there is reason to believe that individual factors (such as local wind forcing or river flow) are non-stationary or trending differently from other factors.Historically, such trends have occurred; for example, over secular time scales, both the seasonal timing and magnitude of river discharge to the ocean has shifted in many river basins, both due to climate shifts and water resources management (Moftakhari et al., 2013;Naik & Jay, 2011).But, the influence on near-coastal estuarine stations is generally small during most conditions (<10 cm), even in high discharge river basins such as the Columbia River, USA (Talke et al., 2020).Similarly, the influence of channel deepening and other infrastructure development on mean river water levels is generally small near the coast, except during extreme river flood events (e.g., Talke et al., 2021).Within the Australian context considered here, river discharge magnitudes are generally small and likely have little influence on water levels at most of the coastal and nearcoastal stations considered here.
The climate change impacts of mean sea level on frequent flood threshold exceedances are expected to dwarf the impact of changes in peak river flows (Ghanbari et al., 2021).Changes in rainfall and streamflow may shift annual recurrence intervals from 0.01 to 1 in a small number of locations under high greenhouse gas emission scenarios (He et al., 2022;Hirabayashi et al., 2021).In contrast, sea-level rise in the same scenario causes present-day 1-in-100-year coastal flood expected to occur annually virtually everywhere, and more than 100 days per year in many locations (Boumis et al., 2023;Fox-Kemper et al., 2021;Hague et al., 2020;Taherkhani et al., 2020;Tebaldi et al., 2021).Hence, not explicitly considering freshwater influences on chronic flood hazards should not affect the robustness of this study's findings.Nonetheless, the framework presented here could be further partitioned to examine how hydrological factors impact the emergence of chronic flooding, particularly in regions with large river-induced water level variability like tidal rivers (Hoitink & Jay, 2016).

Future Annual Mean Sea Level Variability
Our projection includes the effect of interannual variations in the annual mean sea level (AMSL).The AMSL anomaly is calculated as the difference between the observed AMSL and an ordinary least squares trendline fitted over the baseline period (i.e., 1995-2014).AMSL anomalies are assumed to be independent of tides and skew surges, and the AMSL trend (Burgos et al., 2018;Lowe et al., 2021;Ritman et al., 2022;Sweet et al., 2016).Like skew surges, estimates of future annual mean sea level variability are obtained by sampling an empirical distribution comprised of annual mean sea level anomalies over the baseline period.We then generate 1000 (i.e., R) realizations of future AMSL anomalies for each year in the projection period.

Future Still Water Levels
Projections of future daily maximum SWLs are obtained by adding the components together: projected mean sea level (Section 2.2), astronomical tides (Section 2.3), sampled skew surges (Section 2.4), and sampled annual mean sea level (AMSL) anomalies (Section 2.5).The same annual value of projected mean sea level is used for Earth's Future 10.1029/2023EF003993 each day in a calendar year.The daily maximum astronomical tide is used as the tidal contribution to daily maximum SWLs.Each of the 1,000 different realizations of skew surges are matched to the corresponding AMSL anomaly (e.g., iteration 23 of skew surge is matched to iteration 23 of AMSL).This provides 1,000 possible future daily maximum water levels under the same predicted astronomical tide for each SLR scenario considered.The components of these future still water level timeseries are provided in Hague (2024) which is described in Supporting Information S1.

Future Flood Days
Annual flood days are the number of days in a calendar year where the daily maximum SWL exceeds a specified flood threshold (Burgos et al., 2018;Hague et al., 2020Hague et al., , 2022;;Sweet et al., 2018;Sweet & Park, 2014;Thompson et al., 2019Thompson et al., , 2021)).Several flood thresholds are considered for each location in the ANCHORS data set.Firstly, we consider the impact-based minor flood levels defined by Hague et al. (2022).Of the 38 ANCHORS locations, 25 have a minor flood threshold defined.However, 15 of these are classified as low-sample thresholds, meaning they are based on less than 10 flood reports.The 10 locations with well-sampled flood thresholds are Cairns, Townsville, Brisbane, Gold Coast, Ballina, Newcastle, Sydney, Melbourne, Port Adelaide, and Fremantle.These impact-based thresholds are supplemented by percentile-based thresholds so projections for all locations can be defined in a consistent way, and flood severities other than minor can be considered.Previous studies have shown that minor floods can occur 10 days per year at many locations (Hague et al., 2022).Accordingly, we consider the 97th, 99th and 99.7th percentiles of daily maximum water levels as thresholds.These are calculated from the observed SWLs over the baseline period, occurring on average 11, ∼4 and 1 day/s per year, respectively.To consider more extreme floods we consider eight additional thresholds ranging from 0.1 to 0.7 m inclusive above the 99.7th percentile level, as most Australian locations have offsets between 1-in-1-and 1-in-100-year return period level of less than 0.7 m (Pattiaratchi et al., 2018;Tebaldi et al., 2021).We also consider future exceedances of the highest water level recorded over the baseline period at each location.The resultant timeseries of threshold exceedances are made available (refer Data Availability Statement).

Chronic Flood Emergence Times
Several studies have sought to quantify the number of annual flood days that are required for flooding to be considered chronic.These all agree that for flooding to be considered chronic it must occur at least several times per year (Ghanbari et al., 2020;Gold et al., 2023;Le Cozannet et al., 2021).Buchanan et al. (2019) found that a scenario where "streets may flood several times per month" led to statistically significant tendency for homeowners to consider relocation.Sweet and Park (2014) proposed a similar definition of 30 flood days per year as representing a tipping point, which once surpassed, results in chronic flooding.Most other published definitions require considerably more flood days to occur in a year for flooding to be considered chronic.For example, de Leo et al. ( 2022) used a definition of 100 flood days per year.Thompson et al. (2019) took a slightly different approach, defining chronic flooding to occur when nine out of 10 consecutive years produced at least 50 flood days.This definition has also been adopted by subsequent studies (Habel et al., 2020;Li et al., 2023).
The emergence time is the first year that a specified number of annual flood days occurs (Hague et al., 2020).To understand how much different definitions of "chronic" affect the results of this and previous studies, we consider four different emergence times of chronic flooding: -The first year where 30 or more flood days occur (i.e., Sweet & Park, 2014).
-The first year where 100 or more flood days occur (i.e., de Leo et al., 2022).
-The first year where at least nine of the next 10 years have had least 50 flood days occur (i.e., Thompson et al., 2019).-The first year where 50 or more flood days occur.
The final definition is included to assess the effect of requiring many consecutive years of flooding for it to be considered chronic (e.g., Thompson et al., 2019), rather than just a single year (e.g., the first two definitions).

Projections Considering Changes in Tidal Ranges and Skew Surges
The structure of our framework allows for straightforward modifications that consider the impact of changes in tidal range and skew surge magnitudes on coastal flood frequencies and emergence times of chronic Earth's Future 10.1029/2023EF003993 flooding.For tides, the annual daily maximum tide levels from Section 2.3 are multiplied by some factor (with respect to mean tide level).For skew surges, this is achieved by multiplying sampled skew surge values from Section 2.4.This approach allows consideration of spatially variable trends in tide and surges if desired.We do not consider changes in the magnitude of AMSL anomalies.Whilst variability in sea surface height and AMSL may increase with climate change, further work is required to understand the flood risk implications (Widlansky et al., 2020).
To demonstrate a sensitivity testing approach, we consider increases and decreases of 1% per decade in both skew surge magnitudes and tidal amplitudes.For example, with a 1% per decade increase in tidal amplitude, a 2 m tide (above annual mean sea level) in 2020 becomes a 2.02 m tide in 2030 and 2.2 m tide in 2120.Similarly, a 0.2 m skew surge in 2020 becomes a 0.202 m skew surge in 2030 and a 0.22 m skew surge in 2120.In absolute terms, larger magnitude tides and skew surge increase more.Whilst trends in tides and surges do vary spatially in Australia (Hague, Grayson, et al., 2023;Mawdsley & Haigh, 2016), considering a spatially uniform trend enables standardized comparisons of the relative importance of future trends in sealevel, tides, and skew surge at different sites, and comparison to other sources of uncertainty in coastal flood projections.
Australian locations have been included in several global and regional studies on changes in tidal range and tidal constituents (Devlin, Jay, Talke, et al., 2017;Devlin, Jay, Zaron, et al., 2017;Mawdsley et al., 2015;Schindelegger et al., 2018;Woodworth, 2010) and two national studies (Hague, Grayson, et al., 2023;Harker et al., 2019).Schindelegger et al. (2018) found that a sea-level rise of 0.5 m produced changes of roughly 1/+2 cm in the M2 and K1 constituents in the coastal Australian ocean, depending on location.Harker et al. (2019) found small positive and negative changes in tidal amplitudes of up to 1% per decade but did not quantify how these changes impacted flood frequencies or heights of sea level extremes.A more recent study (Hague, Grayson, et al., 2023) found a much larger, 29% change, likely due to changes in the dredging regime, at one location-Lakes Entrance, Victoria.This study also found that approximately one-third of all Australian locations had statistically significant trends in the M2 constituent (the major component of tidal range in semidiurnal systems), ranging from 1.24% to +3.54% per decade.These studies justify the consideration of a hypothetical 1% per decade increases and decreases in tidal amplitudes in our sensitivity testing approach.
Fewer studies have considered changes in skew surges in the Australian region.One recent study modeled changes in extreme water levels due to changes in weather patterns (Colberg et al., 2019).The results of this study are difficult to apply within our framework for several reasons.Firstly, non-tidal residuals were used instead of skew surges.Secondly, results are expressed as absolute changes in 1-in-20-year non-tidal residual return levels rather than a multiplying factor that applies to all surge magnitudes.Thirdly, changes in ocean circulation are not considered, and they omit tides from projections.These methodological differences matter because unlike skew surge, the non-tidal residual is not independent of tides, meaning large non-tidal residuals may not be due to atmospheric factors (Williams et al., 2016).This makes it is hard to interpret the impact of changes in non-tidal residual on flood hazards without the accompanying tidal heights.A global study that included Australian locations has more useful results for our purposes.Mawdsley and Haigh (2016) found statistically significant decreases in extreme skew surge of order 1%-2% across northern Australia, and positive trends of similar magnitudes at isolated locations in the south.These results suggest changes of around 1% per decade are plausible.

Climatological Projections
Our projection method differs to the approach used by recent studies by explicitly estimating the tide, surge, and annual mean sea level components in projected water levels.Previous studies have estimated future changes in threshold exceedances under SLR.These studies define a 20 calendar-year climatology (i.e., an empirical distribution) of daily maximum SWLs.For each year and future climate pathway, all the values in this climatology are increased by some amount corresponding to the relevant SLR increment (Dusek et al., 2022;Hague, McGregor, et al., 2023;Ritman et al., 2022).The average number of annual flood days is then computed from this new climatology.This is the number of days where the daily maximum water exceeds a threshold of interest divided by the number of data-years in the climatology.This approach retains the autocorrelation structure of the water level timeseries but doesn't model surge and tide separately.We generate projections using this climatological method to understand how these methodological differences impact estimates of chronic flood emergence.

Impact of Coinciding Independent Tidal and Non-Tidal Water Level Contributions on Flood Frequencies
A key feature of our approach (i.e., Section 2.6) is that flood frequencies can be estimated probabilistically by taking percentiles of the R realizations (here 1000) of future annual exceedances of water levels.This enables quantification of the probability of joint events, such as the probability that large storm surges coincide with high spring tides.As also observed in other joint-probability approaches (e.g., Baranes et al., 2020), including this aleatoric uncertainty improves the replication of between-year variability like what is seen in observed flood rates (Figures S1 and S2 in Supporting Information S1).Here we primarily consider the 5th, 50th (median) and 95th percentile estimates of the possible maximum number of annual exceedances over a decade (Figure 3).This was computed by first finding the 50th, 500th, and 950th largest of the 1,000 realizations of annual flood days for each site then taking the largest annual estimate across a decade.We also provide a national median value for each of these (i.e., median of all 5th percentile estimates in Figure 3a).These probabilities are conditional on the mean sea level scenario specified, as are all other projections of future flood hazards (e.g., Thompson et al., 2021).Future changes in tides and skew surges are not considered here.
We show that variability in realized flooding is large, even when a single SLR scenario is considered (SSP3-7.0median).For example, 26 (of 38) locations have a 95% chance of experiencing at least 1 year with 20 exceedances of the present-day once a year level (i.e., the 99.7th percentile of daily maxima) in the 2060s (Figure 3b).However, 23 of these locations have only a 5% chance of this occurring in the 2040s, two decades earlier (Figure 3g).There is a national median estimate that 12 (5%-95% range: 6-25) exceedances of the 99.7th percentile threshold will occur in at least 1 year in the 2040s.This increases to 44 (24-72) days in the 2060s and 205 (165-249) days in the 2090s.
The variability in possible flood days due to random variations in how tides and surge extremes coincide can be larger than the difference in flood days due to following a higher emission scenario.Under the SSP5-8.5 median SLR scenario, there is a national median estimate that 14 (5%-95% range: 7-27) exceedances of the 99.7th percentile threshold will occur in at least 1 year in the 2040s.This increases to 57 (34-84) days in the 2060s and 253 (215-290) days in the 2090s.The national median estimates for 2040 and 2070 sit within the 5%-95% range for SSP3-7.0provided above and in Figure 3. Median estimates based on standard climatological projections (Section 2.10) are 13, 43, and 212 days for 2040, 2060, and 2090 respectively, under SSP3-7.0.These estimates increase to 13, 56, and 256 days under the higher SSP5-8.5 scenario.This indicates that our projection framework is unbiased with respect to the standard practice of climatological projections (i.e., Section 2.10).However, our projections that consider combinations of tides and surge (i.e., Section 2.6) can provide a possible range of values, not just a single estimate.
The sensitivity of differences in decade-dependent threshold exceedances on how tide and surge extremes coincide highlights the importance of defining risk tolerance when defining chronic flood adaptation triggers.For example, if it is decided that an adaptive action is to be taken when there is 5% chance of some flood frequency occurring, critical decisions will be required to be made decades earlier than if planners were content with waiting until there was a 50% or 95% chance of some annual flood frequency occurring.If this trigger involves anything other than a 50% probability, climatological projection frameworks based only on hazard assessments from a baseline period are unsuitable.These only provide a median estimate based on the observed coincidence of tides and storm surges at a location.
The differences in the degree to which tide and surge extremes' coincidence impact flood rates varies by location, and by the time period being considered.For example, Bunbury has the 11th smallest 5th-95th percentile range for maximum annual flood days during the 2040s but the largest 5th-95th percentile range for the 2090s.However, some locations, such as Mackay, Broome, Port Hedland, Townsville, and Darwin, have the smallest 5th-95th percentile range for all thresholds and decades (e.g., compare Figures 3c and 3i, or Figures 3b and 3h).This variability in projected flood dates is related to tidal range.The latter locations are among those with the largest tidal ranges in the ANCHORS data set (Hague et al., 2022).These large tidal ranges more strongly constrain the future frequency of flooding.This means that the coincidence of storm surges with large spring tides is less important in modulating flood days, as flooding generally occurs only on days with the very highest spring tides anyway (Hague, McGregor, et al., 2023).It is notable that the Australian locations with large tidal ranges and Earth's Future 10.1029/2023EF003993 HAGUE AND TALKE less variability due to how tide and surge extremes coincide are all subject to tropical cyclone driven storm surges at present.Tropical cyclones are very often the focus when considering coastal flood risk in these locations (e.g., Haigh, MacPherson, et al., 2014;O'Grady et al., 2022).The most extreme future floods will continue to be driven by tropical cyclones.However, impacts currently associated with tropical cyclones will likely occur chronically due to tides in future.

Impact of Chronic Flood Definition on Emergence Times
There is substantial variability in emergence times of chronic minor flooding between locations, with some locations facing an imminent chronic flood threat.Based on the minor flood thresholds defined by Hague et al. (2022), Ballina, Brisbane, Gladstone, Gold Coast, Newcastle, Port Kembla, and Sydney have a 95% chance of 30 annual flood days of flood occurring at least once during the 2020 decade.In contrast, the threat is less immediate at Fremantle and Port Adelaide, where there is 95% chance that chronic flooding will not emerge until after 2050, even under high emissions scenarios.Of the 10 locations with well-established flood thresholds based on multiple observations, median estimates of the first year of 50-day-per-year flooding vary from 2020 (or earlier) (Newcastle) to 2074 (Port Adelaide).This variability in coastal flood frequency occurs despite relatively uniform rates of SLR and approximately uniform changes in the heights of sea level extremes across Australia (McInnes, Church, et al., 2015;McInnes, Monselesan, et al., 2015).Tidal variability modulates how frequency of threshold exceedances changes respond to SLR but not how extreme water level heights respond to SLR (Hague, McGregor, et al., 2023).Arguably, improving estimates of spatial variations in SLR is not very important for understanding how coastal flood frequencies will evolve this century.
Variability in emergence times of chronic minor flooding also arises from using different definitions of emergence times.Considering the median of all stations, the first year when 30 flood days is projected (Figure 4c) is 15 years earlier than the first year when 50 flood days is projected (Figure 4a).The first year of 50 minor flood days typically occurs 12 years earlier than the first year of 100 annual minor flood days (Figure 4d).Defining chronic flood emergence years as the first of nine out of 10 consecutive years where at least 50 minor flood days are projected (Figure 4b) pushes back the median estimated emergence years of chronic minor flooding by about 5 years, compared to the first year of 50 flood days.As the emergence year is defined as the first of the 9 years, the observed 5 years difference is explained by year-to-year variability in flood days (which is driven by random variations in how tide and skew surge extremes coincide).Thus, persistent chronic flooding (9 out of 10 years) requires more SLR than a single year of the same threshold (here, 50 days).Future changes in tides and skew surges are not considered here.
The dependence of emergence times on how "chronic" flooding is defined highlights the importance of identifying the most appropriate definitions of chronic flooding and what flood thresholds these pertain to.More broadly, there is a need to define appropriate adaptation triggers so science can inform coastal decision-making.For example, are single or consecutive years of extreme flood counts more likely to trigger adverse effects and/or adaptative actions?Another consideration is the relative importance of frequency and severity of floods in determining these adaptation triggers.The definitions of chronic flooding used here and by previous studies is focused on chronic nuisance or chronic minor flooding, in which flood water depths are relatively small (<20-30 cm).It is possible that increasingly frequent moderate or major flooding require adaptation measures well before minor floods occur 30, 50 or 100 days per year.For example, managed retreat has been implemented in response to recurrent hazards that have occurred less frequently than 30 days per year (Hino et al., 2017).It is important to identify the most locally relevant definition of chronic flooding as part of coastal flood hazard assessments, based on chronic flood adaptation triggers.
The effect of considering different SLR scenarios on emergence times is dependent on both the definition of chronic and the emergence time itself.The effect is generally modest.If chronic minor flooding is expected prior to 2050, considering a different scenario doesn't have much effect.For example, our projections suggest that Melbourne has a 95% chance of seeing 30 days per year of minor flooding by 2049 under SSP3-7.0 and 2047 under SSP5-8.5.If chronic minor flooding is expected to emerge only after 2050, then considering a different SLR scenario has a greater effect.For example, Port Hedland has 95% chance of seeing 30 days of minor flooding by 2114 under SSP3-7.0,or by 2104 under SSP5-8.5.The same principle applies if emergence times are delayed by considering a different definition of chronic flooding.For example, Melbourne has 95% chance of seeing 100 days per year of minor flooding by 2072 under SSP3-7.0 and 2068 under SSP5-8.5.

Impact of Changes in Tidal Range and Skew Surge Magnitudes and Sea-Level Rise Scenarios on Emergence Times
Increases (decreases) in tidal ranges and skew surge magnitudes of 1% per decade lead to larger (smaller) increases in flood frequencies than would be expected based on SLR alone.Increasing tide and storm surge magnitudes bring forward emergence times of chronic flooding (Figure 5), and increase the probability of chronic flooding emerging by a given year.In general, percentage-wise changes in tidal range have a much larger effect than the equivalent percentage-wise change in skew surge.For example, 1% per decade changes in skew surge magnitudes change emergence years of chronic flooding by one or 2 years.Such a low value suggests that even quite large changes in the daily skew surge have a quite insignificant impact on coastal flood frequencies.In contrast, 1% per decade changes in tidal amplitudes change emergence years of chronic flooding by a median of 5 years and by decades at locations with very large tidal ranges.
Results suggest a symmetry in the sensitivity of flood frequency to changes in sea level variability.The impact of 1% decrease in tidal range has the same magnitude but different sign as a 1% increase.Thus, reducing tidal range through engineering interventions or nature-based solutions that return bathymetry and tidal inlets to predevelopment depths could be a viable flood risk mitigation technique in locations where flood frequencies are sensitive to changes in tidal range (e.g., Pareja-Roman et al., 2023).Given that SLR and depth increases can strongly increase tidal range in estuaries where tidal ranges are currently damped (Khojasteh et al., 2023), our findings mean that these locations may be more at risk of future flooding than anticipated based on assumptions of stationary tidal ranges (e.g., Hanslow et al., 2018).
For 22 of the 38 ANCHORS gauges, the emergence time changes due to 1% per decade changes in tidal amplitudes are comparable to those associated with different emission scenarios.At locations with large tidal ranges, a 1% per decade increase in tidal range brings forward emergence times further than following a higher emissions scenario does.To demonstrate this, we consider the median emergence year of 50 annual exceedances of the 99th percentile under SSP3-7.0(Figure 5).These emergence years vary from 2046 at Carnarvon and Newcastle to 2093 at Broome, with a national median of 2055.If a 1% per decade increase in tidal amplitudes is applied, the median emergence time becomes 2053.The same 2 years shift in median emergence time occurs when the SSP5-8.5 SLR scenario was considered instead of SSP3-7.0.The difference is much larger for locations with large tidal ranges.At Darwin under the SSP3-7.0median SLR scenario, there is 50% chance of chronic flooding occurring by 2076.If the higher SSP5-8.5 median SLR scenario is considered the emergence year becomes 2072.An increase in tidal amplitudes of 1% per decade results in an emergence time of 2061-a full decade earlier than if stationary tides are considered.
The impact of changes in skew surge magnitudes on emergence times of chronic flooding is much less.For example, increasing skew surges by 1% per decade does not change the national median emergence time of 50 annual exceedances of the 99th percentile under SSP3-7.0(Figure 5).No locations experienced a change in emergence time greater than 1 year.This includes locations such as Melbourne and Fremantle where exceedances Earth's Future 10.1029/2023EF003993 HAGUE AND TALKE of the 99th percentile currently occur exclusively during large storm surge event at present (Hague et al., 2022).Based on our approach (Section 2.9), a 1% change per decade in tidal amplitude means that all high tide heights (relative to annual mean sea level) increase by 1% per decade and all low tide heights decrease by 1% per decade, relative to the mean tidal level.The same applies to storm surges.It is likely that tidal range and storm surges will alter in a less spatially coherent and symmetric way, meaning these results should only be interpretated a sensitivity test, not locally applicable projected future changes in tidal ranges.
Several flood hazard studies have considered changes other than mean sea level but have primarily focused on non-tidal, rather than tidal, sources of sea-level variability (e.g., McInnes et al., 2013;Muis et al., 2020;Vousdoukas, Bouziotas, et al., 2018;Vousdoukas, Mentaschi, et al., 2018).Furthermore, recent studies have shown that it is local, rather than global or regional, changes in tidal range that tend to be largest; nonetheless, such tidal changes are often driven by factors such as channel deepening that are occurring world-wide (Haigh et al., 2020;Talke & Jay, 2020).Local shifts in tidal properties, even if they occurred decades or a century ago, have produced the greatest impact on flood frequencies (datum exceedances) over the recent past due to sea-level rise (Hague, Grayson, et al., 2023;De Leo et al., 2022;Li et al., 2021;Pareja-Roman et al., 2023).Studies that have considered future changes in tides and their impact on flood hazards have used tidal models which do not generally resolve processes that have led to these local-scale changes, as they require detailed local knowledge (Harker et al., 2019;Schindelegger et al., 2018;Vousdoukas, Mentaschi, et al., 2018).

Impact of Flood Severities on Emergence Times
The key findings of Sections 3.1-3.3apply regardless of the flood threshold being considered: annual flood counts are strongly modulated by whether extreme storm surges coincide with large tides, projections of chronic flood emergence years depend on how chronic flooding is defined, and changes in tides can lead to changes in these emergence years.This is observed in projections of annual flood rates and emergence times of chronic flooding for higher thresholds provided as Supporting Information S1.The risk of frequent major flooding this century is real.For example, under the SSP3-7.0median SLR scenario there is a 50% chance that the level 0.3 m above the present-day 1-in-1-year level (i.e., 99.7th percentile) will be exceeded at least 50 times in a single year in the 2090s at many locations on the Australian east coast (Figure 5d).For Sydney, this level (i.e., 0.3 m above the 99.7th percentile) has been estimated as occurring less than once per millennium in the current climate (Watson, 2022).In other words, the present-day 1-in-1000-year level will be likely be occurring at 50 days per year before 2100 under SSP3-7.0,a 50,000-fold increase in frequency.
The main difference between results for different flood thresholds is that the emergence of chronic flooding occurs earlier for lower flood thresholds and later for higher flood thresholds.If the 99.7th percentile of daily maxima is considered as a flood threshold and the 50-day criterion is used, the emergence years of chronic flooding vary from 2054 (Newcastle) to 2105 (Booby Island), with a median of 2067 (Figure 6b).In contrast, all locations are projected to have seen chronic exceedances of the 97th percentile before 2075, with this occurring in most locations before 2042 (Figure 6a).If a threshold 0.3 m above the 99.7th percentile is considered, the median emergence year increases from 2067 to 2098 (Figure 6d).Relative differences between different locations' emergence years in Figure 5 further demonstrate the threshold-dependence of emergence years.For example, Wyndham has the earliest emergence time for chronic exceedances of the 97th percentile (Figure 6a) but ranks 7th when considering the level 0.3 m above the 99.7th percentile (Figure 6d).This means that the locations most at risk of chronic minor flooding may not be the locations most at risk for chronic moderate or major flooding.For example, based on the 97th percentile Wyndham would be assessed as most at-risk, which is a different conclusion that would be drawn if the level 0.3 m above the 99.7th percentile is used.The inconsistent emergence of chronic flooding of different thresholds means that past or recent experience may not be a good guide to where the focus for adaptation should lie.
Our findings highlight that the accuracy of flood thresholds should be a key consideration when developing flood hazard studies and assessments.For example, comparing Figures 6c and 6d shows that a 0.2 m difference in flood threshold typically causes a two -decade difference in chronic flood emergence times.The national median emergence tide of 50 days exceeding a threshold of 0.1 m above the 99.7th percentile is 2079.If the threshold is 0.3 m above the 99.7th percentile, the emergence year is 2098 instead.This observation holds for different SLR scenarios and differences in tidal trends.Under SSP5-8.5,The national median emergence tide of 50 days exceeding a threshold 0.1 m above the 99.7th percentile is 2075.This is delayed until 2092 if a 0.2 m higher Earth's Future 10.1029/2023EF003993 HAGUE AND TALKE threshold is considered.Similarly, SSP3-7.0 with a 1% per decade increase in tidal amplitudes has a national median emergence year of 2075 for a threshold of 0.1 m above the 99.7th percentile.For 0.3 m above the 99.7th percentile, the median emergence time is 2095.
This threshold sensitivity means that a 0.2 m difference, uncertainty or error in flood thresholds results in projections of flood frequencies and emergence times that can be offset by two decades.Thus, locations with a relatively small, 0.2 m difference in elevation may require different planning and adaptation horizons.But, another challenge for adaptation planning is that coastal flood threshold uncertainties are typically at least 0.2 m in most coastal flood studies.For example, the frequently used 'common impact threshold' approach to estimate flood thresholds (Dusek et al., 2022;Hague, McGregor, et al., 2023;Thompson et al., 2021) has errors of 0.19 m, 0.25, and 0.39 for minor, moderate and major flood thresholds, respectively (Sweet et al., 2018).Digital elevation models (DEMs) are also used to relate water levels to their terrestrial footprints.The vertical root mean square errors of global DEMs are of order meters (Kulp & Strauss, 2018, 2021), or tens of centimeters for regional DEMs (Hanslow et al., 2018).In Australia, vertical bias of CoastalDEM is 0.11 m (Kulp & Strauss, 2018).Further work Earth's Future 10.1029/2023EF003993 HAGUE AND TALKE is required to understand if an uncertainty of several decades is tolerable for identifying and predicting the onset of triggers that may elicit an adaptive action, such as managed retreat (Buchanan et al., 2019;Haasnoot et al., 2013Haasnoot et al., , 2021;;Stephens et al., 2018).

Limitations of Methods and Results
Our projection framework also shows potential for use in understanding how other sources of non-stationarity, other than mean sea-level rise and changes in tidal range and skew surge magnitudes, may impact flood hazards.In this context, our framework is similar to the storylines approach adopted in many climate change studies (Shepherd et al., 2018).This approach develops physically plausible, self-consistent future pathways for decisionmakers by prioritizing different impact outcomes over different climate model outputs.We have demonstrated this for changes in tidal range and skew surges, but several other avenues could be explored in future work.Firstly, we have not considered the changes in magnitudes of AMSL anomalies through time (Widlansky et al., 2020).Secondly, we have discounted the possibility of local mean sea level changes differing from CMIPbased SLR scenarios due to local vertical land motion and subsidence (Karegar et al., 2017;Tay et al., 2022).Thirdly, we have not considered compound floods and possible changes to river flow, local wind forcing, and other factors that influence water level anomalies in estuaries (e.g., Moftakhari, Salvadori, et al., 2017;Wahl et al., 2015).Our work could be extended to assess the sensitivity of flood frequencies to such changes, using a percentage per decade approach like the one explored here.
Future work could also consider whether preserving autocorrelation in skew surges would improve the method.Skew surges exhibit statistically significant autocorrelation at multiple time lags (Figure S3 in Supporting Information S1), but annual mean sea level anomalies do not (Figure S4 in Supporting Information S1).Autocorrelation of tides is retained when using the approach described in Section 2.3.Similar to the convolution of distributions used in joint probability methods (Baranes et al., 2020;Batstone et al., 2013;Pugh & Vassie, 1978), the random sampling of skew surges used here does not preserve autocorrelation properties in the timeseries.Future improvements could include using an extreme value distribution for skew surge and applying an extremal index that accounts for correlation (e.g., Baranes et al., 2020;Batstone et al., 2013).Additionally, a block sampling approach may improve on the daily sampling used here.It is likely that different block lengths would need to be used for different locations due to large differences in the autocorrelation functions at each location (Figure S3 in Supporting Information S1).This has the potential to lead to inconsistencies when inter-site comparisons are made.Hence, detailed further work is required to evaluate the merits of such an approach, and is beyond the scope of this study.

Concluding Remarks and Opportunities for Future Work
We have presented a new way to estimate the coastal flood implications of local sea-level rise.Our approach advances upon existing coastal flood projection frameworks by considering how the coincidence of independent tidal, storm surge and AMSL extremes lead to year-to-year variability in flood frequencies.Our framework can also be used to understand the impact of future changes in sea level variability (e.g., tidal trends), not just the mean rise, on coastal flood frequencies.We find that increases in tidal range increase flood rates and bring forward emergence times, with decreases having a commensurate but opposite effect.In many cases, non-stationary tides have an equivalent effect to lowering the impact threshold or using a higher sea-level rise scenario.A similar but much reduced effect is seen for changes in skew surge magnitudes.
More generally, we present a compelling argument for improving understanding of the sensitivity to flood hazard metrics to sources of uncertainty beyond the common emission scenarios and their SLR responses.Different realizations of these sources of uncertainty can have commensurate impacts on flood frequencies as different SLR scenarios can.Applying our projection framework allows for first-pass assessments of the relative sensitivities of coastal flood frequencies to different sources of uncertainty in future changes in mean sea level and sea level variability at a tide gauge location.An example of where improvement is needed is in relating water levels to their flood impacts using digital elevation models or common impact threshold approximations.These methods can have uncertainty of at least 0.2 m, which corresponds to 20 years in chronic flood emergence times at most Australian locations.Appropriate and centimeter-accurate impact-based flood thresholds may be urgently needed to correctly identify the water levels associated with impacts of local concern.
Earth's Future 10.1029/2023EF003993We highlight the importance of considering all sources of future changes in coastal hazards, not just emission scenario and the possible climate system responses to these.Estimates of SLR-induced tidal changes exist for many locations including Australia (Harker et al., 2019), but have not been included in projections of coastal flood frequencies.This adds another source of uncertainty not typically quantified in flood projections: the unquantified uncertainty inherent in assuming that sea level variability will not change from its present-day values.Where practical, we suggest that future projections of coastal flood heights and frequencies should dynamically model possible tidal changes, both due to alterations in the oceanic response to sea-level rise (e.g., Pickering et al., 2017;Schindelegger et al., 2018), and due to coastal management and bathymetric changes in estuaries (e.g., Holleman & Stacey, 2014;Lee et al., 2017;Talke & Jay, 2020).Tides and other long-waves such as storm surge are influenced by river flow (e.g., Kukulka & Jay, 2003;Moftakhari et al., 2013), by channel dredging and inlet modifications (e.g., Hague, Grayson, et al., 2023;Pareja-Roman et al., 2023), and by other processes that alter the roughness and geometric dimensions of estuaries (Talke et al., 2021).Hence projection frameworks that include both local and far-field responses of tides to changing conditions are needed to better project possible future flooding and the variance of possible pathways.
Our results highlight the need to consider how different responses to one element of risk can exacerbate or mitigate risks posed by another (Simpson et al., 2021).For example, channel dredging can reduce the flood threat posed by heavy rainfall and river runoff but potentially increase chronic flooding on high tides through increased tidal ranges (de Leo et al., 2022;Talke et al., 2021).Similarly, this work could also be leveraged to help make decisions around the efficacy of implementing possible flood risk mitigation measures.For example, the results of Table 1 imply that if a seawall currently protects against a 99.7th percentile water level is raised by 0.2 m, the onset of chronic flooding may be delayed by 20 years.However, this may only reduce flooding from direct overtopping, but not from backflow up storm water drains, a common pathway of nuisance flooding (e.g., Hague, Grayson, et al., 2023;Habel et al., 2020).
The use of an Australian case study increases the local significance of our results-ours are the first national estimates of changes in flood frequency for Australia.Prior to this study, sea level allowances (changes in height of present-day extremes) were the only metric for which projections were available to inform coastal decision-making.The successful implementation of the allowance approach is predicated on being able to keep pace with sea level rise through engineering interventions such that flood frequencies remain unchanged (McInnes, Church et al., 2015).Even if practical and economic, the relative benefits and risks of such engineering actions depend on location and the rate of SLR (Ghanbari et al., 2020(Ghanbari et al., , 2021;;Hague, Grayson, et al., 2023;Stephens et al., 2021).In addition, the influence of interventions on local tidal range must also be considered to limit the chance of maladaptation occurring (Nunn et al., 2021;Schipper, 2020).Our approach does not require such assumptions, allowing for the consideration of an alternative viewpoint where coastal managers may choose for certain assets to fully depreciate and be subsequently abandoned (Haasnoot et al., 2021).
Finally, our study highlights that further work is required to properly incorporate the impacts of coinciding tide and surge extremes and changes in tidal ranges into flood hazard assessment.This includes national projections of changes in tidal range at tide gauge locations, developing centimeter-accurate flood thresholds or elevations models, and identifying the adaptive triggers that are most relevant to coastal decisionmakers.To support this, changes in heights of present-day extreme water levels under changes in tidal range and skew surge magnitudes should be similarly assessed as changes in their frequency have been here.Our results highlight the urgency of this future work being undertaken.Even without considering changes in tidal range, present-day annual extremes could occur chronically as early as the 2040s in several Australian locations.Hence, planning should already be underway for what adaptive actions will be required to protect, accommodate, or retreat from this emerging chronic flooding hazard.

Figure 2 .
Figure 2. Locations of Australian tide gauges in the ANCHORS data set.Reproduced with author permission from Hague et al. (2021) under CC-BY 4.0 license.

Figure 3 .
Figure 3. Projections of maximum annual exceedances of the present-day average annual maximum water level (99.7th percentile of daily maxima) for three different decades, under SSP3-7.0: the 2040s (left panels), 2060s (center vertical panels) and 2090s (right panels).The 5th (upper panels), median (center horizontal panels) and 95th (lower panels) percentile counts are provided for each decade.

Figure 4 .
Figure 4. Impact of chronic flood definition on chronic minor flood emergence time.The median estimate of emergence year of chronic flooding based on four different definitions of emergence (a-d) based on SSP3-7.0 median SLR scenario from our projections.

Figure 5 .
Figure 5. Change in emergence year resulting from a 1% per decade increases in tidal amplitudes (a, upper left) or skew surge magnitudes (b, upper right) and 1% per decade decreases in tidal amplitudes (a, lower left) or skew surge magnitudes (b, lower right).Brown dots denote emergence times being delayed relative to the case where no changes in tidal amplitudes or skew surge magnitudes were applied.Blue dots denote emergence times occurring earlier compared to the case where no changes in tidal amplitudes or skew surge magnitudes were applied.

Figure 6 .
Figure 6.Impact of chronic flood threshold on the emergence year of 50 annual flood days, based exceedances of that threshold using our projection method.The median based estimate of emergence year of chronic flooding for four different definitions of emergence (a-d) based on SSP3-7.0 median SLR scenario.Locations where chronic flooding does not occur by 2150 are not indicated.

Table 1
Projected Emergence Years of 30 Days Per Year With Exceedances of Two Different Thresholds: The Historical Period 99.7th Percentile (Left Side of Table) and the Water Level 0.2 m Above the Historical Period 99.7th Percentile (Right Side of Table)