Tropical Atmospheric Intraseasonal Oscillations Leading to Sea Level Extremes in Coastal Indonesia during Recent Decades

William Kamp aDepartment of Atmospheric and Oceanic Sciences, University of Colorado Boulder, Boulder, Colorado

Search for other papers by William Kamp in
Current site
Google Scholar
PubMed
Close
,
Weiqing Han aDepartment of Atmospheric and Oceanic Sciences, University of Colorado Boulder, Boulder, Colorado

Search for other papers by Weiqing Han in
Current site
Google Scholar
PubMed
Close
,
Lei Zhang aDepartment of Atmospheric and Oceanic Sciences, University of Colorado Boulder, Boulder, Colorado

Search for other papers by Lei Zhang in
Current site
Google Scholar
PubMed
Close
,
Shoichiro Kido bApplication Laboratory, Research Institute for Value-Added-Information Generation, Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan

Search for other papers by Shoichiro Kido in
Current site
Google Scholar
PubMed
Close
, and
Julian P. McCreary cUniversity of Hawai‘i at Mānoa, Honolulu, Hawaii

Search for other papers by Julian P. McCreary in
Current site
Google Scholar
PubMed
Close
Open access

Abstract

Coastal flooding induced by sea surface high extreme (HEX) events is an increasing risk to human society and infrastructure as both urban growth in coastal areas and anthropogenic sea level rise continue, especially for island nations like Indonesia. This paper investigates the role of atmospheric intraseasonal oscillations (ISOs), which are dominated by the Madden–Julian oscillation (MJO), in forcing HEXs on the coasts of Indonesia bordering the Indian Ocean. We use satellite altimetry data from 1993 to 2021 and tide gauge observations to detect HEXs, and modeling experiments using both the Regional Ocean Modeling System and a Bayesian dynamic linear model to understand the forcing and processes. We find that HEXs exhibit strong seasonality, with most events occurring during boreal winter (December–February) and spring (March–May) that are dominated by seasonal-to-decadal and intraseasonal variability respectively. In 32% of the 56 HEX events detected, the amplitude of ISO-induced sea level anomalies (SLAs) exceeds that of seasonal-to-decadal SLAs. Surface wind stress associated with atmospheric ISOs is the major forcing for intraseasonal SLAs, and both the remote westerly wind stress from the Indian Ocean equator and northwesterly longshore wind stress at the Indonesian coasts play important roles in driving the HEXs. The MJO is the dominant cause of ISO-dominated HEXs and its impact shows strong seasonal differences. Spring MJOs are associated with stronger convective anomalies over the eastern Indian Ocean equator that drive stronger zonal winds across the equatorial basin that lead to more HEX events compared to winter MJOs when the convection is shifted southward.

© 2024 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: William Kamp, william.kamp@colorado.edu

Abstract

Coastal flooding induced by sea surface high extreme (HEX) events is an increasing risk to human society and infrastructure as both urban growth in coastal areas and anthropogenic sea level rise continue, especially for island nations like Indonesia. This paper investigates the role of atmospheric intraseasonal oscillations (ISOs), which are dominated by the Madden–Julian oscillation (MJO), in forcing HEXs on the coasts of Indonesia bordering the Indian Ocean. We use satellite altimetry data from 1993 to 2021 and tide gauge observations to detect HEXs, and modeling experiments using both the Regional Ocean Modeling System and a Bayesian dynamic linear model to understand the forcing and processes. We find that HEXs exhibit strong seasonality, with most events occurring during boreal winter (December–February) and spring (March–May) that are dominated by seasonal-to-decadal and intraseasonal variability respectively. In 32% of the 56 HEX events detected, the amplitude of ISO-induced sea level anomalies (SLAs) exceeds that of seasonal-to-decadal SLAs. Surface wind stress associated with atmospheric ISOs is the major forcing for intraseasonal SLAs, and both the remote westerly wind stress from the Indian Ocean equator and northwesterly longshore wind stress at the Indonesian coasts play important roles in driving the HEXs. The MJO is the dominant cause of ISO-dominated HEXs and its impact shows strong seasonal differences. Spring MJOs are associated with stronger convective anomalies over the eastern Indian Ocean equator that drive stronger zonal winds across the equatorial basin that lead to more HEX events compared to winter MJOs when the convection is shifted southward.

© 2024 American Meteorological Society. This published article is licensed under the terms of the default AMS reuse license. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: William Kamp, william.kamp@colorado.edu

1. Introduction

Sea surface height extreme (HEX) events result from sea level variations across time scales from hours to decades. Such events are of special interest because of their destructive potential to coastal communities. Due to global sea level rise (Church et al. 2014) and population growth in coastal areas (Muis et al. 2015), HEX-induced coastal flooding will become more damaging in the future. By 2100 it is estimated that up to 52% of the global population and 46% of global assets will be at flooding risk (Kirezci et al. 2020). We chose the Indonesian islands as our study region because of their growing coastal populations and the associated rapid urbanization (Muis et al. 2015).

Historically, most research on HEXs has focused on how synoptic-scale storm surges and high tides cause coastal flooding (Muis et al. 2016; Church et al. 2014; Sweet et al. 2018; Oppenheimer et al. 2019). At longer time scales, Han et al. (2022) looked at how anthropogenically induced global sea level rise combined with decadal climate variability affect the frequency of occurrence of HEX events along the Indonesian coast of the Indian Ocean. At intraseasonal time scales (variability from 10 to 90 days), studies of HEX events have been neglected.

The coasts of Indonesia that border the Indian Ocean are susceptible to wind-forced sea level anomalies (SLA) due to their location at the eastern coast of the tropical Indian Ocean (see Fig. 1 for area of interest). Winds over both the equatorial Indian Ocean and the Indonesian coasts affect coastal sea level (Clarke and Liu 1993; Yamagata et al. 1996; Iskandar et al. 2005; Iskandar and McPhaden 2011; Pujiana and McPhaden 2020). Westerly winds along the Indian Ocean equator cause surface Ekman mass convergence to the equator and raise sea level; the high SLAs propagate eastward via equatorial Kelvin waves to the west coast of Sumatra. There they subsequently propagate southeastward along the Indonesian coasts as coastal Kelvin waves, increasing sea level along the southern coasts of Java and the Nusa Tenggara island chain (Chen et al. 2015). Additionally, northwesterly and westerly longshore wind anomalies along the coasts of Sumatra and from Java to Nusa Tenggara also increase coastal sea level by causing coastal Ekman convergence and generating downwelling coastal Kelvin waves, enhancing the SLAs remotely generated from the equator (Iskandar et al. 2005).

Fig. 1.
Fig. 1.

The location of the four tide gauge stations (Padang: purple; Cilacap: blue; Prigi: teal; Benoa: magenta), and the areas where the satellite SLA and longshore wind data are taken along the coasts of Sumatra (orange area), Java (red area), and the equatorial region (2.5°S–2.5°N, 65°–95°E; boxed area). Color contours in the equatorial box show the lead days of equatorial zonal wind stress leading Java coastal SLA, when the correlation between zonal wind at a specific location within the boxed region and Java SLA obtains the maximum. These lead days are used for obtaining the equatorial zonal wind predictor in Eq. (2), as discussed in section 2d.

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

Sea level along the Indonesian coasts exhibits evident seasonal variations. The Australian–Indonesian monsoon winds shift from longshore northwesterlies in boreal winter to southeasterlies in boreal summer, causing an annual cycle with high SLAs in winter and low SLAs in late summer and early fall. Along the equator, semiannual zonal winds during Indian monsoon transition seasons induce high SLAs in May and November and low SLAs in February and August. Together the semiannual and annual cycles define the seasonal cycle in SLA in coastal Indonesia. A recent study by Han et al. (2022) explored HEX events and their compounding marine heatwaves using monthly data and focusing on the roles of seasonal, interannual, and decadal climate variability. They found that the northwest monsoon wind anomalies during northern winter play a crucial role in driving these HEXs; meanwhile, the equatorial westerlies and longshore northwesterlies during La Niña and/or negative Indian Ocean Dipole (IOD) enhance the seasonal northwest monsoon winds during December–February, generating strong HEX events. Since Han et al. (2022) used monthly data, which cannot resolve atmospheric intraseasonal oscillations (ISOs), the effects of ISOs on HEXs remain explored.

Atmospheric ISOs, which are associated with basin-scale surface wind patterns across the tropical Indian Ocean and the Maritime Continent, can force SLAs along the coast of Indonesia through both the remote and local processes described above (Iskandar et al. 2005). Of particular interest is the Madden–Julian oscillation (MJO) (Madden and Julian 1971, 1972), the dominant ISO mode in the tropical troposphere. Many MJO events are generated in the tropical Indian Ocean where they propagate eastward across the Maritime Continent to the western tropical Pacific Ocean (Zhang 2005). Impacts on HEXs along the Indonesian coasts are therefore expected.

In this paper, we examine the role of atmospheric ISOs in generating HEX events along the Indonesian coasts of the Indian Ocean since 1993, when satellite altimeter data are available. To achieve this goal, we analyze daily tide gauge observations and satellite altimetry data, and we perform ocean general-circulation-model experiments using the Regional Ocean Modeling System (ROMS). To assess the roles of remote wind forcing from the equatorial Indian Ocean and coastal wind forcing along the shores of Indonesian islands, we also utilize a Bayesian Dynamic Linear Model (DLM). Finally, we specifically examine the large and seasonally variable effects of MJOs on HEX events. Our results will help understand the processes leading to sea level surges and therefore improve their prediction. This may contribute to more informed decision making and risk assessments of coastal flooding.

2. Data and methodology

a. Observational and reanalysis datasets

1) Tide gauge and satellite observed sea level data

The daily observations from four tide gauge stations at the Sumatra, Java, and Bali coasts (see Fig. 1 for their locations) are analyzed. The Padang station is on southwestern coast of Sumatra (1.00°S, 100.37°E) and has data from 2005 to 2018. The Cilacap B station is on the southern coast of Java (7.75°S, 109.02°E) and has data from 2007 to 2017. The Prigi station is also on the southern coast of Java (8.28°S, 111.73°E) and has data from 2007 to 2018. The Benoa station is on the southern tip of Bali (8.75°S, 115.21°E) and has data from 2006 to 2018. The daily data from all four stations are from the Permanent Service for Mean Sea Level (PSMSL; https://psmsl.org/data/obtaining/) (Holgate et al. 2013). The data are corrected for tidal and inverted barometer effects as well as glacial isostatic adjustments by PSMSL.

Gridded, daily, satellite altimeter data for January 1993–December 2021 from the Data Unification and Altimeter Combination System (DUACS) (Taburet et al. 2019) are used to document sea level variability and identify HEX events. Data for from January to February 2022 are used for filtering (section 2b) to prevent edge effects, but not for any other analyses. This dataset uses all satellite data (all-sat) that contain daily sea level values with a resolution of 0.25° × 0.25° over the global ocean (https://duacs.cls.fr/duacs-data/duacs-products/). A comparable dataset based on two satellites (two-sat) is the Copernicus Climate Change Service (C3S) dataset (https://cds.climate.copernicus.eu/cdsapp#!/dataset/satellite-sea-level-global?tab=form).

The DUACS dataset has the advantage of providing the best estimates of sea level values, particularly for smaller-spatial-scale variability during the period when multiple satellites are available, whereas the C3S dataset has the advantage of homogeneity and stability throughout the data record. Area-averaged SLA values along the Java coast (Fig. 1) are used for comparing the two datasets. The standard deviations are 0.140 and 0.143 m for the all-sat and two-sat data respectively, and their correlation is 0.996 (Fig. S1 in the online supplemental material). Given that the two datasets agree well, hereafter we use the all-sat DUACS data for our analysis. Due to land contamination, satellite altimetry measures sea level over the shelf instead of right on the coast, unlike the tide gauges. Satellite data at the closest grid points to the tide gauge locations are used to compare with tide gauge observations.

2) Surface wind data

Daily 10-m wind data on a 0.25° × 0.25° grid are used to understand the role of surface winds in driving HEXs. The dataset is from the fifth generation of the European Center for Medium-Range Weather Forecasts (ECMWF) reanalysis (ERA5) for global climate and weather available from 1959 to the present (https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels?tab=form) (Hersbach et al. 2020). The daily values are the mean of the 24 hourly observations each day. The Cross-Calibrated Multi-Platform Winds version 3.0 (CCMP) available for 1993–2019 are also analyzed to validate the ERA5 winds. CCMP winds are a blended satellite product with a resolution of 0.25° × 0.25° and 6 h (https://data.remss.com/ccmp/v03.0/daily/) (Mears et al. 2022). The daily values are the mean of the four 6-hourly observations each day. Neither dataset contained wind stress variable, so the zonal and meridional wind stress components are calculated using the bulk formulas
τx=ρcduu2+υ2,τy=ρcdυu2+υ2
where ρ = 1.175 kg m−3 is air density, cd = 0.0015 is the drag coefficient, and u and υ are the 10-m zonal and meridional components of wind velocity. Figure 1 shows the grid cells used to find the coastal wind for Sumatra (orange) and Java (red).

3) Outgoing longwave radiation data

The 2.5° × 2.5°, daily interpolated satellite outgoing longwave radiation (OLR) product from the National Oceanic and Atmospheric Administration (NOAA) (https://psl.noaa.gov/data/gridded/data.olrcdr.interp.html) (Liebmann and Smith 1996) is used to document convection associated with the MJO events.

b. Isolation of intraseasonal variability

A Butterworth filter is used to isolate the intraseasonal variability of atmospheric and oceanic variables. The intraseasonal wind and OLR anomalies (OLRA) are calculated by subtracting the monthly climatology and then applying a 10–90-day bandpass Butterworth filter. For sea level, a high-pass filter of <105 days is used to retain the SLA spectral peak in the equatorial Indian Ocean at the 90-day period (Han 2005). Note that although satellite sea level data are daily, their effective temporal resolution is ∼10 days (Ballarotta et al. 2019). Therefore, no filtering of synoptic signals is necessary since they cannot be resolved.

c. Definition of HEX and identification of co-occurring MJO events

A HEX event is defined to occur when the magnitude of an SLA dataset exceeds the 90th percentile level in the 29-yr record of 1993–2021 and lasts for at least 5 consecutive days. Gaps between events of 2 days or less are considered as a continuous event. For instance, 10 extreme days of high SLA followed by 2 days of low SLA (<90th percentile) and then 11 extreme sea level days is defined as a 23-day HEX event. In contrast, 10 days of extreme SLA followed by 3 days of low SLA and then only 2 extreme SLA days is defined as a single, 10-day HEX event.

To identify MJO events co-occurring with HEX, both the Real-time Multivariate MJO index (RMM) (Wheeler and Hendon 2004) and OLR-based MJO Index (OMI) (Kiladis et al. 2014) were used. The RMM is defined using both zonal wind and OLR and is downloaded from Australian Bureau of Meteorology (http://www.bom.gov.au/climate/mjo/). The OMI is calculated from OLR only and is downloaded from the NOAA Physical Sciences Laboratory (https://www.psl.noaa.gov/mjo/mjoindex/).

Several criteria were applied to identify an MJO event that occurs during a HEX. First, the amplitude of one of the two indices must be ≥1.0 for at least 20 days, with an allowance for a short-lasting (≤3 days) and small decrease (≤0.2) below 1.0 (Figs. 2a,b) (Lu and Hsu 2017). Second, the event must propagate eastward, which is equivalent to counterclockwise rotation in the RMM phase diagram defined as the position of the first and last day above an amplitude 1.0 observed. Note that to compare RMM and OMI directly so an MJO will have the same phase the OMI axes are changed so that OMI PC2 is analogous to RMM PC1 and −OMI PC1 is analogous to RMM PC2. Third, the event must also satisfy the condition that the −10 W m−2 contour of OLRA shows at least 30° of eastward propagation (Fig. 2c) (Feng et al. 2015). Finally, since the SLAs forced by MJO-associated wind anomalies in the equatorial Indian Ocean can lead Java SLAs by up to 25 days (Iskandar et al. 2005), an MJO event was counted if it was observed within 25 days before the peak of a HEX event.

Fig. 2.
Fig. 2.

The three plots used to identify MJO events associated with HEXs. As an example, the MJO event associated with the 22 May 2002 HEX is shown. (a),(b) MJO phase diagrams for (a) RMM and (b) OMI. The first date is a month before HEX, the second date is the HEX, and the last date is 10 days after the HEX. (c),(bottom) Hovmöller diagram of 10°S–5°N-averaged OLR anomalies (color contours) with the −10 W m−2 value showing in black line contour; (top) area (red) where the OLR anomalies are shown.

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

d. ROMS and Bayesian DLM experiments

1) ROMS experiments

ROMS was set up from 25°S to 25°N with a horizontal resolution of 0.33° × 0.33° and 40 vertical sigma layers. It was forced using 3-hourly atmospheric forcing fields and daily river runoff of the Japanese 55-year atmospheric reanalysis–drive ocean (JRA55-do) data for the period 1958–2021 (Tsujino et al. 2018). Along the northern and southern open ocean boundaries, a mixed radiation/nudging boundary condition was used, in which temperature, salinity, and horizontal velocity were relaxed to the monthly values of ECMWF Ocean ReAnalysis System 4 (ORAS4) (Balmaseda et al. 2013) reanalysis data from 1958 to 2017 and to ORAS5 reanalysis data (Zuo et al. 2019) from 2018 to 2021, with a nudging time scale of 360 days (3 days) for the outflow (inflow) case. ORAS5 was used from 2018 to 2021 because ORAS4 is only available from 1958 to 2017. The open ocean boundary conditions allow the influence of global sea level rise on the Indonesian coast because these reanalysis datasets assimilated observed data (including satellite altimeter data), and there is no constraint for volume conservation over a specific ocean basin. For more detailed description of the model, readers may refer to Han et al. (2022) and Kido et al. (2019).

To isolate the effects of surface wind stress versus buoyancy fluxes in causing SLAs and HEXs, we perform two separate ROMS experiments. One is referred to as the ROMS main run (MR), which is a complete solution forced by surface wind stress, atmospheric surface temperature/humidity, shortwave/longwave radiation, and precipitation from the JRA55-do dataset. The ROMS wind stress run (WS) is similarly forced as the MR, but the forcing fields are set to their climatology except for the wind stress. Therefore, SLAs from ROMS WS experiment are mainly forced by surface wind stress.

2) Bayesian dynamic linear model

A Bayesian DLM is used to quantify the effects of remote and local wind stress forcing on SLAs along the Indonesian coasts. The Bayesian DLM has an observation [Eq. (2)] and state [Eq. (3)] equations.
Y(t)=b0(t)+b1(t)X1(t)+b2(t)X2(t)+b3(t)X3(t)+ϵ(t),ϵ(t)N[0,V(t)],
bi=bi(tΔt)+wi(t),wiN[0,Wi(t)],i=0,1,2,3.
The observation Eq. (2) is equivalent to a conventional linear regression model except that the regression coefficients vary with time. The state Eq. (3) controls the dynamic evolution of the coefficients bi and updates the predictive distribution of bi at each time step t based on the distribution of the previous time step t − Δt and the probability of the observed Y conditional on bi at time t using Bayes theorem (Petris et al. 2009). Specifically, each coefficient bi is found using Kalman filtering and smoothing with the regression coefficient of the conventional linear regression model as its initial guess (Han et al. 2017). The terms ϵ(t) and wi(t) represent the independent white noise errors with a normal distribution around zero and variances of V(t) and Wi(t). Discussions on the advantage of Bayesian DLM over the conventional regression can be found in Han et al. (2017).

In this paper, the predictand Y(t) is SLA averaged over the Java coast and the predictors Xi(t), with i = 1, 2, 3, are remote zonal wind stress averaged over the equatorial Indian Ocean; longshore wind stress averaged along the western coast of Sumatra; and local longshore wind stress averaged along the southern coast of Java (Fig. 1). As we shall see in section 3a below, SLAs along the Indonesian coasts are highly coherent especially between the coasts of Java and Nusa Tenggara. Therefore, we focus on discussing the SLAs along Java coasts using the Bayesian DLM.

Because it takes time for SLA signals to propagate from the equatorial and Sumatra regions to the Java coast, we allow for lead times between the equatorial and Sumatra winds and coastal SLA. To find the lead times for the equatorial region we first calculate the lead/lag correlation between the meridionally averaged zonal wind stress for at each longitude j in the equatorial region (2.5°S–2.5°N, 65°–95°E; boxed area in Fig. 1) and the mean Java coastal SLA, and define tj to be the time that gives the maximum correlation (Fig. 1, color contours in the box). It follows that at longitude j wind stress τjx(ttj) most strongly impacts Java SLA. The time series of the remote equatorial wind stress predictor X1(t) is then obtained by averaging τjx(ttj) for each longitude in the boxed region. The Sumatra longshore wind stress predictor X2(t) is obtained similarly. The Java longshore wind stress predictor X3(t) is used with a lag of zero days.

We use satellite-observed SLAs as the predictand and the three predictors calculated using ERA5 winds to obtain coefficients bi of the Bayesian DLM. Then the b1(t)X1(t), b2(t)X2(t), and b3(t)X3(t) terms represent Java SLAs induced by equatorial zonal wind, Sumatra longshore wind, and Java longshore wind, respectively, and the b0(t) term represents the SLA that cannot be explained by the three wind predictors.

3. Results

In this section, we discuss properties and causes of HEX events. We begin by examining the sea level variability along the Indian Ocean coasts of Indonesia and define a representative SLA for the region (section 3a). Next, we use the SLA data from section 3a to detect HEXs (section 3b). We then analyze the relative impacts of wind versus buoyancy in forcing sea level, especially during HEXs (section 3c). Next the overall contributions from ISOs versus longer variability are compared (section 3d), and the intraseasonal forcing from remote wind, Sumatra longshore wind, and Java longshore are analyzed using a Bayesian DLM (section 3e). Finally, the specific role of the MJO and its seasonal pattern are examined (section 3f).

a. Measure of SLAs along the coasts of Indonesia bordering the Indian Ocean

We used daily satellite altimetry data from 1993 to 2021 to track SLAs along the coasts of Indonesian Islands that border the Indian Ocean: Sumatra, Java, and Nusa Tenggara. To validate the satellite-data performance, we compared tide-gauge observations at the four stations with the satellite altimeter data at the closest locations during their overlapping period of 2006–19 (Figs. 3a–d). For this comparison, the mean values from 2006 to 2019 are removed from each record.

Fig. 3.
Fig. 3.

Sea level anomaly (SLA) data from the tide gauges (blue) and the closest points for the satellites altimetry observations (orange) during 2006–18 for (a) Cilacap, (b) Prigi, (c) Benoa, and (d) Padang tide gauge stations. See Fig. 1 for their locations. The means of the tide gauge and satellite data for their overlapping periods are removed from each dataset. The standard deviations (std) in mm and the correlation coefficients are shown. (e) The tide gauge data from (a)–(d) are shown together.

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

The tide gauge and satellite records agree well at all four locations, with correlation coefficients of 0.95–0.97 along Java-to-Nusa Tenggara coasts and 0.90 at the coast of southwestern Sumatra (Fig. 3). The good agreements suggest that, even though the effective temporal resolution of satellite data is ∼10 days (Ballarotta et al. 2019) and the tide gauge is daily, the satellite data accurately captured the SLA signals at intraseasonal and longer time scales, and thus are suitable for detecting HEX events. Spectral analysis confirmed that both the satellite and tide gauge data show consistent annual, semiannual, and intraseasonal spectral peaks, except for the 14-day peak that appeared only in tide gauge data. This is expected because satellite data have an effective resolution of 10 days and thus cannot resolve the 14-day cycle (Fig. S2). We repeated the comparisons shown in Fig. 3 using 9-day running means for both the satellite and tide gauges, and all correlations are essentially the same as those of the unfiltered data (see Fig. S3). One difference between the tide-gauge and satellite records is that the former has more high-frequency variability (noise). Another is that the tide-gauge signals have greater magnitudes, likely owing to the gridded satellite data being spatially and temporally averaged. Neither of these differences significantly impact the intraseasonal signals of interest here.

SLAs along the coasts of Java and Nusa Tenggara are highly coherent with each other (Figs. 3a–c), but not with the record off southwestern Sumatra (Fig. 3d). Indeed, SLA correlations for the Java and Nusa Tenggara tide gauges range from 0.93 to 0.98, whereas their correlations with the Sumatra tide gauge are only 0.53–0.66 (Fig. 3e). Given the high coherency of coastal SLAs from Java to Nusa Tenggara (also see Han et al. 2022), we use the Java SLAs as a representative for the region. We note that the average of all the satellite SLAs along the southern coast of Java (Fig. 1, red area) agrees well with the daily tide gauge anomalies from Cilacap and Prigi, with a correlation of 0.97 at both locations.

For further analyses we seek to understand the effect of atmospheric ISOs on HEXs, so we focus on internal climate variability independent from anthropogenic induced sea level rise. To accomplish this, we take the average of the satellite SLAs along the southern coast of Java and subtract the linear trend to produce Java SLAs (JSLA). Sea level rise is removed from JSLA because a linear trend has been shown to be an effective approximation for anthropogenic global sea level rise in this region (Han et al. 2022). JSLA is used for subsequent analyses, including the detection of HEXs. JSLA and the HEXs include sea level variability from intraseasonal-to-decadal time scales, including the mean seasonal cycle. The mean seasonal cycle is kept because it is part of total sea level that controls coastal inundation and erosion.

b. Detection and seasonality of HEX events

Applying the definition of HEX from section 2c to JSLA, a total of 56 HEX events along Indonesia are identified for the 29-yr period of 1993–2021 (Fig. 4), with a maximum amplitude of 0.4 m. This 0.4 m amplitude is at the lower end of the 0.5–1 m surges due to tropical storms and high tides with a return period of 100 years along the Indonesian coasts (Muis et al. 2016). The frequency of HEX occurrences by month exhibits a clear seasonal pattern, with most events occurring during November–February and April–June, which correspond to the seasons when the mean seasonal cycle of JSLA is higher (Fig. 4, black curve). (See section 3d for a discussion of the different colors in each bar.)

Fig. 4.
Fig. 4.

The number of HEX events for each month (bar graph), together with the monthly mean SLA climatology for the 1993–2021 period (black curve). The HEX events are divided into two categories (see section 3d): The first with ISO signals (105-day high-pass filtered SLA) contributing less than the 105-day low-pass signals (blue) and the second with ISO signals contributing greater than the low-pass signals (red).

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

c. Wind versus buoyancy forcing

To understand the causes of HEXs, we first analyze results from ROMS experiments. ROMS MR, which is forced by both wind stress and buoyancy flux (see section 2d), successfully simulated JSLA with a correlation coefficient of 0.96 and standard deviation of 0.16 m, the latter compared to the 0.14 m from JSLA (Fig. 5a). Since the model results were 3-day means, a 3-day mean was taken for the satellite data for when comparing it to ROMS. Intraseasonal SLAs from ROMS MR and JSLA also agree, with a correlation of 0.86 and standard deviations of 0.062 and 0.056 m for JLSA, respectively (Fig. 5b). Similarly, the low-passed seasonal-to-decadal SLAs from ROMS MR and JSLA have a correlation of 0.97 and standard deviations of 0.14 and 0.13 m respectively (Fig. 5c). Additionally, ROMS MR captures all 56 HEX events, using the definition of ROMS MR having an SLA peak above the satellite 90th-percentile threshold within 10 days of the satellite-detected HEX events. The ability of ROMS MR to simulate sea level variability and HEX events suggests that the model captures the major processes that control HEXs and therefore is suitable for this study.

Fig. 5.
Fig. 5.

Time series of the Java coastal mean SLAs from ROMS (MR) simulation (blue) and 3-day mean SLAs from satellite observation (orange) for 1993–2021. (a) The original unfiltered data and the 90th percentile of the satellite data, the threshold that defines HEX. All the HEX dates are marked with vertical blue lines. (b) As in (a), but for the intraseasonal (105-day high-pass filtered) SLAs. (c) As in (a), but for the 105-day low-passed SLAs that include seasonal-to-decadal variability.

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

To identify the contribution of buoyancy forcing, we perform the ROMS WS experiment in which only nonseasonal variations in wind stress force SLAs. The difference between ROMS MR and ROMS WS, ROMS MR-ROMS WS, then estimates the influence of buoyancy forcing. The total SLAs (including intraseasonal and longer period variability) from ROMS WS agree well with that of ROMS MR (Figs. 6a,b), with correlation of 0.96 and standard deviation of 0.165 m compared to the 0.157 m of ROMS MR from 1993 to 2021. The intraseasonal SLAs of ROMS WS also agree well with that of ROMS MR, with correlation of 0.91 and standard deviations of 0.072 and 0.062 m for the ROMS WS and ROMS MR, respectively (Fig. 6c). Comparisons of these two experiments show that surface wind stress forcing (from ROMS WS) can either reproduce the full magnitude of HEX events simulated in ROMS MR (Fig. 6d, red and blue dots overlap and red dots above blue dots) or is the major cause for the HEXs (Fig. 6d, red dots below blue dots). Buoyancy forcing (difference between red and blue dots) can also contribute a few centimeters for some events, either positively (e.g., HEX numbers 2, 4, and 21 in Fig. 6d) or negatively (e.g., HEX numbers 18 and 31 in Fig. 6d). We conclude that wind stress forcing is the primary cause for SLAs at both intraseasonal and seasonal-to-decadal time scales.

Fig. 6.
Fig. 6.

(a) Time series of 3-day mean SLAs from the ROMS MR (blue) and ROMS wind stress (WS) run (red) for 1993–2021, with the 1993–2021 mean removed but linear trend retained. (b) As in (a), but with linear detrend removed from each experiment. (c) As in (b), but for intraseasonal (105-day high-pass filtered) SLAs from ROMS MR and ROMS WS. (d) Peak SLA values (y axis) from ROMS MR (blue dot) and ROMS WS (red dot) for each of the 56 HEX events in chronological order (x axis).

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

d. Impacts of atmospheric ISOs

To investigate the impact of forcing by atmospheric ISOs, we separate JSLA into intraseasonal (105-day high-pass filtered, HP; see section 2b) and seasonal-to-decadal (105-day low-pass filtered, LP) periods using a Butterworth filter. According to this split, two signals contribute to JSLA: an intraseasonal (HP) one and everything else (LP), that is, JSLA = HP + LP. Then, the blue bars in Fig. 4 indicate the number of HEX events that occur when HP < LP. The red bars indicate the number of HEX events when HP > LP. Thus, the two bars show HEX events that vary in the relative magnitude of their intraseasonal contribution: varying from weak (blue) to strong (red).

The maximum JSLA in the high-passed band is 0.28 m, which exceeds the 90th percentile of 0.18 m for the unfiltered daily JSLAs and is somewhat larger than the maximum value of 0.26 m from the low-passed JSLAs. On average, during the peak days of HEX events the LP signals are larger (mean = 0.15 m) compared to HP (ISO) signals (mean = 0.10 m). However, ISO signals are strong for many of the events and exceed LP signals in 32% (18/56) of the HEX (Table 1). As noted earlier, the frequency (i.e., number) of HEX events by month follows the mean seasonal cycle of SLA (Fig. 4, black curve), which suggests that the 105-day low-passed seasonal-to-decadal signals control the seasonal distribution of HEXs. Contrary to this idea, however, while the boreal-winter HEXs are generally dominated by LP signals (blue bars in Fig. 4), the boreal-spring ones are dominated by HP signals (red bars). (Hereafter we refer to the seasons as those of the Northern Hemisphere, such as boreal winter as “winter”). The seasonality in Fig. 4 is somewhat unexpected, since MJOs are often considered to be strongest during winter (Zhang 2005; Marshall et al. 2015). In fact, 12 of the 18 ISO dominated HEXs occur during spring while only 2 of the 18 occur during winter (see section 3f). A similar figure to Fig. 4 was produced using the mean SLA along southwestern coast of Sumatra (Fig. S4), and they both show peaks of ISO dominated HEXs in spring and LP dominated HEXs in winter, although there are some differences in the number of HEX events and seasonal cycle of SLA between Java and Sumatra coasts. While annual cycle dominates Java SLA due to the Australian–Indonesian monsoon wind forcing, semiannual cycle dominates Sumatra SLA due to its adjacency to the equator, where oceanic response resonates with semiannual wind forcing (Han et al. 1999).

Table 1.

Three categories for contributions from ISO forcing to HEXs defined in Fig. 4, showing properties for all the events (row 2), and events with ISO < LP (row 3) and ISO > LP (row 4). The columns show the total number of events that fit each category (column 2); the rate at which they co-occur with MJOs (column 3); and Java coastal SLAs, taken from the Bayesian DLM and averaged over all HEX events, that are induced by intraseasonal equatorial zonal wind (column 4), Sumatra LSW (column 5), and Java LSW (column 6). The percentage values in parentheses in column 3 show the rates of HEXs associated with MJOs relative to the number of HEXs shown in column 2.

Table 1.

e. Remote versus local forcing

To quantify the effects of remote versus local wind forcing on HEXs, we use the Bayesian DLM, with JSLA as the predictand (Y) and equatorial zonal wind stress, Sumatra longshore wind stress, and Java longshore wind stress from ERA5 reanalysis as predictors (X1, X2, and X3, respectively; see section 2d). Note that all three wind stress predictors act independently in driving Java coastal SLA. Both Y and the Xis are filtered to isolate the intraseasonal signals. Therefore, all results from the Bayesian DLM are only for the intraseasonal time scales.

The resulting SLAs due to wind forcing from the Bayesian DLM analysis (Y′) is Y without b0 and the noise term ϵ(t). The time series of Y′ has a correlation of 0.91 with JSLA and 0.78 with the Indonesian SLAs from ROMS WS (Fig. 7), demonstrating the success of both the Bayesian DLM analysis and ROMS solution, and confirming the dominance of intraseasonal wind forcing on intraseasonal SLAs. Using the three predictors, the Bayesian DLM explains most of the variability of JSLA. The good agreement between the Bayesian DLM results and observed/simulated SLAs demonstrate the robustness of the relationship between remote and local wind stress forcing and JSLA on the intraseasonal time scale.

Fig. 7.
Fig. 7.

The 105-day high-pass filtered SLAs averaged for the Java coast for the satellite observation (blue curve), the total wind forcing from the Bayesian DLM (orange curve), and the ROMS WS (purple) from 1993 to 2021. The figure is divided into four panels to clearly show the peaks. The dates of HEX events are marked with vertical dotted lines, with the red (blue) vertical lines marking the HEX events with intraseasonal SLAs greater (smaller) than the low-passed SLAs.

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

The Bayesian DLM results reveal that SLAs forced by the remote equatorial zonal wind are generally stronger compared to those forced by the longshore winds (second row of Table 1). During the peaks of HEX events, equatorial-forced SLAs are about 50% larger than Sumatra-forced SLAs and 200% larger than the locally forced (Java) SLAs. A possible reason they are larger is that the fetch of the equatorial winds is larger than those along the Sumatra and Java coasts.

To quantify ISO impacts further, we separately examine the HEX events for the two categories discussed earlier in Fig. 4. During strong ISO conditions (red bars in Fig. 4; marked by red vertical lines in Fig. 8), SLAs forced by the remote equatorial wind (purple curves) are most enhanced compared to the weak ISO condition (blue bars in Fig. 4; marked by blue vertical lines in Fig. 8). The effects of Sumatra and Java, longshore wind forcing are also larger for HEXs with strong ISOs, but by considerably less than the effects due to equatorial wind forcing as seen in Table 1. While the remote equatorial wind is generally the most important, Fig. 8 shows that for some individual events, SLAs induced by Sumatra and Java longshore winds can be larger than those forced by equatorial wind during HEXs. For instance, Java longshore wind almost entirely forced the SLA during the 4 December 2001 HEX and Sumatra longshore wind almost entirely forced the 1 April 2006 HEX.

Fig. 8.
Fig. 8.

The 105-day, high-pass JSLAs from the Bayesian DLM forced by remote equatorial zonal wind (purple), Sumatra longshore wind (LSW; blue), and Java LSW (orange) from 1993 to 2021. The figure is divided into four panels for clarity. The dates of HEX peaks are marked with vertical dotted lines, with the red (blue) vertical lines marking the HEX events with intraseasonal SLAs greater (smaller) than the low-pass SLAs.

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

f. Impacts of MJOs on HEXs and the observed seasonal patterns

Given that MJO is the dominant ISO mode, here we investigate its impact on HEXs. Table 1 (third column) shows that HEX events co-occurring with MJOs are relatively more common as the contribution of ISO forcing to them increases, with the percentages increasing from 39% to 66%. The frequent occurrence of ISO-dominated HEXs during spring season (Fig. 4) combined with the fact that MJOs are linked to larger ISO forcing (Table 1) implies that spring MJOs likely excite stronger HEXs compared to the winter MJOs, and this pattern is confirmed by Table 2. In addition, of the eight HEXs with the strongest ISO-induced SLAs, six are associated with MJOs and all of them occur during March–May (MAM) except for one in June. No such events occur during December–February (DJF). Note that while more MJOs co-occur with HEX during DJF (12) compared to MAM (9), LP-driven SLAs dominate in DJF. In fact, the average ISO contribution in spring (144 mm) is 59 mm more than that in DJF (85 mm) for HEX co-occurring with MJOs. Since MAM is the monsoon transition season, LP surface winds are overall weak. Therefore, MJOs dominate surface wind anomalies (Lee et al. 2013) and thus have larger contributions to HEXs than LP winds. While LP SLAs are weak, ISO contributions must be larger to generate HEXs because the 90th-percentile threshold of HEXs is based on daily SLAs that include seasonal-to-decadal variability.

Table 2.

Sea level anomalies associated with the 27 HEX events that co-occur with the MJOs (row 2), 9 of which occur in MAM (row 3) and 12 in DJF (row 4) are shown. The columns show Java coastal SLAs (mm) induced by intraseasonal equatorial zonal wind (column 2), Sumatra LWS (column 3), and Java LSW (column 4) from the Bayesian DLM averaged over the events in each category.

Table 2.

To understand why MJOs tend to force stronger HEX events along Indonesia during spring rather than winter, we perform composite analyses for surface wind stress, SLA, and OLRA associated with the HEX events that co-occurred with MJOs during DJF and MAM, respectively (Figs. 9 and 10), together with the differences between the seasons (Figs. S5 and S6). Day 0 represents the HEX peak day and days −3, −6, …, −15 represent 3, 6, …, 15 days prior to the HEX peak. The same composites for all MJOs events in DJF and MAM observed from 1993 to 2021 (1993 to 2019) are shown in Figs. S7a and S9a (Figs. S7b and S9b) using ERA5 (CCMP) winds. The composites in Fig. 9 show that the winds in the equatorial basin are stronger and the zonal extent of the equatorial winds is wider during spring than in winter. Even though the differences in zonal wind stress between the MAM and DJF composites are below 90% significance level in most regions of the equatorial basin (Fig. S6) due to the limited number of HEX-MJO co-occurrence cases, the differences in the corresponding SLAs are statistically significant (Fig. S5). This is because the SLA magnitudes of equatorial Kelvin waves are proportional to the zonal integral of zonal wind stress along the equator (i.e., xwxeτxeikxdx), consistent modest differences in wind stress that accumulate overtime as Kelvin waves propagate eastward can result in large differences in SLAs along the Indonesian coasts. While the limited record length constrains the statistical significance in the HEX-MJO composite differences, the similar patterns of their wind stress and OLRA compared to that of the all-MJO composites (cf. Figs. S6 and S9) suggest the robustness of our results. The stronger zonal wind and its wider zonal extent over the equator are most notable from composite day −15 to day −6, which correspond to the MJO phases 3–5 (Figs. S7 and S9). The stronger zonal wind anomalies and their larger zonal extent in the equatorial basin for spring MJOs are why MJOs, and therefore ISOs overall, have a greater impact on the SLAs during HEX events along Java coast in spring compared to winter. Similar composites using CCMP winds (Figs. S7b and S9b) yield the same results, despite the CCMP winds having a somewhat shorter observation period from 1993 to 2019.

Fig. 9.
Fig. 9.

Composites for intraseasonal wind stress (vectors) and SLA (color contours) for HEX events co-occurred with the MJOs from 1993 to 2021 during (a) DJF, based on 12 co-occurring HEX-MJO events and (b) MAM, based on 9 HEX-MJO events. The 10–90-day bandpass filtered wind stress data are calculated from ERA5 and the SLAs are from 105-day low-pass filtered satellite observed SLAs. Day 0 represents the day of the HEX peak and days −3, −6, …, −15 represent 3, 6, …, 15 days prior to the HEX peak, respectively.

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

Fig. 10.
Fig. 10.

As in Fig. 9, but with color contours showing composite of MJO-associated OLR anomalies.

Citation: Journal of Climate 37, 9; 10.1175/JCLI-D-23-0374.1

The composite OLR anomalies for the spring and winter HEXs with co-occurring MJOs (Fig. 10) resemble the composites of all MJO events (Fig. S8), with stronger and equatorially more symmetric convective anomalies in the eastern tropical Indian Ocean driving stronger and equatorially symmetric zonal winds across the equatorial basin in spring compared to winter (days −12, −9, and −6). Hovmöller diagrams for the MJO composites co-occurring with HEXs during spring and winter (Fig. S10) confirmed that stronger OLR (convection) anomalies occur in the eastern equatorial Indian Ocean during spring. In winter, the center of convection is shifted southward so the maximum zonal winds occur south of the equator where they do not force equatorial Kelvin waves. This southward shift reduces the equatorial wind forcing, but the longshore winds along the coasts of Sumatra and Java are minimally impacted (Figs. S6 and S9a). This is consistent with the small differences in JSLA forced by longshore winds relative to the remote equatorial winds (Table 2). The strong seasonality of MJO convection causes the differences in the surface winds between spring and winter, which results in the seasonal patterns of ISO contributions to the sea level of HEX events shown in Fig. 4.

4. Summary and discussion

In this paper, we investigate the roles of atmospheric intraseasonal oscillations in causing sea level height extreme events along the Indonesian coasts of the Indian Ocean from 1993 to 2021, using observational analysis combined with model experiments. Two experiments are performed using a regional ocean model: ROMS MR, which is the complete solution, and ROMS WS, which isolates the surface wind stress forcing. Their difference, MR − WS, assesses buoyancy forcing by heat and freshwater fluxes. A Bayesian DLM is used to estimate the effects of remote versus local wind stress forcing on the HEX events.

Daily satellite altimetry data, which agree well with daily tide gauge observations from all four tide gauge stations along the coasts of South Sumatra, Java, and Bali (Figs. 1 and 3), are used to detect the HEX events. Both satellite and tide gauge data show highly coherent sea level variability along the Indonesian southern coast, particularly between the Java and Bali coasts (Fig. 3). Consequently, we use the SLAs averaged along Java coast (JSLA) as a representative measure to examine the HEX events.

A total of 56 HEX events are detected, with SLAs exceeding a threshold value of the 90th percentile of satellite observation for the 1993–2021 period (0.18 m) (Figs. 3 and 4). ROMS MR and ROMS WS show that surface wind stress forcing is the dominant cause for HEXs along the Java coast, and both intraseasonal and seasonal-to-decadal SLAs are important in causing HEXs (Fig. 6). Among the 56 HEX events, 18 are ISO-dominated events and they primarily occur during boreal spring (MAM); in contrast, the HEXs dominated by the seasonal-to-decadal SLAs primarily occur in winter (Fig. 4). Results from the Bayesian DLM show that for ISO-dominated HEX events wind forcing from the equatorial Indian Ocean, along the southern Sumatra coast, and along the Java coast are all important in driving HEX events. The SLAs induced by remote equatorial winds have the largest overall impact (Table 1), but the Sumatra and Java winds dominate JSLA during some HEX events (Fig. 8).

MJOs are the dominant component of the ISOs and surface wind stress associated with the MJOs is the dominant cause for the ISO-dominated HEXs. The seasonality of MJO convection and surface wind patterns accounts for the seasonality of the ISO-forced HEXs shown in Fig. 4. During spring, convection associated with MJOs is strong in both the Bay of Bengal and Maritime Continent, whereas in winter the convection shifts southward (Fig. 10 and Fig. S8). The stronger convections over the equator during spring drive stronger zonal winds across the equatorial basin, accounting for the larger impacts of springtime MJOs (and thus ISOs) on HEXs.

As global sea level continues to rise due to anthropogenic warming and coastal populations grow, coastal regions and island nations are increasingly susceptible to HEX events. On the other hand, anthropogenic warming can potentially alter the MJO behavior, making the MJOs stay shorter in the tropical Indian Ocean but longer in the western tropical Pacific (Roxy et al. 2019). How the changing MJO behavior may change its impact on HEX events around Indonesia and other countries surrounding the tropical Indian and western Pacific Oceans is an important area of future study. A thorough understanding of the roles played by internal climate variability from intraseasonal to interdecadal time scales and their interplay with global sea level rise is critical for a successful mitigation of disastrous coastal flooding events.

Acknowledgments.

W. K. and W. H. are supported by NASA Ocean Surface Topography Science Team award 0NSSC21K1190 and NASA International Ocean Vector Wind Science Team award 80NSSC23K0982. A portion of the work was done when W. H. was visiting the International Space Science Institute (ISSI), Bern, Switzerland. She thanks the Johannes Geiss Fellowship from ISSI for providing the travel support, and Dr. Anny Cazenave for stimulating discussions. S. K. is supported by JSPS KAKENHI JP21K13997 and 23H01250. We thank Danni Du for providing the code that identifies the MJO events.

Data availability statement.

All the observational datasets used in this research are publicly available from links provided in section 2. Model outputs are available upon request.

REFERENCES

  • Ballarotta, M., and Coauthors, 2019: On the resolutions of ocean altimetry maps. Ocean Sci., 15, 10911109, https://doi.org/10.5194/os-15-1091-2019.

    • Search Google Scholar
    • Export Citation
  • Balmaseda, M. A., K. Mogensen, and A. T. Weaver, 2013: Evaluation of the ECMWF ocean reanalysis system ORAS4. Quart. J. Roy. Meteor. Soc., 139, 11321161, https://doi.org/10.1002/qj.2063.

    • Search Google Scholar
    • Export Citation
  • Chen, G., W. Han, Y. Li, D. Wang, and T. Shinoda, 2015: Intraseasonal variability of upwelling in the equatorial eastern Indian Ocean. J. Geophys. Res. Oceans, 120, 75987615, https://doi.org/10.1002/2015JC011223.

    • Search Google Scholar
    • Export Citation
  • Church, J., and Coauthors, 2014: Sea level change. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1137–1216, https://doi.org/10.1017/CBO9781107415324.026.

  • Clarke, A. J., and X. Liu, 1993: Observations and dynamics of semiannual and annual sea levels near the eastern equatorial Indian Ocean boundary. J. Phys. Oceanogr., 23, 386399, https://doi.org/10.1175/1520-0485(1993)023<0386:OADOSA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Feng, J., T. Li, and W. Zhu, 2015: Propagating and nonpropagating MJO events over maritime continent. J. Climate, 28, 84308449, https://doi.org/10.1175/JCLI-D-15-0085.1.

    • Search Google Scholar
    • Export Citation
  • Han, W., 2005: Origins and dynamics of the 90-day and 30–60-day variations in the equatorial Indian Ocean. J. Phys. Oceanogr., 35, 708728, https://doi.org/10.1175/JPO2725.1.

    • Search Google Scholar
    • Export Citation
  • Han, W., J. P. McCreary Jr., D. L. T. Anderson, and A. J. Mariano, 1999: Dynamics of the eastern surface jets in the equatorial Indian Ocean. J. Phys. Oceanogr., 29, 21912209, https://doi.org/10.1175/1520-0485(1999)029<2191:DOTESJ>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Han, W., G. A. Meehl, A. Hu, J. Zheng, J. Kenigson, J. Vialard, B. Rajagopalan, and Yanto, 2017: Decadal variability of the Indian and Pacific walker cells since the 1960s: Do they covary on decadal time scales? J. Climate, 30, 84478468, https://doi.org/10.1175/JCLI-D-16-0783.1.

    • Search Google Scholar
    • Export Citation
  • Han, W., and Coauthors, 2022: Sea level extremes and compounding marine heatwaves in coastal Indonesia. Nat. Commun., 13, 6410, https://doi.org/10.1038/s41467-022-34003-3.

    • Search Google Scholar
    • Export Citation
  • Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 19992049, https://doi.org/10.1002/qj.3803.

    • Search Google Scholar
    • Export Citation
  • Holgate, S. J., and Coauthors, 2013: New data systems and products at the permanent service for mean sea level. J. Coastal Res., 29, 493504, https://doi.org/10.2112/JCOASTRES-D-12-00175.1.

    • Search Google Scholar
    • Export Citation
  • Iskandar, I., and M. J. McPhaden, 2011: Dynamics of wind-forced intraseasonal zonal current variations in the equatorial Indian Ocean. J. Geophys. Res., 116, C06019, https://doi.org/10.1029/2010JC006864.

    • Search Google Scholar
    • Export Citation
  • Iskandar, I., W. Mardiansyah, Y. Masumoto, and T. Yamagata, 2005: Intraseasonal Kelvin waves along the southern coast of Sumatra and Java. J. Geophys. Res., 110, C04013, https://doi.org/10.1029/2004JC002508.

    • Search Google Scholar
    • Export Citation
  • Kido, S., T. Tozuka, and W. Han, 2019: Anatomy of salinity anomalies associated with the positive Indian Ocean dipole. J. Geophys. Res. Oceans, 124, 81168139, https://doi.org/10.1029/2019JC015163.

    • Search Google Scholar
    • Export Citation
  • Kiladis, G. N., J. Dias, K. H. Straub, M. C. Wheeler, S. N. Tulich, K. Kikuchi, K. M. Weickmann, and M. J. Ventrice, 2014: A comparison of OLR and circulation-based indices for tracking the MJO. Mon. Wea. Rev., 142, 16971715, https://doi.org/10.1175/MWR-D-13-00301.1.

    • Search Google Scholar
    • Export Citation
  • Kirezci, E., I. R. Young, R. Ranasinghe, S. Muis, R. J. Nicholls, D. Lincke, and J. Hinkel, 2020: Projections of global-scale extreme sea levels and resulting episodic coastal flooding over the 21st century. Sci. Rep., 10, 11629, https://doi.org/10.1038/s41598-020-67736-6.

    • Search Google Scholar
    • Export Citation
  • Lee, J.-Y., B. Wang, M. C. Wheeler, X. Fu, D. E. Waliser, and I.-S. Kang, 2013: Real-time multivariate indices for the boreal summer intraseasonal oscillation over the Asian summer monsoon region. Climate Dyn., 40, 493509, https://doi.org/10.1007/s00382-012-1544-4.

    • Search Google Scholar
    • Export Citation
  • Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77, 12751277.

    • Search Google Scholar
    • Export Citation
  • Lu, W., and P. C. Hsu, 2017: Factors controlling the seasonality of the Madden-Julian Oscillation. Dyn. Atmos. Oceans, 78, 106120, https://doi.org/10.1016/j.dynatmoce.2017.04.002.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci., 28, 702708, https://doi.org/10.1175/1520-0469(1971)028<0702:DOADOI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1972: Description of global-scale circulation cells in the tropics with a 40–50 day period. J. Atmos. Sci., 29, 11091123, https://doi.org/10.1175/1520-0469(1972)029<1109:DOGSCC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Marshall, A. G., H. H. Hendon, T. H. Durrant, and M. A. Hemer, 2015: Madden Julian Oscillation impacts on global ocean surface waves. Ocean Modell., 96, 136147, https://doi.org/10.1016/j.ocemod.2015.06.002.

    • Search Google Scholar
    • Export Citation
  • Mears, C., T. Lee, L. Ricciardulli, X. Wang, and F. Wentz, 2022: Improving the accuracy of the Cross-Calibrated Multi-Platform (CCMP) ocean vector winds. Remote Sens., 14, 4230, https://doi.org/10.3390/rs14174230.

    • Search Google Scholar
    • Export Citation
  • Muis, S., B. Güneralp, B. Jongman, J. C. Aerts, and P. J. Ward, 2015: Flood risk and adaptation strategies under climate change and urban expansion: A probabilistic analysis using global data. Sci. Total Environ., 538, 445457, https://doi.org/10.1016/j.scitotenv.2015.08.068.

    • Search Google Scholar
    • Export Citation
  • Muis, S., M. Verlaan, H. C. Winsemius, J. C. Aerts, and P. J. Ward, 2016: A global reanalysis of storm surges and extreme sea levels. Nat. Commun., 7, 11969, https://doi.org/10.1038/ncomms11969.

    • Search Google Scholar
    • Export Citation
  • Oppenheimer, M., and Coauthors, 2019: Sea level rise and implications for low lying islands, coasts and communities. The Ocean and Cryosphere in a Changing Climate, H.-O. Pörtner et al., Eds., Cambridge University Press, 321–445.

  • Petris, G., S. Petrone, and P. Campagnoli, 2009: Dynamic linear models. Dynamic Linear Models with R, Springer, 31–84.

  • Pujiana, K., and M. J. McPhaden, 2020: Intraseasonal Kelvin waves in the equatorial Indian Ocean and their propagation into the Indonesian seas. J. Geophys. Res. Oceans, 125, e2019JC015839, https://doi.org/10.1029/2019JC015839.

    • Search Google Scholar
    • Export Citation
  • Roxy, M. K., P. Dasgupta, M. J. McPhaden, T. Suematsu, C. Zhang, and D. Kim, 2019: Twofold expansion of the Indo-Pacific warm pool warps the MJO life cycle. Nature, 575, 647651, https://doi.org/10.1038/s41586-019-1764-4.

    • Search Google Scholar
    • Export Citation
  • Sweet, W. W. V., G. Dusek, J. Obeysekera, and J. J. Marra, 2018: Patterns and projections of high tide flooding along the U.S. coastline using a common impact threshold. NOAA Tech. Rep. NOS CO-OPS 086, 56 pp., https://www.tidesandcurrents.noaa.gov/publications/techrpt86_PaP_of_HTFlooding.pdf.

  • Taburet, G., A. Sanchez-Roman, M. Ballarotta, M. I. Pujol, J. F. Legeais, F. Fournier, Y. Faugere, and G. Dibarboure, 2019: DUACS DT2018: 25 years of reprocessed sea level altimetry products. Ocean Sci., 15, 12071224, https://doi.org/10.5194/os-15-1207-2019.

    • Search Google Scholar
    • Export Citation
  • Tsujino, H., and Coauthors, 2018: JRA-55 based surface dataset for driving ocean-sea-ice models (JRA55-do). Ocean Modell., 130, 79139, https://doi.org/10.1016/j.ocemod.2018.07.002.

    • Search Google Scholar
    • Export Citation
  • Wheeler, M. C., and H. H. Hendon, 2004: An all-season real-time multivariate MJO index: Development of an index for monitoring and prediction. Mon. Wea. Rev., 132, 19171932, https://doi.org/10.1175/1520-0493(2004)132<1917:AARMMI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Yamagata, T., K. Mizuno, and Y. Masumoto, 1996: Seasonal variations in the equatorial Indian Ocean and their impact on the Lombok throughflow. J. Geophys. Res., 101, 12 46512 473, https://doi.org/10.1029/95JC03623.

    • Search Google Scholar
    • Export Citation
  • Zhang, C., 2005: Madden-Julian Oscillation. Rev. Geophys., 43, RG2003, https://doi.org/10.1029/2004RG000158.

  • Zuo, H., M. A. Balmaseda, S. Tietsche, K. Mogensen, and M. Mayer, 2019: The ECMWF operational ensemble reanalysis-analysis system for ocean and sea ice: A description of the system and assessment. Ocean Sci., 15, 779808, https://doi.org/10.5194/os-15-779-2019.

    • Search Google Scholar
    • Export Citation

Supplementary Materials

Save
  • Ballarotta, M., and Coauthors, 2019: On the resolutions of ocean altimetry maps. Ocean Sci., 15, 10911109, https://doi.org/10.5194/os-15-1091-2019.

    • Search Google Scholar
    • Export Citation
  • Balmaseda, M. A., K. Mogensen, and A. T. Weaver, 2013: Evaluation of the ECMWF ocean reanalysis system ORAS4. Quart. J. Roy. Meteor. Soc., 139, 11321161, https://doi.org/10.1002/qj.2063.

    • Search Google Scholar
    • Export Citation
  • Chen, G., W. Han, Y. Li, D. Wang, and T. Shinoda, 2015: Intraseasonal variability of upwelling in the equatorial eastern Indian Ocean. J. Geophys. Res. Oceans, 120, 75987615, https://doi.org/10.1002/2015JC011223.

    • Search Google Scholar
    • Export Citation
  • Church, J., and Coauthors, 2014: Sea level change. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1137–1216, https://doi.org/10.1017/CBO9781107415324.026.

  • Clarke, A. J., and X. Liu, 1993: Observations and dynamics of semiannual and annual sea levels near the eastern equatorial Indian Ocean boundary. J. Phys. Oceanogr., 23, 386399, https://doi.org/10.1175/1520-0485(1993)023<0386:OADOSA>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Feng, J., T. Li, and W. Zhu, 2015: Propagating and nonpropagating MJO events over maritime continent. J. Climate, 28, 84308449, https://doi.org/10.1175/JCLI-D-15-0085.1.

    • Search Google Scholar
    • Export Citation
  • Han, W., 2005: Origins and dynamics of the 90-day and 30–60-day variations in the equatorial Indian Ocean. J. Phys. Oceanogr., 35, 708728, https://doi.org/10.1175/JPO2725.1.

    • Search Google Scholar
    • Export Citation
  • Han, W., J. P. McCreary Jr., D. L. T. Anderson, and A. J. Mariano, 1999: Dynamics of the eastern surface jets in the equatorial Indian Ocean. J. Phys. Oceanogr., 29, 21912209, https://doi.org/10.1175/1520-0485(1999)029<2191:DOTESJ>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Han, W., G. A. Meehl, A. Hu, J. Zheng, J. Kenigson, J. Vialard, B. Rajagopalan, and Yanto, 2017: Decadal variability of the Indian and Pacific walker cells since the 1960s: Do they covary on decadal time scales? J. Climate, 30, 84478468, https://doi.org/10.1175/JCLI-D-16-0783.1.

    • Search Google Scholar
    • Export Citation
  • Han, W., and Coauthors, 2022: Sea level extremes and compounding marine heatwaves in coastal Indonesia. Nat. Commun., 13, 6410, https://doi.org/10.1038/s41467-022-34003-3.

    • Search Google Scholar
    • Export Citation
  • Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 19992049, https://doi.org/10.1002/qj.3803.

    • Search Google Scholar
    • Export Citation
  • Holgate, S. J., and Coauthors, 2013: New data systems and products at the permanent service for mean sea level. J. Coastal Res., 29, 493504, https://doi.org/10.2112/JCOASTRES-D-12-00175.1.

    • Search Google Scholar
    • Export Citation
  • Iskandar, I., and M. J. McPhaden, 2011: Dynamics of wind-forced intraseasonal zonal current variations in the equatorial Indian Ocean. J. Geophys. Res., 116, C06019, https://doi.org/10.1029/2010JC006864.

    • Search Google Scholar
    • Export Citation
  • Iskandar, I., W. Mardiansyah, Y. Masumoto, and T. Yamagata, 2005: Intraseasonal Kelvin waves along the southern coast of Sumatra and Java. J. Geophys. Res., 110, C04013, https://doi.org/10.1029/2004JC002508.

    • Search Google Scholar
    • Export Citation
  • Kido, S., T. Tozuka, and W. Han, 2019: Anatomy of salinity anomalies associated with the positive Indian Ocean dipole. J. Geophys. Res. Oceans, 124, 81168139, https://doi.org/10.1029/2019JC015163.

    • Search Google Scholar
    • Export Citation
  • Kiladis, G. N., J. Dias, K. H. Straub, M. C. Wheeler, S. N. Tulich, K. Kikuchi, K. M. Weickmann, and M. J. Ventrice, 2014: A comparison of OLR and circulation-based indices for tracking the MJO. Mon. Wea. Rev., 142, 16971715, https://doi.org/10.1175/MWR-D-13-00301.1.

    • Search Google Scholar
    • Export Citation
  • Kirezci, E., I. R. Young, R. Ranasinghe, S. Muis, R. J. Nicholls, D. Lincke, and J. Hinkel, 2020: Projections of global-scale extreme sea levels and resulting episodic coastal flooding over the 21st century. Sci. Rep., 10, 11629, https://doi.org/10.1038/s41598-020-67736-6.

    • Search Google Scholar
    • Export Citation
  • Lee, J.-Y., B. Wang, M. C. Wheeler, X. Fu, D. E. Waliser, and I.-S. Kang, 2013: Real-time multivariate indices for the boreal summer intraseasonal oscillation over the Asian summer monsoon region. Climate Dyn., 40, 493509, https://doi.org/10.1007/s00382-012-1544-4.

    • Search Google Scholar
    • Export Citation
  • Liebmann, B., and C. A. Smith, 1996: Description of a complete (interpolated) outgoing longwave radiation dataset. Bull. Amer. Meteor. Soc., 77, 12751277.

    • Search Google Scholar
    • Export Citation
  • Lu, W., and P. C. Hsu, 2017: Factors controlling the seasonality of the Madden-Julian Oscillation. Dyn. Atmos. Oceans, 78, 106120, https://doi.org/10.1016/j.dynatmoce.2017.04.002.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1971: Detection of a 40–50 day oscillation in the zonal wind in the tropical Pacific. J. Atmos. Sci., 28, 702708, https://doi.org/10.1175/1520-0469(1971)028<0702:DOADOI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Madden, R. A., and P. R. Julian, 1972: Description of global-scale circulation cells in the tropics with a 40–50 day period. J. Atmos. Sci., 29, 11091123, https://doi.org/10.1175/1520-0469(1972)029<1109:DOGSCC>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Marshall, A. G., H. H. Hendon, T. H. Durrant, and M. A. Hemer, 2015: Madden Julian Oscillation impacts on global ocean surface waves. Ocean Modell., 96, 136147, https://doi.org/10.1016/j.ocemod.2015.06.002.

    • Search Google Scholar
    • Export Citation
  • Mears, C., T. Lee, L. Ricciardulli, X. Wang, and F. Wentz, 2022: Improving the accuracy of the Cross-Calibrated Multi-Platform (CCMP) ocean vector winds. Remote Sens., 14, 4230, https://doi.org/10.3390/rs14174230.

    • Search Google Scholar
    • Export Citation
  • Muis, S., B. Güneralp, B. Jongman, J. C. Aerts, and P. J. Ward, 2015: Flood risk and adaptation strategies under climate change and urban expansion: A probabilistic analysis using global data. Sci. Total Environ., 538, 445457, https://doi.org/10.1016/j.scitotenv.2015.08.068.

    • Search Google Scholar
    • Export Citation
  • Muis, S., M. Verlaan, H. C. Winsemius, J. C. Aerts, and P. J. Ward, 2016: A global reanalysis of storm surges and extreme sea levels. Nat. Commun., 7, 11969, https://doi.org/10.1038/ncomms11969.

    • Search Google Scholar
    • Export Citation
  • Oppenheimer, M., and Coauthors, 2019: Sea level rise and implications for low lying islands, coasts and communities. The Ocean and Cryosphere in a Changing Climate, H.-O. Pörtner et al., Eds., Cambridge University Press, 321–445.

  • Petris, G., S. Petrone, and P. Campagnoli, 2009: Dynamic linear models. Dynamic Linear Models with R, Springer, 31–84.

  • Pujiana, K., and M. J. McPhaden, 2020: Intraseasonal Kelvin waves in the equatorial Indian Ocean and their propagation into the Indonesian seas. J. Geophys. Res. Oceans, 125, e2019JC015839, https://doi.org/10.1029/2019JC015839.

    • Search Google Scholar
    • Export Citation
  • Roxy, M. K., P. Dasgupta, M. J. McPhaden, T. Suematsu, C. Zhang, and D. Kim, 2019: Twofold expansion of the Indo-Pacific warm pool warps the MJO life cycle. Nature, 575, 647651, https://doi.org/10.1038/s41586-019-1764-4.

    • Search Google Scholar
    • Export Citation
  • Sweet, W. W. V., G. Dusek, J. Obeysekera, and J. J. Marra, 2018: Patterns and projections of high tide flooding along the U.S. coastline using a common impact threshold. NOAA Tech. Rep. NOS CO-OPS 086, 56 pp., https://www.tidesandcurrents.noaa.gov/publications/techrpt86_PaP_of_HTFlooding.pdf.

  • Taburet, G., A. Sanchez-Roman, M. Ballarotta, M. I. Pujol, J. F. Legeais, F. Fournier, Y. Faugere, and G. Dibarboure, 2019: DUACS DT2018: 25 years of reprocessed sea level altimetry products. Ocean Sci., 15, 12071224, https://doi.org/10.5194/os-15-1207-2019.

    • Search Google Scholar
    • Export Citation
  • Tsujino, H., and Coauthors, 2018: JRA-55 based surface dataset for driving ocean-sea-ice models (JRA55-do). Ocean Modell., 130, 79139, https://doi.org/10.1016/j.ocemod.2018.07.002.

    • Search Google Scholar
    • Export Citation
  • Wheeler, M. C., and H. H. Hendon, 2004: An all-season real-time multivariate MJO index: Development of an index for monitoring and prediction. Mon. Wea. Rev., 132, 19171932, https://doi.org/10.1175/1520-0493(2004)132<1917:AARMMI>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Yamagata, T., K. Mizuno, and Y. Masumoto, 1996: Seasonal variations in the equatorial Indian Ocean and their impact on the Lombok throughflow. J. Geophys. Res., 101, 12 46512 473, https://doi.org/10.1029/95JC03623.

    • Search Google Scholar
    • Export Citation
  • Zhang, C., 2005: Madden-Julian Oscillation. Rev. Geophys., 43, RG2003, https://doi.org/10.1029/2004RG000158.

  • Zuo, H., M. A. Balmaseda, S. Tietsche, K. Mogensen, and M. Mayer, 2019: The ECMWF operational ensemble reanalysis-analysis system for ocean and sea ice: A description of the system and assessment. Ocean Sci., 15, 779808, https://doi.org/10.5194/os-15-779-2019.

    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    The location of the four tide gauge stations (Padang: purple; Cilacap: blue; Prigi: teal; Benoa: magenta), and the areas where the satellite SLA and longshore wind data are taken along the coasts of Sumatra (orange area), Java (red area), and the equatorial region (2.5°S–2.5°N, 65°–95°E; boxed area). Color contours in the equatorial box show the lead days of equatorial zonal wind stress leading Java coastal SLA, when the correlation between zonal wind at a specific location within the boxed region and Java SLA obtains the maximum. These lead days are used for obtaining the equatorial zonal wind predictor in Eq. (2), as discussed in section 2d.

  • Fig. 2.

    The three plots used to identify MJO events associated with HEXs. As an example, the MJO event associated with the 22 May 2002 HEX is shown. (a),(b) MJO phase diagrams for (a) RMM and (b) OMI. The first date is a month before HEX, the second date is the HEX, and the last date is 10 days after the HEX. (c),(bottom) Hovmöller diagram of 10°S–5°N-averaged OLR anomalies (color contours) with the −10 W m−2 value showing in black line contour; (top) area (red) where the OLR anomalies are shown.

  • Fig. 3.

    Sea level anomaly (SLA) data from the tide gauges (blue) and the closest points for the satellites altimetry observations (orange) during 2006–18 for (a) Cilacap, (b) Prigi, (c) Benoa, and (d) Padang tide gauge stations. See Fig. 1 for their locations. The means of the tide gauge and satellite data for their overlapping periods are removed from each dataset. The standard deviations (std) in mm and the correlation coefficients are shown. (e) The tide gauge data from (a)–(d) are shown together.

  • Fig. 4.

    The number of HEX events for each month (bar graph), together with the monthly mean SLA climatology for the 1993–2021 period (black curve). The HEX events are divided into two categories (see section 3d): The first with ISO signals (105-day high-pass filtered SLA) contributing less than the 105-day low-pass signals (blue) and the second with ISO signals contributing greater than the low-pass signals (red).

  • Fig. 5.

    Time series of the Java coastal mean SLAs from ROMS (MR) simulation (blue) and 3-day mean SLAs from satellite observation (orange) for 1993–2021. (a) The original unfiltered data and the 90th percentile of the satellite data, the threshold that defines HEX. All the HEX dates are marked with vertical blue lines. (b) As in (a), but for the intraseasonal (105-day high-pass filtered) SLAs. (c) As in (a), but for the 105-day low-passed SLAs that include seasonal-to-decadal variability.

  • Fig. 6.

    (a) Time series of 3-day mean SLAs from the ROMS MR (blue) and ROMS wind stress (WS) run (red) for 1993–2021, with the 1993–2021 mean removed but linear trend retained. (b) As in (a), but with linear detrend removed from each experiment. (c) As in (b), but for intraseasonal (105-day high-pass filtered) SLAs from ROMS MR and ROMS WS. (d) Peak SLA values (y axis) from ROMS MR (blue dot) and ROMS WS (red dot) for each of the 56 HEX events in chronological order (x axis).

  • Fig. 7.

    The 105-day high-pass filtered SLAs averaged for the Java coast for the satellite observation (blue curve), the total wind forcing from the Bayesian DLM (orange curve), and the ROMS WS (purple) from 1993 to 2021. The figure is divided into four panels to clearly show the peaks. The dates of HEX events are marked with vertical dotted lines, with the red (blue) vertical lines marking the HEX events with intraseasonal SLAs greater (smaller) than the low-passed SLAs.

  • Fig. 8.

    The 105-day, high-pass JSLAs from the Bayesian DLM forced by remote equatorial zonal wind (purple), Sumatra longshore wind (LSW; blue), and Java LSW (orange) from 1993 to 2021. The figure is divided into four panels for clarity. The dates of HEX peaks are marked with vertical dotted lines, with the red (blue) vertical lines marking the HEX events with intraseasonal SLAs greater (smaller) than the low-pass SLAs.

  • Fig. 9.

    Composites for intraseasonal wind stress (vectors) and SLA (color contours) for HEX events co-occurred with the MJOs from 1993 to 2021 during (a) DJF, based on 12 co-occurring HEX-MJO events and (b) MAM, based on 9 HEX-MJO events. The 10–90-day bandpass filtered wind stress data are calculated from ERA5 and the SLAs are from 105-day low-pass filtered satellite observed SLAs. Day 0 represents the day of the HEX peak and days −3, −6, …, −15 represent 3, 6, …, 15 days prior to the HEX peak, respectively.

  • Fig. 10.

    As in Fig. 9, but with color contours showing composite of MJO-associated OLR anomalies.

All Time Past Year Past 30 Days
Abstract Views 169 169 0
Full Text Views 343 343 227
PDF Downloads 265 265 161