Storm-induced saltwater intrusion responds divergently to sea level rise in a complicated estuary

Global warming and sea level rise (SLR) not only increase the intensity and frequency of coastal hazards but also complicate associated dynamics. The exacerbated saltwater intrusion in this context will further be adversely affected by storms with deepening distances and growing duration, aside from the simultaneous coastal flooding they cause. Here, we investigate storm-induced saltwater intrusion and its responses to SLR in the Pearl River Estuary by numerical simulation. Predominant in competition with river runoffs, typhoons passing by cause fast stratification and dramatic increase of saltwater intrusion lengths via wind mixing. Stronger destratification and longer recovery time are linked to a narrow long channel, where the tidal excursion is weak owing to bay/channel-shape modulation. The rising sea levels enhance the tidal prism and shift the saline water universally to the upper reaches, and this impact tends to be amplified in the upper part of the bays owing to the narrowing bay shape and shoaling bathymetry. The saltwater intrusion length could be expressed as a linear relationship with the water level, but with divergent responses to storms, depending on bay/channel shapes. Amplification of saline intrusion is indicated in the channel-shaped estuary, but the farthest distance during a storm is less sensitive to SLR than in a bell-shaped estuary. The present study reveals the potential importance of storm-induced compound hazards to coastal communities, and highlights the notably specific salinity responses whereby tributary morphology.


Introduction
Estuaries are one of the most vulnerable systems with low-lying terrain (Harley et al 2006, Tessler et al 2015), their varying environment is jointly modulated by terrestrial and oceanic processes and has wide-ranging impacts on human society and ecosystems.Flooding and saltwater intrusion in estuaries have always been vital concerns associated with sea level rise (SLR) (Nicholls andCazenave 2010, Bhuiyan andDutta 2012), which causes higher extreme water levels and migrates saline water upstream (Chua and Xu 2014).Particularly exacerbated by increasing demands for freshwater in coastal zones (Werner et al 2013), the saltwater intrusion exerts crucial socioeconomic and ecological impacts by worsening water quality, penetrating underwater and shrinking estuarine habitats (Hilton et al 2008, Werner and Simmons 2009, Nicholls and Cazenave 2010).As another driver from the sea-side, storm surge induced by tropical cyclones (TCs) is intensely studied for its significant contribution to coastal flooding (e.g.Sheng et al 2010, Sebastian et al 2014, Bilskie et al 2016, Bacopoulos et al 2017, Santiago-Collazo et al 2019, Gori et al 2020), but the perturbation it brings to salinity does not appeal to much attention since high flushing capacity in TC-active seasons and the day-scale nature of storm wave generally only cause a short-term issue of water supply.
Nevertheless, it is worth noting that apart from the rising sea levels, climate change and anthropogenic activities are also reported to alter geophysical factors, e.g.TC activities, land subsidence, and groundwater supply (Elsner et al 2008, Marfai and King 2008, Wang et al 2012, Tessler et al 2015, Yamaguchi et al 2020, Zhang et al 2020), which can exacerbate the frequency and severity of saltwater intrusion coinciding with TC events.Take the Pearl River Estuary (PRE) for example, TCs generated in the western North Pacific (WNP) tend to occur more frequently in the dry season (from October to next March, figure S1).This increases the joint probability of late typhoons accompanied by low river discharges.Despite Typhoon Nesat (2022), which happened in the middle of October, turned to the west approximately 390 Km away from the PRE, but still caused severe saltwater intrusion in the Modaomen Waterway (hereafter MW) with concurrent low discharge (Zhongshan Water Affairs Bureau 2022).Not only that, but anthropogenic factors associated with water resources management (e.g.pumping, landuse change) might affect the severity of compound hazards in terms of water safety and water environment (Nicholls andCazenave 2010, Werner et al 2013) in the flood season.For instance, to positively cope with more frequent extreme floods and droughts (e.g. the severest drought in the Pearl River Basin in 2021 since 1961, the Catastrophic flood in the North River in 2022 since 1915) along with changing climate, emergency operation methods of a reservoir in upper reaches are highly correlated with water supply and flood risk.To specify, releasing reservoir storage to prevent watershed floods from heavy rainfall might not provide enough runoff to repulse saline water induced by a sequential typhoon event.Previous research suggested that the simultaneous or consecutive occurrence of moderate-intensity events may lead to a compound event and enhance the overall hazard (Moftakhari et al 2017, Ward et al 2018, Zscheischler et al 2018).Thus, it is a significant and urgent issue to include storm-induced saltwater intrusion into drivers for assessing compound hazards in the PRE.
Many studies focus on spatial and temporal variations of saltwater intrusion by discharge and tidal mixing.The increase in saltwater intrusion is linked with a decrease in discharge and an increase in tidal amplitude (Lv and Du 2006).The relationship between saltwater intrusion and river discharge follows a power law with an exponent of n, which depends on estuaries (Bowen and Geyer 2003, Banas et al 2004, Lerczak et al 2009).Stronger salt intrusion occurs during neap tides in a well-mixed or salt wedge estuary (Turrell et al 1996, Brockway et al 2006, Xue et al 2009, Gong and Shen 2011), while stronger salt intrusion happens during spring tides in partially mixed estuaries (Bowen and Geyer 2003, Banas et al 2004, MacCready 2007, Lerczak et al 2009).The effects of bathymetric change (e.g.channel deepening, land reclamation), winds and waves on saltwater intrusion have also been intensively examined (Wu et al 2016, Xie and Li 2018, Gong et al 2018a,  2018b, Lange and Burchard 2019).With the concern about global warming, the effects of SLR on salinity have been evaluated in several estuaries.Bhuiyan and Dutta (2012) suggested a climatic effect of 1.5 PSU per meter SLR in the Gorai river network, Bangladesh.According to simulation results of Rice et al (2012), an SLR of 1 m would largely threaten the tidal freshwater wetlands along the James River, and make the drinking-water intake in the Chickahominy River exceed local standard.Featuring multiple tributaries, the PRE presents increases in salinity, stratification and tidal range responding to SLR, though flow condition and tidal phase would alter the specific increase rate of salinity (Yuan et al 2015, Hong et al 2020).As described earlier, there is a knowledge gap as to how storm plays a role in saltwater intrusion in an estuary and how SLR exerts further impacts.Therefore, the present study aims to understand the saltwater intrusion mechanisms associated with storms in changing climate, which can complete knowledge corresponding to dynamic drivers with different time scales (slow varying SLR and short-term storm surge) and provide scientific information supporting future mitigation and adaptation measures.

Study area and affecting TCs
The Pearl River Delta (figure 1) is one of the biggest deltas in the world, the rapid development of the Guangdong-Hong Kong-Macau Greater Bay Area made it the most densely populated and dynamic economic coastal system.Mainly dominated by the interplay of river runoffs and tides, the estuary features a funnel shape and forms one of the most complicated river networks worldwide (Yuan et al 2015).The estuary comprises three tributaries, the West, North, and East Rivers, which discharge freshwater to the sea via eight outlets from south to north: Yamen, Hutiaomen, Jitimen, Modaomen, Hengmen, Hongqimen, Jiaomen, and Humen (figure 1(b)).Among them, MW transports the largest portion of freshwater (approximately 26.6%) and supplies water for domestic and industrial usage for many cities such as Zhuhai and Macau.Due to distinct seasonal variability of runoff, the annual river discharge in the dry season (4000 m 3 s −1 , Yuan et al 2015) is one-fifth of that in the flood season, and makes this estuary particularly susceptible to intruding saltwater.
From the CMABST (CMA-STI best track dataset for TCs over the WNP) dataset (Ying et al 2014(Ying et al ) between 1949(Ying et al and 2021, 604 , 604 TCs were determined by filter criteria of TCs stronger than tropical depressions have ever entered into a neighboring circle of 800 Kmradius (HKO 2018, SMG 2018) centered with MW (figure 1(b)), as candidates affecting the PRE coasts.Among them, 293 TCs are categorized as the primary track type that moved toward the shoreline heading in a west-northwest (WNW) direction and made landfalls (figure 1(a)).Since the shoreline of the PRE faces southeastward, this WNW-track type threatens it with high frequency (approximately 4 per year) and potential destructiveness of strong surges and saltwater intrusion (all TC types displayed in figure S2).Typhoon Mangkhut ( 2018) is one of the latest and strongest storms that hit the PRE with wide-ranging surges and flooding.To address possible differences in saltwater transport due to TC landfall locations, Typhoon Hagupit (2008) was chosen as another typical case but with comparative intensity and size.They both made landfalls at the southern coasts of the PRE (figure 1(a)), putting the majority of the estuary at high risk of surge and saltwater intrusion by falling into the dangerous right-front quadrant of TC tracks.

Hydrodynamic model and configurations
The flexible, primitive equation, hydrostatic model SCHISM (Semi-implicit Cross-scale Hydroscience Integrated System Model) (Zhang et al 2016), which is implemented in hybrid finite-element/finite-volume algorithm and semi-implicit time-stepping scheme, can efficiently and robustly resolve creek-to-ocean flows with water elevations, velocities, and mass concentrations along with them on unstructured grids.It was adopted in the present study to simulate surge processes and saltwater transport influenced by storm weather.
The computation domain (figure S3) comprises the tidal river network and nearshore region of the South China Sea.The river runoffs are prescribed at Shizui, Gaoyao, Shijiao, Laoyagang, and Boluo as input.The Etopo1 Global Relief Model dataset with a resolution of one arc-minute was employed in mapping the bathymetry outside the continental shelves, while the nautical charts from the Navigation Guarantee Department of Chinese Navy Head Quarters and cruise data were utilized for nearshore regions and river channels.The vertical discretization uses a 25-layer hybrid σ-z scheme that can fully resolve the surface of the water body.
The flux transfer via sea surface was prepared from the European Centre for Medium-Range Weather Forecasts Reanalysis 5 (ERA5) dataset (Hersbach et al 2020), including downwelling shortwave/longwave radiation flux, downwards precipitation flux, near-surface air temperature at a height of 2 m, wind at a height of 10 m and mean sea level pressure, except the specific humidity at a height of 2 m (not directly archived in ERA5) was obtained from the Japanese 55 year Reanalysis (JRA-55) dataset.As high winds around cyclone eyes are usually underestimated in ERA5 reanalysis (De Dominicis et al 2020, Yang et al 2021), the hourly frames of wind fields associated with the TC period were merged with the ones generated from the parametric Holland vortex model (Holland 1980) and validated against observations.Note that the Holland model is efficient though errors might be larger than a physicsbased complete wind profile model (Wang et al 2022).The spatiotemporal-varying salinity and temperature profiles along the open sea lateral boundaries were extracted from the global data assimilative HYCOM (Hybrid Coordinate Ocean Model, Halliwell et al 1998, 2000, Bleck 2002) simulation products, which provides accurate 3D temperature, salinity and current structure in 0.08 • × 0.08 • latitude-longitude grid every 3 h.

SLR scenarios
Uncertainty about future sea level change might arise from reliability, accuracy, and other aspects in terms of data, prediction methods, and models.Considering greenhouse gas emission scenarios in future climate, sea level trends were predicted from climate model results of the Coupled Model Intercomparison Project Phase 5 (Meehl et al 2000(Meehl et al , 2005) ) and adjusted by glacial isostatic adjustment (Argus et al 2014, Peltier et al 2015), corrections owing to ocean thermal expansion and land ice mass loss were also made according to IPCC's Fifth Assessment Report (IPCC AR5 2014).A uniform SLR of 0.4 m and 0.9 m by 2050 and 2100 under the RCP8.5 scenario (Yang and Chen 2022) are adopted in the present study.

Quantitative measures 2.4.1. Squared buoyancy frequency N 2
The Brunt Väisälä Frequency N (Knauss 1996), also as buoyancy frequency, represents the intrinsic frequency of internal waves, its square is commonly employed to measure stratification strength and is expressed as: where g is gravitational acceleration, ρ i is the density at depth of z i , and ∂ρ ∂z is the denisty gradient at depth z i .A higher value of N 2 implies higher static stability and a more stratified water column (Li et al 2020).

Saltwater intrusion length
To quantify the saltwater intrusion induced by a storm, saltwater intrusion length is defined as the distance from the estuary entrance to the landward limit of the bottom 0.5 PSU isohaline comparatively along the pre-selected longitudinal sections (Gong et al 2018a).

Water age
The water age (Deleersnijder et al 2001, Delhez andDeleersnijder 2002) of freshwater is defined as the time elapsed since its entry into the simulation domain, and it can assess the mixing of freshwatersaltwater.The water age can be solved similarly as for salinity in the equations governing a pair of state variables, tracer concentration C and age concentration α, respectively: where h is the water depth, u is the velocity vector, and κ denotes the eddy diffusivity tensor.Following a = α/C, the mean water age a (hereafter as water age) can be derived.The initial condition and the inlet boundary condition were set to zero for both the tracer and age concentrations, except the tracer concentration at the inlet boundary has a continuous unit value.

Results and discussions
The hindcasted simulation results influenced by Typhoons Hagupit and Mangkhut were first conducted and validated against measured water levels separately (figures S4 and S5).Due to the availability of salinity data, the capability of our model for modeling salinity and velocity variations were checked by simulation during other periods (figure S6).The model in the present study reproduces hydrodynamics and mass transports with satisfaction.

Saltwater intrusion patterns from a planar view of the estuary
Due to the facts that the intensites and the wind radii of 15.5 m s −1 for these two typhoons are comparative (both are severe typhoons with a maximum wind speed of 41.5-50 m s −1 and 15.5 m s −1 wind radius of 400 Km) near landfalling hours, the PRE is prone to higher surges caused by the closer typhoons.The shortest distance of Typhoon Hagupit (2008) to MW is ∼132 Km when approaching the shoreline, which is 75 Km farther than that of Typhoon Mangkhut (2018).According to the hindcasts of these two typhoons (figure 2), high storm surges occur in the whole estuary, especially with a magnitude surpassing 3 m in heads of all bays during Typhoon Mangkhut.
In the case of Typhoon Hagupit, the most affected area shifts to the west of the estuary where the maximum surge height also reaches 3 m.The snapshots of winds in figures 2(a) and (b) are chosen correspondence to instants when the saline water transports the farthest distance upstream, the anti-clockwise wind stresses superimposed with storm translation is prone to surge generation and strengthened water mixing.
The instantaneous distributions of surface/bottom salinity in figures 2(c)-(f) demonstrate the farthest locations where saltwater intrusion takes place.Due to the freshwater plume in summer, stratification forms indicated by remarkable surface/bottom isohaline differences (25 h averaged isohalines are obtained 3 d before the storm's landfall, indicating influences by moderate tides; figure S7).However, stratification is destroyed during the typhoon's passage, the intruding saltwater expands rapidly to the water column from the bottom due to wind mixing and vertical shear, and achieves a fully mixing state, the freshwater-saltwater interface transports upstream therewith (figures S8 and S9).Being eastward-facing estuaries, storms coming from the WNP and those landing to the west side of the PRE potentially lead to up-estuary winds and destratify entrainment, whose magnitudes depend on specific landfall locations.Despite the differences in storm surges for the two typhoons (maximum surges in figures 2(a) and (b) and instantaneous surge distribution in supplementary figure S10), the storm-induced saltwater intrusion shows less discrepancy in horizontal pattern rather than the surge itself.During the state transition of salinity distribution, the surface isohalines experience broader variations in horizontal positions.Owing to differences in the surge distribution, the saltwater intrusion patterns differ slightly in dominant pathways of the Lingding Bay (hereafter LDB): compared with the case that the main pathway is in the West Channel during Hagupit (see figure 1

Behavior differences due to bay morphology
The significant difference in bay/channel shapes is one of the key reasons that contribute to the distinctive behavior of saltwater transport toward upstream.As displayed in figure 1(b), MW in long narrow shape suffers from the severest seawater intrusion, especially in the dry season, thus section S1 is selected for analysis purposes.Besides, as a bell-shaped bay, LDB has the highest tidal prism capacity in the PRE, we choose another section S2 along its West Channel, which is the primary channel transporting water.Both MW and LDB are partially mixed estuaries.To understand the physical processes in MW and LDB caused by a storm event, here we look into the temporal variations of vertical profiles at A1 and A3 (see figure 1(b) for locations) during Typhoon Mangkhut (2018).Figure 3 displays the results of the squared buoyancy frequency N 2 , salinity, and water age.When the peak of the surge wave arrives, N 2 drops sharply to very small values below 0.001, impling strong mixing and this instability lasts for about two days.Correspondingly, the stratification dynamically balanced by river discharge and tides is devastated by mixing due to storms, and the recovery time can be several times of the surge wave period.Interestingly, during the recovery period, relatively larger N 2 appears from modulation of the intrinsic frequency of the water body.The intruding water volume of high salinity is confined and results in a moderate value of salinity in MW whereby narrow mouth width and the existence of entrance sand bar, but this narrow shape also leads to a slower recovery back to stratification (figure 3(c)).Comparatively, larger magnitudes of salinity and age maintain in lower part of LDB and are strengthened by Typhoon Mangkhut to 30 PSU and 39 d, respectively.The differences of salinity concentration and flushing capacity in MW and LDB is tightly linked with open sea forcing strength that competes with river runoffs, deciding by their interplay with channel/bay shapes.Generally, weaker tidal energy penetrates MW and stabilizes the water column there, indicating by higher N 2 than that in LDB (figures 3(a) and (b)); but destratification by storms leads to stronger pre-/poststorm variation of N 2 in MW.

Impacts of rising sea levels
To exclude the possible impacts of discharge variation on freshwater/seawater mixing, the responses of saltwater intrusion to SLR are simulated and investigated replacing with an annual mean discharge during flood season (from April to September; e.g. 10 745 m 3 s −1 and 2,030 m 3 s −1 at Gaoyao and Shijiao based on the daily records between 1961 and 2019, and the baseline scenarios are the cases without sea level changes. In figures 4(a)-(d) we discuss the results in the case of Typhoon Mangkhut (similar results associated with Typhoon Hagupit are displayed in figure S11).Due to the strong mixing induced by typhoons, the intruded saline water in the sea surface and bottom reaches equivalent lengths, but differs in terms of longitudinal variation ranges among different estuaries (figures 4(e)-(h)).Wider variation ranges are observed along the section S2 in LDB, which is correlated with a larger tidal prism.Similar as results in figure 2, the 25 h averaged isohalines represent influences by tides.The rising sea levels further enlarge the tidal prism and enhance the saline water transport for both surface and bottom waters.The narrowing bay shape and shoaling bathymetry are likely to amplify the impacts of SLR, in particular in upper part of bays.This amplication phenomenon is also observed when a saltwater intrusion happens due to storms, that tongue-shaped isohalines in both surface and bottom waters form and move farther because of fully vertical mixing.Figure 4 also implies that the West Channel in LDB is the main pathway for  b)).However, on the other aspect, the peak values mitigate under these SLR scenarios, implying that the severity of storm-induced saltwater intrusion in this kind of narrow channel weakens its sensitivity to sea level changes.Comparatively, the extra intrusion lengths in LDB (along section S2) are shorter, ranging between from 20 to 25 Km (figures 5(c) and (d)).Meanwhile, the impulse of saltwater intrusion length is migrated to upstream by similar magnitude of distance, no matter whether the driving forces are tides or typhoons, indicating that the saltwater intrusion lengths influenced by tides and typhoons response to SLR similarly in LDB.Furthermore, the slightly increased amplitude due to tides by rising sea levels can be explained with amplified tidal range in the upper region of this estuary (see supplementary figure S12), similarly as reported by Hong et al (2020).
Further interpretation of behavior of saltwater intrusion length is investigated by correlating it with water levels (reflecting amplitude referring to mean sea level) in figures 5(e)-(h).A simple linear relationship can be expressed during tides and periods collectively impacted by typhoons.When the estuary is solely influenced by tides, the similar slopes (0.131-0.163) indicate that a same water level change does not show a significant difference in increment of saltwater intrusion length (figures 5(e) and (f)).This proportional relationship also holds in LDB when typhoons pass by (figures 5(g) and (h)), however, remarkable amplification (about 1/2 − 2/3 of the slopes) of saltwater intrusion length is found in MW (figures 5(e) and (f)).When sea levels continue to rise, the variation of intercepts demonstrates a direct response of intruding strength to SLR, and 1 mrise in sea level raises 16 Km of saltwater intrusion length in LDB, which is several times of that (about 5 Km) in MW.At the same time, tiny increases in slopes due to SLR imply that the saltwater tends to intrude shorter distances with rising sea levels in both LDB and MW.Being very small this indirect response compensates for the variation of saltwater intrusion length owing to SLR.Quite similar patterns can be observed if we correlate the saltwater intrusion length with total water depths (figure S13), but without splitting SLR's impacts.

Summary
This paper gives first emphasis on typhoon-induced saltwater intrusion, which is the other key aspect of hazards rather than coastal flooding it causes, in one of the largest delta prone to compound coastal hazard in the context of climate change.Simulations under different SLR scenarios were conducted in a numerical framework implemented for the South China Sea region.
As a typical partially mixed estuary, the salinity field in the PRE represents a brackish water spreading seaward owing to river plume and the saline water intrudes landward from the bottom due to sea-side drivers.The intrusion pattern depends on the competition between fluvial and coastal driving forces.The intrusion responses to storms differ from other dynamic factors (e.g.tides) by a dramatic increase in saltwater intrusion length.The typhoons landing to the west of the estuary cause up-estuary winds and erase stratification by wind mixing and vertical shear.Not only that closer typhoons with similar intensities cause high surges but also lead to larger saltwater intrusion lengths.However, the specific salinity responses differ according to relative strengths between river runoffs and open sea forces that are modulated by bay/channel shapes, resulting in a discrepancy of salinity concentration, flushing properties and recovery capacity of the water column.The narrow shape of MW undergoes stronger state transition of stratification and slower recovery time than the bell-shaped LDB.As a key pathway of saline water by tidal excursion, the West Channel in LDB presents tongue-shaped isohalines during typhoons' passage.
The rising sea levels enhance the tidal prism and shift the saline water universally to the upper reaches.Under influences of both tides and storms, this impact tends to be amplified by SLR in the upper part of bays owing to narrowing bay shape and shoaling bathymetry.Quantification using the saltwater intrusion length shows a decreased sensitivity of peak values to SLR in a long narrow channel.By fitting the correlation between the saltwater intrusion length and water level at the river mouth, the linear relationship describes a significant amplification of intruding magnitude by typhoons than by tides in the channelshaped estuary.Induced by rising sea levels, slightly increasing slopes of the linear relationship imply an inherent inhibition of saltwater intrusion, this impact is small and cannot change the universally enhanced saltwater intrusion lengths stimulated by the direct impact of SLR.These specific relationships can be used to assess potential intrusion lengths in these estuaries from water level data including tidal and surge components.

Figure 1 .
Figure 1.(a) Historical TC tracks affecting the PRE and trajectories of Typhoons Hagupit and Mangkhut (data source: CMABST dataset during 1949-2021) and (b) zoomed bathymetry of the PRE.TCs in solid and dashed curves indicate that move along the WNW direction and other track types.O1-O8 represent eight outlets; S1-S2 display the selected analysis sections in red curves and their entrance positions (A1-A2) in purple dots, and A3 is an extra analysis site along S2. (Abbreviation: HMB-the Huangmao Bay; MW-the Modaomen Waterway; LDB-the Lingding Bay).

Figure 2 .
Figure 2. Distribution of maximum storm surges (a), (b) and corresponding surface/bottom isolines before and during storms' landfall for (c)-(f) salinity and (g)-(j) water age.The dashed lines indicate 25 h averaged isolines for pre-storm instants, and the colored contours present the cases at 23:00 on 23 September 2008, and 11:00 on 16 September 2018, respectively.The instantaneous winds (white arrows) in a-b are corresponding to the above instants for both typhoons.
(b) for channel locations), the East Channel also plays a key role in moving saline water toward the upper bay during Mangkhut, which attributes to a closer TC landfall location.Figures 2(G)-(j) shows the corresponding results of age isolines, which represent the flushing property of fresh water.The maximum age difference between the surface and bottom ranges from 20 to 35 d outside the PRE, but decreased to around zero during typhoon periods (figures S9(e)-(h)), which results in an average flushing time of 20 and 45 d (e.g. in Mangkhut's case) at river mouths of MW and LDB, respectively.

Figure 3 .
Figure 3. Temporal variations of vertical profiles for (a), (b) N 2 , (c), (d) salinity, and (e), (f) water age at analysis sites A1 (left panel) and A3 (right panel) in the case of the Typhoon Mangkhut.

Figure 4 .
Figure 4. Comparison of surface/bottom isohalines for Typhoon Mangkhut (2018)'s case under different SLR scenarios (a)-(d) and time series of surface/bottom salinity along sections S1 and S2 for the baseline (present-day) case (e)-(h).Plots in (a), (b) and (c), (d) represent results affected by tidal forcing and that combined with the passage of Mangkhut, respectively; and the former is calculated by a 25 h average during moderate tides.

Figure 5 .
Figure 5.Time series of saltwater intrusion length (a)-(d) and its relationship with water level at estuary mouth (e)-(h) based on results during Typhoon Hagupit (2008) and Typhoon Mangkhut (2018) under different SLR scenarios.Solid dots represent data associated with periods during typhoon passage and circles represent other data; thick and thin lines demonstrate linear regression fits of above two situations affected by typhoons and tides, respectively.S1-S2 represent two sections in figure 1(b).
Zhang G, Murakami H, Knutson T R, Mizuta R and Yoshida K 2020 Tropical cyclone motion in a changing climate Sci.Adv. 6 eaaz7610 Zhang Y J, Ye F, Stanev E V and Grashorn S 2016 Seamless cross-scale modeling with SCHISM Ocean Model 102 64-81 Zhongshan Water Affairs Bureau 2022 Report on Situation of Salt Water Intrusion in Zhongshan City on October 25 2022 (available at: http://water.zs.gov.cn/gkmlpt/content/2/2176/post_2176123.html#2775)Zscheischler J et al 2018 Future climate risk from compound events Nat.Clim.Change 8 469-77