A
Investigating Historical Storm Surge Occurrences in the Philippines from Tide Gauge Observations
Anjela
A.
Ilagan
1,3✉
Emailanjela.ilagan@pagasa.dost.gov.ph
Olivia
C.
Cabrera
2
Marcelino
Q.
Villafuerte
1
II
1
1
Department of Science and Technology – Philippine Atmospheric, Geophysical and Astronomical Services Administration (DOST-PAGASA)
1100
Quezon City
Philippines
2
Institute of Environmental Science and Meteorology
University of the Philippines Diliman
1101
Quezon City
Philippines
3
DOST-PAGASA, Science Garden Complex
Sen. Miriam Defensor-Santiago Ave
1100
Central, Quezon City
Philippines
Anjela A. Ilagan1, Olivia C. Cabrera2, and Marcelino Q. Villafuerte II1
1Department of Science and Technology – Philippine Atmospheric, Geophysical and Astronomical Services Administration (DOST-PAGASA), Quezon City 1100 Philippines
2Institute of Environmental Science and Meteorology, University of the Philippines Diliman, Quezon City 1101 Philippines
Correspondence to:
Anjela A. Ilagan
DOST-PAGASA, Science Garden Complex, Sen. Miriam Defensor-Santiago Ave., Central, Quezon City 1100 Philippines
Email: anjela.ilagan@pagasa.dost.gov.ph
Acknowledgements
We gratefully acknowledge the National Mapping and Resource Information Authority’s Hydrography Branch for providing the tide gauge data used in this study. We extend our thanks to Mr. Mark Anthony Valencia of the Oceanographic Data Management Section for his valuable guidance in the interpretation and processing of the data. We also thank the Department of Science and Technology – Science Education Institute for the scholarship and graduate fellowship granted to Ms. Ilagan.
Abstract
Tropical cyclones (TCs) frequently affect the Philippines, making its coastal areas highly vulnerable to TC-induced hazards such as storm surges. As an archipelagic country with an extensive and complex coastline, local variations in coastal characteristics can influence surge magnitude and extent. Despite this, most of the previous studies have focused on individual extreme events or model-based simulations of specific TCs, with few directed toward systematically compiling storm surge histories from instrumental records across the entire country. This study addresses this gap by investigating historical storm surge events (SSE) in the Philippines using tide gauge observations from 1947 to 2024. A variable annual threshold was applied to the residual sea level derived from tide gauge data, identifying 133 SSEs across multiple coastal locations. The identified SSE exceeding 1 m above mean sea level, occurred in areas characterized by shallow and wide continental shelves that amplify surge response, which included the stations located along the east, north, and west coasts of the country. In contrast, SSEs were largely absent in the southern part of the country (Mindanao), likely due to its proximity to the equator where the weak Coriolis force limits TC intensification. Five selected TC case studies of high storm tides further illustrated that surge magnitude depends not only on TC proximity but also on coastal geometry and bathymetry. The resulting database enhances understanding of storm surge variability across the Philippine coasts and provides a foundation for improving localized SSE forecasting, hazard mapping, and risk assessment frameworks.
Keywords
Storm Surge ⸱ Storm Tide ⸱ Historical events ⸱ Tide gauge ⸱ UTide
1. Introduction
Storm surge is the abnormal rise in sea level generated by tropical cyclones (TC) over and above the predicted astronomical tide (Pore and Barrientos 1976). It is mainly a coastal hazard that can cause inland flooding of up to 30 feet or more and can last from a few minutes to a few days (McKeever 2020; WMO 2011). With almost 20 TCs, on average, entering the Philippine Area of Responsibility (PAR) annually and 9 making landfall (Cinco et al. 2016), the Philippines is highly susceptible to storm surges. This susceptibility is further amplified by the country's extensive coastline, which spans 37,008 km (PEMSEA 2021), and the presence of Low Elevation Coastal Zones (LECZs), making up approximately 4.5% of its land area.
Arafiles et al. (1984, as cited in Needham et al. 2015) estimated that the Philippines experiences an average of 4 to 6 storm surge events (SSE) annually. Remarkably, the country holds the record of the four highest storm surge heights (SSH) documented in the western North Pacific between 1880 and 2013 (Needham et al. 2015). Among these, Super Typhoon (STY) Haiyan in November 2013 produced the most devastating SSE in more recent times, particularly along the coasts of San Pedro Bay on the northwestern margin of Leyte Gulf. STY Haiyan caused over 7,000 fatalities, destroyed an estimated 1.1 million homes, and displaced approximately 4 million people (Sherwood et al. 2015). Most of the destruction and casualties were attributed to SSH exceeding 5 m (Soria et al. 2016). This catastrophic event became a turning point in raising national awareness of storm surge hazards in the Philippines (Esteban et al. 2016), leading to increased research interest and the integration of storm surge risk into disaster mitigation strategies during TC events.
Globally and regionally, tide gauge observations have been widely used to detect SSE. Because tide gauges sample water levels directly at the coastline, they have a higher temporal resolution than satellite records (Hamlington and Thompson 2022) and are considered the most reliable data sources for observing storm surges (Ma and Li 2023). In such studies, astronomical tides are first removed from the observed sea level (OSL) to isolate the non-tidal residual sea level (RSL), which is then interpreted as the SSH associated with the passage of TCs (e.g., Morin et al. 2016; Kim et al. 2017). For example, Morin et al. (2016) analyzed historical tide gauge records at the Manila South Harbor station from 1960 to 2012 and identified SSE as occurrences of a positive RSL of at least 0.25 meters when a TC was located within 800 km of Metro Manila. Their study identified 75 SSE over the 53-year period. Similarly, Towey et al. (2022) used data from 12 tide gauges along the eastern coast of the United States to detect SSE from 1946 to 2019. Unlike Morin et al. (2016), they applied a 365-day running mean in addition to predicted tides to remove low-frequency trends from the OSL. They also used a more conservative TC proximity threshold of 500 km, which they considered the maximum range at which a TC could realistically generate a storm surge.
Despite the growing relevance of storm surge impacts, previous studies have largely focused on individual extreme events or model-based simulations of notable TC events, and relatively few are directed toward compiling storm surge histories from OSL data. In the Philippines, systematic documentation of historical events based on long-term observations remains limited. One earlier attempt was made by Project NOAH (2014), a nationwide disaster risk and reduction management program, wherein 57 SSE were identified in the Philippines between 1589 and 2013 based primarily on archived newspaper clippings and journals. However, such database has not been updated since its initial release. While anecdotal sources offer valuable insight, especially for earlier times, they are often limited by spatial and temporal uncertainty. Validating modeled SSH against anecdotal accounts also remains challenging when inconsistencies exist between sources. Moreover, eye witnessed SSH by people living in the affected area represent storm tide, which is the combined effect of storm surge and astronomical tide (NOAA n.d.), hence requires careful consideration when used for model validation. Thus, developing a comprehensive and consistent storm surge database based on instrumented records is crucial for improving our understanding of surge-related risks in various parts of the Philippines.
To address the earlier mentioned gaps, the present study investigates storm surges in the Philippines using tide gauge observations. By identifying and characterizing SSE across multiple locations and covering a multi-decade period, this work contributes to long-term coastal hazard assessment, enhance risk communication, and support evidence-based planning that strengthens disaster resilience nationwide.
2. Data and Methods
2.1 Datasets
A
stations have records of observations starting only in the early 2000s (Fig. 1b). Hourly tide measurements were recorded in Local Time (LT) relative to the zero of tide staff (0TS) and expressed in centimeters (cm). Since 0TS is an arbitrary reference point specific to each station (Flores et al. 2022), the OSL datasets were adjusted to their corresponding local mean sea level (MSL) to ensure consistency. A data quality assessment was conducted to identify blank or invalid entries (e.g., '999'), correct misalignment and typographical errors, and detect any offsets or phase shifts that may produce tidal-like patterns in the RSL (Towey et al. 2022; Ma and Li 2023).Figure 1b shows the data completeness for each tide station. Five stations, namely Manila, Cebu, Legazpi, Davao, and Tacloban, have over 73 years of sea level observations. Most of the remaining stations have maintained data for 11 to 20 years. Stations with near-continuous records were generally established in more recent years, with Guimaras being the most complete but covers only two years of data up to 2024. Interestingly, the four longest-operating stations namely, Manila (94%), Legazpi (91%), Cebu (86%), and Davao (85%), all exhibited data completeness above 80%. In contrast, Tacloban, despite having records dating back to 1951, shows a significantly lower completeness of just 32%. Several other stations also exhibited notable data gaps ranging from a few months to several years, like San Fernando, Real, and Tagbilaran. Consultation with NAMRIA indicated that these gaps may be due to power issues, transmission errors, equipment malfunction or sensor interference, physical damage to the station, or interruptions caused by natural calamities.
The International Best Track Archive for Climate Stewardship (IBTrACS) Version 4r01 was used to get the TC data for the western North Pacific basin. Following Towey et al. (2022), which noted that TCs beyond 500 km from a location generally have limited impacts, a convex hull boundary (referred to as the TC boundary) was constructed to enclose all 500-km radius circles centered at each tide station. Only those TCs that existed or entered over this imaginary boundary (delineated by the orange dashed-line in Fig. 2) were included in the analyses. Additionally, only those TCs that reached at least tropical storm (TS) intensity, defined by maximum sustained winds of ≥ 34 knots, were included in the analyses. From 1947 to 2024, a total of 637 TCs met these criteria. To align with the hourly resolution of the OSL data, cubic spline interpolation was used on the TC track and intensity. The timestamps were also converted from Coordinated Universal Time (UTC) to LT (UTC + 8) to align with the time zone used in the OSL records.
To investigate the prevailing atmospheric conditions during the SSEs, we used reanalysis from ERA5 hourly data on single levels converted to LT. Specifically, the 10-m u and v wind components at the nearest ocean grid point to each station during the TC event were extracted, along with the mean sea level pressure (MSLP) corresponding to the time of the recorded maximum storm surge.
Fig. 2
TC boundary region denoted by the orange dashed lines derived using the 500-km radius circles (blue) centered at each tide station
2.2 Methods
An SSE identification algorithm developed by Ma and Li (2023) for the coasts of China demonstrated that annual sea level thresholds based on the Pauta Criterion have the best comprehensive detection ability for storm surges. The method identifies outliers by checking if the difference between the RSL and the mean RSL exceeds the standard deviation by three times. Following this approach, we adopt a similar methodology to identify SSE in this study.
2.2.1 Derivation of tides, RSL, and filtered RSL
Harmonic analysis is a widely used method for tide prediction that can be performed using computer-based tools (Bosboom and Stive 2021). In this study, the Python package UTide (version 0.3.0) by Bowman (2022), a re-implementation of the Matlab-based UTide by Codiga (2011), was used to perform harmonic analysis and generate hourly astronomical tides at each station. Specifically, the OSL was used as input to the solve function to calculate 68 tidal coefficients. These coefficients were supplied to the reconstruct function to generate the corresponding tidal heights for each station. The RSL was then computed by subtracting the predicted tidal heights from the OSL.
To further minimize the influence of short-period oscillations unrelated to TCs, which can mistakenly be flagged as storm surges, a Gaussian low-pass filter was used. This filter allows low-frequency signals to pass while suppressing high-frequency oscillations, such as those caused by tides (Smith 2023; De Oliveira et al. 2009). In this study, a 6-hour Gaussian low-pass filter was applied to the hourly RSL series. This was chosen because it better preserves significant sea level peaks compared to a 12-hour filter used by Ma and Li (2023). However, since it tends to attenuate surge amplitudes with durations shorter than 12 hours, the actual SSH was quantified using the unfiltered RSL values (Ma and Li 2023).
2.2.2 Identification of SSE
In this study, an SSE is considered to have occurred during a TC event when both the RSL and the filtered RSL exceed an annual sea level threshold calculated as the sum of mean RSL and three times the standard deviation of RSL expressed as:
Threshold = Mean RSL + 3σ (1)
For each station, sea level data were aligned with the time window when a TC was present within the defined TC boundary region. If both the RSL and the filtered RSL rose above the annual threshold during that time, the event was classified as a storm surge.
3. Results
We present the results by first giving an overview of the identified SSEs, focusing on their spatial distribution across tide stations, temporal occurrence by year and month, and the annual frequency at each location. In this study, we refer to the OSL as the storm tide and the unfiltered RSL as the SSH. This is followed by a characterization of the TCs that generated the storm surges, and finally, an examination of five selected SSEs that occurred in different coastal regions of the country.
3.1 Frequency and spatio-temporal distribution of SSEs
A total of 133 SSEs were identified across 43 tide gauge stations from 1977 to 2024, based on exceedances of both the RSL and the filtered RSL above annual thresholds during TC events. Although the study period begins in 1947, no SSEs were detected from 1947 to 1976. Manila recorded the highest number of SSEs, with 59 occurrences over 78 years of data. This was followed by Port Irene, with 17 events spanning 38 years, and Currimao with 16 events over 14 years. Meanwhile, 17 stations have no recorded SSEs (Fig. 3).
Fig. 3
Number of detected SSEs per station and their equivalent storm surge rate
However, because tide stations have varying record lengths, absolute counts do not necessarily reflect how frequently a storm surge occurs at a given location. To account for this, it is more representative to look at the annual frequency or surge rate per station. For instance, Masinloc recorded 5 SSEs with only 2.5 years of data, resulting in the highest annual frequency rate of approximately 2 SSEs per year. This is followed by Currimao with approximately one SSE per year. All the remaining stations, including Manila which has the greatest absolute count, experience less than one SSE per year (Fig. 3). The 17 stations with no SSEs, are mostly located in the southernmost major island of the country (Mindanao) and over Western Visayas (Fig. 4a).
The highest SSH across all stations and TC events was recorded in Pasacao during TY Molave in 2020, reaching 211 cm whereas the highest storm tide was recorded in Virac during TY Goni at 239 cm. Notably, the impact of tides on the final sea level reaching the coasts is evident in Fig. 4. Most stations experience storm surges of no more than 50 cm especially on the western side of the Philippines (Fig. 4a). However, they can be amplified by as much as 100 cm when coinciding with high tides (Fig. 4b).
Fig. 4
a Highest storm surge height and b highest storm tide (storm surge + tide), shown by colored markers at each station, with marker size indicating the surge rate
Fig. 5
Distribution of SSEs by year (x-axis) and by month (y-axis). Totals per year and month are displayed at the top row and rightmost column, respectively
The seasonal distribution of SSEs shows a peak in October (25 events), followed by November (21) and July (20), with no occurrences in March (Fig. 5). Interannually, SSEs averaged nearly three per year, though none were recorded from 1947–1976, 1979, 1998, and 2002. Most years noted 1–3 SSEs, while a maximum of 6 SSEs were observed in 2022 and 2024. By decade, most SSEs occurred in the 2010s (35).
3.2 TCs associated with SSEs
To provide an overview of the results across all stations, Table 1 summarizes the highest storm surge and highest storm tide recorded at each station, the TCs associated with these events, and the distance of the TC from the station at the time of the occurrence.
As expected, due to the influence of tides, the TC that generated the highest storm surge did not always correspond to the event with the highest storm tide. This is evident in several stations, including Balanacan, Cagayan De Oro, Coron, and Currimao. In terms of TC distance, storm surges were still detected even when the TC is already more than 500 km away from the station. For instance, Camiguin and Gensan recorded SSH of up to 29 cm and 67 cm, respectively, despite the TC being over 1000 km away. Similarly, Brookes Point experienced an SSH of 135.97 cm when TY Vamco in November 2020 was situated 715 km from the station.
Figure 6 depicts the tracks of TCs that caused an SSE in at least one of the tide stations. Based on the current TC categories operationally being used by the Philippine Atmospheric, Geophysical and Astronomical Services Administration (PAGASA), majority of these TCs were classified as Typhoons (TY; 57), followed by Super Typhoons (STY; 31), Severe Tropical Storms (STS; 30), and Tropical Storms (TS; 15). Of the 133 TCs, 120 originated in the Pacific Ocean and 13 in the South China Sea. Among them, TY Molave (2020; local name Quinta) produced the highest storm surge, STY Goni (2020; local name Rolly) the highest storm tide, while STS Trami (2024; local name Kristine) affected the largest number of stations, with SSEs recorded at 18 locations. These notable TCs are discussed further in the following subsection.
Fig. 6
TC tracks that caused an SSE in at least one of the stations, which originated from the a Pacific Ocean and b South China Sea. The TC category is based on the maximum wind speed attained by the TC during its entire lifetime and classified based on PAGASA
3.3 Selected SSEs from Different Coastal Regions
The impacts of storm surges can be exacerbated when they occur coincidentally with high tides. This section presents five selected SSEs with significant storm tides observed across different coastal regions in the Philippines. The five SSEs were identified by selecting the top five TCs with the highest storm tide (HST) and determining the specific stations where these values occurred. In cases where two TCs produced their HST at the same station, the TC with the lower HST was assigned the station with the next HST. This process was repeated until five unique stations were identified, each representing one SSE.
3.3.1 TY Molave (2020)
Typhoon Molave, locally known in the Philippines as Quinta, developed as a TD east of Mindanao on October 23, 2020 (NDRRMC 2020a). It moved westwards across Southern Luzon and intensified to STS on October 25 generating less than 25 cm SSH at Catbalogan. As the TC strengthened to typhoon and progressed westward the following day, more stations, mainly in the Bicol region, have recorded storm surges. Water levels at Pasacao were significantly higher by as much as 100 cm than in Balanacan, Bulan, Catbalogan, El Nido, Mamburao, and Masbate.
Fig. 7
a Wind vectors and MSLP (shadings) at the time of the maximum storm surge during TY Molave. b Satellite image of the Pasacao coast from Google Earth. c Times series of wind and MSLP (top) and sea level (bottom) during the TC event. Gray-shaded area in c denotes the time when the TC is traversing inside the TC boundary (orange dashed lines in a). Blue arrows denote the wind during storm surges while the red arrow highlights the wind associated with the highest storm surge and storm tide
On October 26 at 1:00 (LT), a peak SSH of 211 cm, the highest among all SSEs identified in this study, with corresponding storm tide of 232 cm were recorded at Pasacao station (Fig. 7c). A snapshot of the large-scale atmospheric conditions during the highest SSH is shown in Fig. 7a. It can be noticed that the center of TY Molave (the area with the lowest MSLP) was approximately west of the station. To further scrutinize what happened during this SSE, the hourly values of MSLP and winds from the nearest grid point of ERA5 to the station (marked in Fig. 7b) were extracted. The time series of extracted atmospheric and sea level data at Pasacao reveal that the peak SSH that occurred in Pasacao happened three hours after the lowest MSLP in the area was registered with prevailing southeasterly winds and when the center of the typhoon was 116 km away from the station. This condition persisted for a few hours generating storm tides of more than 200 cm being recorded in Pasacao until the data became unavailable on October 26, 8:00 (LT) onwards (Fig. 7c). Data from other stations including Balanacan, Bulan, Catbalogan and Masbate remained continuous with less than 40 cm surges and up to 63 cm storm tides.
3.3.2 STY Goni (2020)
Super typhoon Goni, locally known as Rolly, entered PAR on October 29, 2020 and followed a track similar to TY Molave. After developing at the Pacific Ocean, it moved westward across the southern portion of Luzon generating less than 30 cm SSH at Jose Panganiban, Legazpi, and Real stations. It made its first landfall on November 1, 2020 in Bato, Catanduanes at 5:00 (LT) (NDRRMC 2020a). An hour after, at 6:00 (LT), the highest SSH for this TC was recorded within the same province at Virac station where the SSH reached 171 cm, the second highest among all SSEs in this study. Combined with high tide, this corresponded to a storm tide of 239 cm (Fig. 8c). A snapshot of the large-scale atmospheric conditions during this event is shown in Fig. 8a. Similar to TY Molave, the center of STY Goni during its highest SSH was approximately west of Virac station.
To further investigate the conditions that persisted during this event, the hourly MSLP and winds from the nearest grid point of ERA5 to the station (marked in Fig. 8b) were extracted. The time series of extracted atmospheric and sea level data at Virac reveal that the peak SSH that occurred in this station happened one hour after the lowest MSLP was registered and when the center of the TC was 17 km away. Virac’s concave coastline, shown in Fig. 8b, combined with atmospheric winds directed toward it during the peak surge (Fig. 8a, c) have contributed to the buildup of the water. The elevated water levels in this area persisted throughout the morning when the prevailing winds were southerly and MSLP were less than 1005 hPa (Fig. 8c). Meanwhile, other stations that recorded storm surges were Balanacan, Bulan, Jose Panganiban, Legazpi, Masbate, and Real ranging from 17–64 cm and storm tide of at most 126 cm. Compared with TY Molave, STY Goni produced storm surges of almost similar magnitude across these stations. However, no SSE was detected at Pasacao station, which had the highest SSH during TY Molave.
Fig. 8
Same as Fig. 7, but for STY Goni and at Virac station
3.3.3 STY Lynn (1987)
Super typhoon Lynn, locally known as Pepang, developed as a TD over the western North Pacific on October 16, 1987 and followed a west-northwestward course towards the South China Sea (Royal Observatory Hong Kong 1988). Tide station coverage during this period was limited, with only eight operational stations. Storm surges were recorded at Manila and Port Irene stations between the 23rd and 24th of October in 1987. At Manila, SSH ranged from 16 to 30 cm and maximum storm tide reached 85 cm. Higher SSH were observed farther north at Port Irene where SSH ranged 23–114 cm with a maximum storm tide of 173 cm. Both the highest storm tide and storm surge during STY Lynn were recorded at Port Irene on October 23 at 18:00 and 20:00 (LT), respectively. The large-scale atmospheric conditions during its highest SSH in Port Irene are shown in Fig. 9a. It can be observed that the center of STY Lynn was approximately east of the station.
The hourly MSLP and wind data from the nearest grid point of ERA5 to the station (marked in Fig. 9b) were similarly extracted. The time series of atmospheric and sea level data at Port Irene (Fig. 9c) show that the peak SSH of 114 cm happened one hour before the lowest MSLP was registered in the area, under prevailing northwesterly winds, and when the TC’s center was approximately 55 km from the station. Elevated water levels persisted for several hours, generating storm tides of nearly 100 cm. Almost three days after the last SSH observation on October 24, 7:00 (LT), tide gauge data from the Port Irene station became unavailable.
Fig. 9
Same as Fig. 7, but for STY Lynn and at Port Irene station
3.3.4 TY Nesat (2011)
Typhoon Nesat, locally known as Pedring, entered PAR from east of Southern Luzon on September 24, 2011 (Fig. 10a). It followed a west-northwestward track and interacted with the prevailing Southwest Monsoon. By September 27 at 1:00 (LT), it started producing less than 25 cm SSH initially detected at Masbate station and subsequently at Manila, Baler, Subic, and Caticlan stations. On the same day, Manila station recorded the highest SSH of 75 cm with a corresponding storm tide of 162 cm. This peak occurred an hour after the station recorded the highest storm tide of 165 cm (Fig. 10c). Figure 10a shows the large-scale atmospheric conditions during this event. It can be noticed that the center of TY Nesat was approximately northwest of Manila station around that time.
To further assess the event, the hourly MSLP and winds from the nearest grid point of ERA5 to the station (marked in Fig. 10b) were similarly extracted. The time series of this data plotted in Fig. 10c shows southwesterly winds and MSLP less than 1001 hPa conditions were present during the surge occurrences. The highest SSH happened five hours after the lowest MSLP in the area was recorded and when the center of the typhoon was 214 km away from Manila station. Meanwhile, the other stations have experienced smaller storm surges up to 31 cm, with maximum storm tide reaching 95 cm. As the TC moved farther away from the Philippines on September 28, storm surges were still detected at Subic station reaching 26 cm and a total water level of 44 cm.
Fig. 10
Same as Fig. 7, but for TY Nesat, and at Manila station
3.3.5 STS Trami (2024)
Severe Tropical Storm Trami, locally known as Kristine, entered PAR on October 21, 2024 while approaching the Philippines from the Philippines Sea and later tracked west-northwestward toward central Philippines. Being the most recent among the five SSEs investigated, tide gauge coverage was extensive, with storm surges detected in 18 stations from San Fernando in the north to Cagayan De Oro in the south. By October 22, as it intensified into a TS, SSH were detected in the Eastern Visayas region and at Real station in the eastern coast of Luzon, with SSH reaching 42 cm and storm tides of up to 127 cm. As the TC strengthened to an STS by October 23, more stations across Luzon and Visayas recorded storm surges. At Baybay station, the highest storm tide of 148 cm occurred at 2:00 (LT). Meanwhile, the highest SSH of 59 cm was recorded 17 hours after this at 19:00 (LT). A snapshot of the large-scale atmospheric conditions during this time is shown in Fig. 11a. During its highest SSH, it can be noticed that the center of STS Trami is approximately northwest of Baybay station
Fig. 11
Same as Fig. 7, but for STS Trami, and at Baybay station
The hourly values of MSLP and winds from the nearest ERA5 grid point to Baybay station (marked in Fig. 11b) were extracted. The time series of the extracted data (Fig. 11c) indicated that peak SSH happened one day after the lowest MSLP was registered in the area, with prevailing southwesterly winds, and when the TC’s center was 708 km away from Baybay station. As Trami moved northwestward and exited PAR on October 25, storm tides up to 151 cm were observed along the western Philippines, although SSH were relatively low at maximum 34 cm. Over the course of the TC, the highest storm surge of 59 cm was recorded at Baybay station, associated with southwesterly winds of around 10 m/s. However, since this peak surge coincided with low tide, the resulting total water level reached only 51 cm (Fig. 11c).
4. Discussion
From the 40 SSEs documented in Project NOAH’s compilation of storm surge occurrences in the Philippines that coincided with our study period, only 16 were classified as SSEs in this analysis. Among the SSEs our method was not able to identify, the most notable is the surge generated by STY Haiyan in the Eastern Visayas region, particularly in Tacloban, which remains as one of the most extensively studied surge events in the country. Situated deeper within Leyte Gulf, Tacloban was expected to experience amplified surge heights because of the funneling effect of the gulf’s topography. As reported by Soria et al. (2016), the Tacloban tide gauge recorded an initial drawdown in sea level but failed to capture the surge peak due to instrument damage at the height of the typhoon. In contrast, the Guiuan station, located near the mouth of the gulf, successfully captured surge signals exceeding 100 cm, although still smaller than the maximum levels reported in post-event assessments. The San Jose, Samar station also recorded surge signals but at significantly lower magnitudes than Guiuan (Fig. 12). These differences underscore the critical role of tide gauge location in storm surge monitoring and the possibility of instruments to fail under extreme conditions during TC events.
Beyond Haiyan, the absence of the remaining 24 SSEs from Project NOAH in our analysis may be attributed to several factors. As exemplified by Pasacao and Tacloban stations during TY Molave and STY Haiyan, respectively, instrument malfunctions or physical damage during strong typhoons may have resulted in incomplete or missing data. In addition, the limited spatial coverage of the tide stations, especially before the 2000s when tide gauges are few, may have prevented the detection of surges in areas without instruments. The SSE identification algorithm used here could have also excluded events that did not meet the derived thresholds. For instance, Morin et al. (2016) were able to identify 75 SSEs at Manila Station from 1960 to 2012 using a constant 25 cm threshold for the RSL. Meanwhile, the thresholds we used vary per year depending on the calculated RSL. The annual thresholds derived for Manila ranged from 9 to 76 cm resulting to only 45 SSEs detected for the same period.
Fig. 12
Track of STY Haiyan (left, blue line) and sea level data at Tacloban, Guiuan, and San Jose Samar stations (right).
Upon further examination, the unusually high 76 cm threshold derived for 2002 in Manila was due to a phase offset in the tide data from November to December, which inflated the RSL values. Since the exact correction needed to realign the OSL with the tidal phase could not be determined, the data for that year were retained as is. However, to ensure the reliability of surge detection, any SSEs identified during this period were excluded from the analysis. The next highest threshold of 50 cm occurred in 2000, which was a La Niña year. Sea levels in the South China Sea, toward which the Manila Station faces, are typically elevated during such events (Liu et al. 2020, as cited in Sharoni et al. 2024).
While a fixed threshold may capture more events, it risks including minor water level fluctuations not attributed to TCs. Conversely, a variable threshold provides a more adaptive approach and reflects interannual variability at each location, though it may undercount SSEs that fall below higher annual thresholds. Despite this limitation, the method employed in this study was able to identify 69 SSEs within the temporal coverage of Project NOAH that were not documented in its database. The highest among these was STY Lynn (1987) with 114 cm SSH, followed by STS Nina (1978) with 74 cm, and TY Ruby (1988) with 73 cm (see Supplementary Information). Notably, both STS Nina and TY Ruby were also cited in Morin et al. (2016) as among the highest SSEs in Manila.
When the entire SSE database compiled in this study is considered, greater than 1 m surges were predominantly recorded in the provinces of Catanduanes, Camarines Sur, Eastern Samar, Cagayan, Batanes, and southern Palawan. These elevated surges can be attributed to the presence of shallow and wider continental shelves that allow water to climb up easily as opposed to steeper bathymetry. Lapidez et. al (2015) similarly emphasized that Camarines Sur, Samar, and Palawan’s coastal morphology, particularly their shallow bays, made them vulnerable to high storm surges. In contrast, surges were absent in the provinces of Guimaras and Negros Occidental where the presence of barrier islands reduces their direct exposure to TCs despite being near historical TC tracks.
Storm surges were also largely absent in Mindanao, despite the passage of TCs within a few hundred kilometers. This may be associated with the region’s proximity to the equator, where Coriolis force is weak and TCs rarely intensify, resulting in less robust winds necessary for significant storm surge development (Gray 1968; Chan 2005). In addition, the deeper offshore bathymetry surrounding most parts of Mindanao limits water piling toward the coast, unlike the broader and shallower continental shelves observed along the eastern seaboard of Luzon and Visayas, where higher surges were recorded.
Interestingly, storm surges of more than 1 m were still detected even when a TC was already more than 500 km away from the affected location. This finding emphasizes that storm surges are not solely a function of the TC’s proximity to a location, but also of other TC characteristics such as its size, intensity, and forward speed. Previous studies have shown that larger and more intense storms tend to produce higher surges due to stronger and more persistent wind forcing (Irish et al. 2008; Needham and Keim 2012; Mori et al. 2014). Meanwhile, faster-moving storms can also enhance surge heights by increasing wind stress and reducing the time for water to recede (Rego and Li 2009; Lin and Chavas 2012).
Although this study did not delve deeper into the mechanisms behind the monthly or seasonal distribution of SSEs, it was notable that the monthly frequency of SSEs corresponds with the pattern observed by Cinco et al. (2016). Their study of historical TCs in the Philippines from 1971 to 2013 found that October recorded the highest number of typhoons that entered the PAR, and subsequently the highest number of landfalling TCs in the country. This naturally increases the probability of more SSEs happening during this month. Conversely, the period from December to April has considerably fewer TCs compared to other months, and therefore, a lower chance for SSEs to happen.
The occurrence of SSEs averaged around three per year, reaching a maximum of six in recent years. Considering that tide gauge coverage only became extensive after 2010 and that several temporal gaps exist in instrumental records, it can be inferred that these findings are consistent with the estimate of Arafiles et al. (1984 as cited in Needham et al. 2015), who reported approximately four to six SSEs annually in the country.
The five TC cases further illustrate the surge responses across the Philippine coasts. Among them, TY Molave produced the highest SSH of 211 cm above MSL (4.8 m relative to 0TS) at Pasacao, Camarines Sur. This area is characterized by shallow coastal shelf and is exposed to southeasterly winds during the TC’s passage. Similarly, STY Goni generated 171 cm surge (3.4 m relative to the 0TS) at Virac, Catanduanes where the concave coastline and southerly wind enhanced water accumulation. By contrast, TY Nesat and STS Trami produced smaller surges (≤ 75 cm) at Manila and Baybay, respectively. These lower values are likely due to their weaker intensity, faster translation speeds, and steeper coastal profiles that limit surge amplification. In northern Luzon, STY Lynn caused significant surges reaching 173 cm (4.2 m relative to 0TS) at Port Irene, consistent with the influence of open-ocean exposure and pressure gradients noted in other western North Pacific studies (Needham and Keim 2012; Mori et al. 2014).
STY Lynn was the only case where the peak SSH occurred while the TC center was east of the station. Port Irene’s coastline, which faces the open waters of the Luzon Strait, further enhances early surge development because exposure to long fetches and deep shelves allows wind-driven setup to build even at large storm-center distances (Mori et al. 2014; Dietrich et al. 2011). Combined with STY Lynn’s broad circulation, these factors enabled strong northwesterly onshore winds and significant water accumulation well before the TC moved west of the station, unlike the other SSE cases where peak surges required the TC center to shift west or northwest to align with the respective coastal orientations.
5. Conclusion
This study examined historical tide gauge data across 60 tide stations in the Philippines to identify storm surge occurrences using a variable annual threshold specific to a location. This method, which accounts for interannual variability at each location, identified 133 SSEs from 1977 to 2024 affecting 43 tide stations. The results revealed that the highest SSH of 211 cm captured by the instrument happened at Pasacao Station. This corresponded to a total water level of 4.97 m relative to 0TS. Other SSH exceeding 1 m occurred in the eastern coasts of the country (provinces of Catanduanes, Eastern Samar), as well as the northern coasts (Cagayan, and Batanes provinces), and Palawan (west coast) which are characterized by shallow and wide continental shelves that favor surge amplification. In contrast, storm surges were largely absent in areas located in the southern part of the country (Mindanao), likely due to its proximity to the equator where the Coriolis force is weak that limits TC intensification and associated wind forcing.
Notably, storm surges as high as 1.4 m were still detected even when the TCs were more than 500 km from a station. This suggests that other TC characteristics or prevailing atmospheric condition during the TC event might have contributed to the elevated water levels observed. The analysis further emphasized the combined effects of coastal geometry, bathymetry, and tide gauge placement on the detection and magnitude of storm surges. Gauges situated in sheltered harbors may record smaller surges due to reduced exposure, whereas those in open coasts can record extreme events better but with higher risk of damage. However, due to instrument limitations and the limited tide gauge coverage in much earlier years, the results of this study may not capture all historical SSEs in the country. This underscores the need to integrate instrumental records with eyewitness accounts for a more complete archive of historical events.
Five TC cases with the highest storm tides at different coastal regions were examined for a closer look on the dynamics of storm surges in a localized setting. Collectively, the five cases emphasized that storm surge magnitude is not solely determined by TC intensity or proximity but also by coastal geometry and local bathymetry. Stations situated along embayed or concave coastlines (e.g., Pasacao and Virac) were most susceptible to surge amplification, while sites shielded by barrier islands (e.g., Guimaras and Negros Occcidental) experienced limited impacts despite exposure to TC tracks. The SSE during STY Lynn further illustrated the role of coastal orientation in storm surge buildup. As the only north-facing site, Port Irene was exposed to early onshore wind forcing that enabled rapid surge buildup even while the TC was still approaching the station. The impact of storm surges to coastal communities can also be worsened by the timing of the surges with respect to tidal phase. Simultaneous occurrence of peak surge and high tide, as seen in Virac during Goni, can substantially elevate total water levels and exacerbate flooding risk.
Overall, this study has documented a comprehensive analysis of historical storm surges in the Philippines based from instrumental records covering the entire country. The results contribute to the understanding of storm surge dynamics in the Philippine context and advances the understanding of surge behavior in diverse coastal settings. These insights support the development of localized storm surge forecasting and risk assessment frameworks to strengthen nationwide disaster resilience especially for the vulnerable coastal communities in the country.
A
A
Electronic Supplementary Material
Below is the link to the electronic supplementary material
References
Bosboom J, Stive MJF (2021), October 11 3.9: Tidal analysis and prediction. Geosciences LibreTexts. https://geo.libretexts.org/Bookshelves/Oceanography/Coastal_Dynamics_(Bosboom_and_Stive)/03%3A_Ocean_waves/3.09%3A_Tidal_analysis_and_prediction. Accessed 10 April 2025
Bowman W (2022), April 27 UTide: Python distribution of the MatLab package UTide. PyPI. https://pypi.org/project/UTide/. Accessed 10 April 2025
Chan JCL (2005) The physics of tropical cyclone motion. Annu Rev Fluid Mech 37:99–128. https://doi.org/10.1146/annurev.fluid.37.061903.175702
Cinco TA, Guzman RG, Ortiz AMD, Delfino RJP, Lasco RD, Hilario FD, Juanillo EL, Barba R, Ares ED (2016) Observed trends and impacts of tropical cyclones in the Philippines. Int J Climatol 36(14):4638–4650. https://doi.org/10.1002/joc.4659
Codiga DL (2011) Unified tidal analysis and prediction using the UTide Matlab functions. https://doi.org/10.13140/RG.2.1.3761.2008
De Oliveira MMF, Ebecken NFF, De Oliveira JLF, De Azevedo Santos I (2009) Neural network model to predict a storm surge. J Appl Meteorol Climatol 48(1):143–155. https://doi.org/10.1175/2008JAMC1907.1
Dietrich JC, Tanaka S, Westerink JJ et al (2011) Performance of the unstructured-mesh, SWAN + ADCIRC model in computing hurricane waves and surge. J Sci Comput 52:468–497. https://doi.org/10.1007/s10915-011-9555-6
Esteban M, Valenzuela VP, Matsumaru R, Mikami T, Shibayama T, Takagi H, Thao ND, Leon MD (2016) Storm surge awareness in the Philippines prior to Typhoon Haiyan: A comparative analysis with tsunami awareness in recent times. Coast Eng J 58(1):1640009–1640001. https://doi.org/10.1142/S057856341640009X
Flores PC, Reyes R, Amedo-Repollo CL, Rediang A, Alfante RM, Bauzon MDA, Pasaje N, Bringas D (2022) Local tide and geoid corrections significantly improve coastal retracked Jason sea surface heights in the Philippines. Phil J Sci 151(6B). https://doi.org/10.56899/151.6B.12
Gahtan J, Knapp KR, Schreck CJ, Diamond HJ, Kossin JP, Kruk MC (2024) International Best Track Archive for Climate Stewardship (IBTrACS) Project, Version 4r01: Western Pacific Basin. NOAA National Centers for Environmental Information. https://doi.org/10.25921/82ty-9e16. Accessed 05 May 2025
Gray WM (1968) Global view of the origin of tropical disturbances and storms. Mon Wea Rev 96:669–700. https://doi.org/10.1175/1520-0493(1968)096%253C0669:GVOTOO%253E2.0.CO;2
Hamlington B, Thompson P (2022), September 9 The Climate Data Guide: Tide gauge sea level data. NCAR Climate Data Guide. https://climatedataguide.ucar.edu/climate-data/tide-gauge-sea-level-data. Accessed 06 October 2025
Irish JL, Resio DT, Ratcliff JJ (2008) The influence of storm size on hurricane surge. J Phys Oceanogr 38(9):2003–2013. https://doi.org/10.1175/2008JPO3727.1
Kim DY, Park SH, Woo SB, Jeong KY, Lee EI (2017) Sea level rise and storm surge around the southeastern coast of Korea. J Coast Res 79:239–243. https://doi.org/10.2112/SI79-049.1
A
Knapp KR, Kruk MC, Levinson DH, Diamond HJ, Neumann CJ (2010) The International Best Track Archive for Climate Stewardship (IBTrACS): Unifying tropical cyclone best track data. Bull Am Meteorol Soc 91:363–376. https://doi.org/10.1175/2009BAMS2755.1Lapidez JP, Tablazon J, Dasallas L et al (2015) Identification of storm surge vulnerable areas in the Philippines through the simulation of Typhoon Haiyan-induced storm surge levels over historical storm tracks. Nat Hazards Earth Syst Sci 15:1473–1481. https://doi.org/10.5194/nhess-15-1473-2015
Lin N, Chavas D (2012) On hurricane parametric wind and applications in storm surge modeling. J Geophys Res Atmos 117(D9):D09120. https://doi.org/10.1029/2011JD017126
Ma P, Li G (2023) Comparison and analysis of detection methods for typhoon-storm surges based on tide-gauge data—Taking coasts of China as examples. Int J Environ Res Public Health 20(4):3253. https://doi.org/10.3390/ijerph20043253
McKeever A (2020), August 27 Why hurricane storm surges are so dangerous. National Geographic. https://www.nationalgeographic.com/environment/article/storm-surges. Accessed 16 June 2025
Mori N, Kato M, Kim S, Mase H, Shibutani Y, Takemi T, Yasuda T (2014) Local amplification of storm surge by Super Typhoon Haiyan in Leyte Gulf. Geophys Res Lett 41(14):5106–5113. https://doi.org/10.1002/2014GL060689
Morin VM, Warnitchai P, Weesakul S (2016) Storm surge hazard in Manila Bay: Typhoon Nesat (Pedring) and the SW monsoon. Nat Hazards 81(3):1569–1588. https://doi.org/10.1007/s11069-016-2146-y
A
Nationwide Operational Assessment of Hazards (NOAH) (2014) Compilation of storm surge occurrences in the Philippines. Project NOAH Open-File Rep 2:7–11National Disaster Risk Reduction and Management Council (NDRRMC) (2020a) SitRep No. 07 re Preparedness Measures for Super Typhoon Rolly (I.N. Goni). Quezon City, Philippines. https://ndrrmc.gov.ph/attachments/article/4135/Sitrep_no_7_re_STY_Rolly_as_of_NOV062020.pdf. Accessed 12 September 2025
National Disaster Risk Reduction and Management Council (NDRRMC) (2020b) SitRep No. 11 re Preparedness Measures and Effects for Typhoon Quinta (I.N. Molave). Quezon City, Philippines. https://ndrrmc.gov.ph/attachments/article/4130/SitRep_no_11_re_Typh00n_Quinta_as_of_09NOV2020.pdf. Accessed 12 September 2025
National Disaster Risk Reduction and Management Council (NDRRMC) (2020c) SitRep No. 13 for TC Kristine (2024). Quezon City, Philippines. https://ndrrmc.gov.ph/attachments/article/4271/SitRep_No_13_for_TC_
KRISTINE_2024.pdf (2025) Accessed 18 September
Needham HF, Keim BD (2012) A storm surge database for the Gulf of Mexico and Atlantic coast of the United States. Bull Am Meteorol Soc 93(4):485–496. https://doi.org/10.1175/BAMS-D-11-00150.1
Needham HF, Keim BD, Sathiaraj D (2015) A review of tropical cyclone-generated storm surges: Global data sources, observations, and impacts. Rev Geophys 53:545–591. https://doi.org/10.1002/2014RG000477
NOAA (n d.) What is storm surge? https://oceanservice.noaa.gov/facts/stormsurge-stormtide.html. Accessed 03 September 2025
Partnerships in Environmental Management for the Seas of East Asia (PEMSEA) (2021) National State of Oceans and Coasts 2018: Blue Economy Growth of Philippines. http://pemsea.org/sites/default/files/NSOC%20
Philippines%202019%2001082021.pdf (2025) Accessed 06 April
Pore NA, Barrientos CS (1976) Storm surge. Mesa New York Bight Atlas Monograph
Rego JL, Li C (2009) On the importance of the forward speed of hurricanes in storm surge forecasting: A numerical study. Geophys Res Lett 36(7):L07609. https://doi.org/10.1029/2008GL036953
Royal Observatory Hong Kong (1988) Tropical Cyclones in 1987. https://www.hko.gov.hk/en/publica/tc1987.pdf. Accessed 17 September 2025
Sherwood A, Bradley M, Rossi L, Guiam R, Mellicker B (2015), June 15 Resolving post-disaster displacement: Insights from the Philippines after Typhoon Haiyan (Yolanda). The Brookings Institution. https://reliefweb.int/report/philippines/resolving-post-disaster-displacement-insights-philippines-after-typhoon-haiyan. Accessed 03 September 2025
Smith A (2023), December 10 Understanding low pass filters: Definition, types, and applications. SoundSCAPEHQ. https://soundscapehq.com/what-is-a-low-pass-filter/. Accessed 22 May 2025
Soria JLA, Switzer AD, Villanoy CL, Fritz HM, Bilgera PHT, Cabrera OC, Siringan FP, Sta Maria YY, Ramos RD, Fernandez IQ (2016) Repeat storm surge disasters of Typhoon Haiyan and its 1897 predecessor in the Philippines. Bull Am Meteorol Soc 97(1):31–48. https://doi.org/10.1175/BAMS-D-14-00245.1
Towey KL, Booth JF, Rodriguez Enriquez A, Wahl T (2022) Tropical cyclone storm surge probabilities for the east coast of the United States: A cyclone-based perspective. Nat Hazards Earth Syst Sci 22(4):1287–1300. https://doi.org/10.5194/nhess-22-1287-2022
World Meteorological Organization (WMO) (2011) Guide to storm surge forecasting (2011 ed). World Meteorological Organization
Statements and Declarations
A
Funding
The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
A
Competing InterestsThe authors have no relevant financial or non-financial interests to disclose.