Volume 49, Issue 22 e2022GL101513
Research Letter
Open Access

Record-Breaking Precipitation in Indonesia's Capital of Jakarta in Early January 2020 Linked to the Northerly Surge, Equatorial Waves, and MJO

Sandro W. Lubis

Corresponding Author

Sandro W. Lubis

Pacific Northwest National Laboratory, Richland, WA, USA

Correspondence to:

S. W. Lubis,

[email protected]

Contribution: Conceptualization, Methodology, Software, Validation, Formal analysis, ​Investigation, Resources, Writing - original draft, Writing - review & editing, Visualization, Supervision

Search for more papers by this author
Samson Hagos

Samson Hagos

Pacific Northwest National Laboratory, Richland, WA, USA

Contribution: Supervision

Search for more papers by this author
Eddy Hermawan

Eddy Hermawan

National Research and Innovation Agency (BRIN), Jakarta, Indonesia

Contribution: ​Investigation, Supervision

Search for more papers by this author
Muhamad R. Respati

Muhamad R. Respati

School of Earth, Atmosphere and Environment, Monash University, Melbourne, VIC, Australia

Contribution: Conceptualization, Formal analysis, Writing - original draft

Search for more papers by this author
Ainur Ridho

Ainur Ridho

Search Engine for Risk and Actions on Resilience, Bandung, Indonesia

Contribution: Software, Formal analysis, Writing - original draft, Visualization

Search for more papers by this author
Risyanto

Risyanto

National Research and Innovation Agency (BRIN), Jakarta, Indonesia

Contribution: Formal analysis, ​Investigation, Resources, Software, Visualization

Search for more papers by this author
Jaka A. I. Paski

Jaka A. I. Paski

Indonesia Agency for Meteorology Climatology and Geophysics, Jakarta, Indonesia

Contribution: ​Investigation

Search for more papers by this author
Fadhlil R. Muhammad

Fadhlil R. Muhammad

School of Earth Sciences, University of Melbourne, Melbourne, VIC, Australia

Contribution: ​Investigation, Supervision

Search for more papers by this author
Siswanto

Siswanto

Indonesia Agency for Meteorology Climatology and Geophysics, Jakarta, Indonesia

Contribution: ​Investigation, Resources

Search for more papers by this author
Dian Nur Ratri

Dian Nur Ratri

Indonesia Agency for Meteorology Climatology and Geophysics, Jakarta, Indonesia

Wageningen University and Research, Wageningen, The Netherlands

Contribution: Writing - review & editing

Search for more papers by this author
Sonny Setiawan

Sonny Setiawan

Department of Geophysics and Meteorology, IPB University, Bogor, Indonesia

Contribution: ​Investigation

Search for more papers by this author
Donaldi S. Permana

Donaldi S. Permana

Indonesia Agency for Meteorology Climatology and Geophysics, Jakarta, Indonesia

Contribution: ​Investigation

Search for more papers by this author
First published: 07 November 2022
Citations: 7

This article was corrected on 24 DEC 2022. See the end of the full text for details.

Abstract

A record-breaking extreme rainfall event, the highest amount recorded since 1866, hit Indonesia's capital, Jakarta, in early January 2020. This torrential rainfall was mainly caused by the convergence of moisture-rich air due to an unusual blocking of cross-equatorial northerly surge over Northwest Java by the southerly winds induced by a cyclonic flow over the Indian Ocean. This condition caused a local increase in the amount of water vapor over Jakarta that eventually led to the formation of clouds and heavy precipitation. In addition, the concurrent occurrences of convectively active phases of equatorial waves (Kelvin, TD-type, and eastward propagating inertia-gravity waves) and Madden-Julian Oscillation during the event also partly contributed to the enhanced local moisture over the region by increasing low-level moisture flux convergence. Together, these large-scale dynamical drivers provided a convective environment that fostered the development of a massive rain-producing mesoscale convective system and, consequently, extreme rainfall over the region.

Key Points

  • An exceptionally high daily rainfall accumulation of up to 377 mm caused severe flash floods in Jakarta in early January 2020

  • The extreme rainfall was mainly caused by convergence of moisture-rich air due to an unusual blocking of northerly surges by southerly winds

  • The occurrences of equatorial waves and Madden Julian oscillation partly contributed to the enhanced local moisture by increasing low-level moisture convergence

Plain Language Summary

A record-breaking heavy rainfall event, the highest in the 156-year historical records, hit Indonesia's capital, Jakarta, in early January 2020. The flooding induced by extreme rainfall caused tremendous damage to infrastructure and had significant socioeconomic impacts. This study examines the atmospheric driving mechanisms underlying this extreme event. We found that heavy rains were triggered by an increase in humid air due to the blocking of the northerly surge over Northwest Java by the southerly winds. This condition created conditions for the development of deep convection and heavy rain over the region. In addition, the large-scale moisture convergence induced by some types of equatorial waves and Madden-Julian Oscillation during the event further enhanced local moisture and supported the development of convective cells and extreme rainfall. Deepening the understanding of the atmospheric mechanisms driving this event may provide valuable information for forecasting precipitation extremes over Jakarta in the future.

1 Introduction

Jakarta, the capital megacity of Indonesia located in the northwest of Java Island, Indonesia (Figure 1a), experienced an extraordinary heavy rainfall event in early January 2020. The highest amount reached up to 377 mm (14.83 inches) per day, making it a record-breaking number in observations since 1866. This extreme rainfall subsequently triggered widespread disastrous flooding in Jakarta and its surroundings in the early morning of 1 January 2020, leading to catastrophic losses. It is estimated that at least 173,000 people were evacuated, 66 people died, more than 60% of the residential areas were submerged, and the economic loss reached over US$700 million (Berlinger & Yee, 2020; Nisa, 2020). Because of the high vulnerability of Jakarta to rainfall extremes, a better understanding of the physical processes of heavy rainfall is needed to establish a reliable extended-range flood forecast system for this region.

Details are in the caption following the image

(a) Total precipitation (mm) estimated by C-band Doppler radar located in Soekarno-Hatta meteorological station from 31 December 2019 to 1 January 2020. Location of the city of Jakarta is qualitatively depicted by a yellow box, covering 6.00°–6.42°S, 106.62°–107.04°E. (b) Mean hourly accumulated precipitation from Global Precipitation Measurement satellite, C-band Doppler radar averaged over the box and 5 rain gauge stations. Time is in Universal Time Coordinated. (c) Daily mean accumulated precipitation based on GPM satellite, radar, and 5 rain gauge stations. Vertical bars in (b and c) denote a standard deviation of the rainfall variation among the stations. (d) Time series of annual maximum daily precipitation (RX1 day) from 5 rain gauge stations from 1960 to 2019 (see Figure S1 in Supporting Information S1 for the longer records back to 1866). The trend line is calculated for the Halim PK rain gauge station. (e) Return period of Jakarta RX1day for the period of 1960–2019. Blue lines correspond to the generalized extreme value (GEV) fit parameters for 2019 with a 95% confidence interval estimated with a non-parametric bootstrap following Siswanto et al. (2015).

Numerous studies have examined the effects of large-scale atmospheric circulation on precipitation extremes and major floods in Jakarta. For example, the major flooding event in February 2007 (the second highest record-breaking precipitation event) was attributed to an intense and persistent cross-equatorial northerly surge (CENS) that created an intensive low-level wind convergence and favorable dynamic conditions for the development of convective cells over the region (Hattori et al., 2011; Trilaksono et al., 2011; Wu et al., 2007). Other studies also showed the potential role of Madden Julian oscillation (MJO) in driving extreme rainfall and major floods in Jakarta (Aldrian, 2008; Lestari et al., 2022; Nuryanto et al., 2019; Wu et al., 20072013). In particular, Wu et al. (2013) reported that the extreme precipitation that caused a major flooding event in January 2013 was associated with the strong low-level convergence of winds induced by the active phase of the MJO. The enhanced convection induced by MJO contributed to the growth of the mesoscale convective system (MCS) and brought heavy rainfall from its activity over the region (Nuryanto et al., 20192021). Other major Jakarta flooding events in February 2002, 2008, and 2015, were also linked to similar causes, with unusual northerly winds associated with a CENS and the convective phases of MJO (Aldrian, 2008; Siswanto et al., 2015; Wu et al., 2007).

Notwithstanding the importance of CENS and MJO in extreme rainfall events, recent studies have found that equatorially trapped waves, including Kelvin waves, equatorial Rossby waves, tropical-depression (TD)-type waves, eastward propagating inertio-gravity (EIG) waves, and mixed Rossby-Gravity (MRG) waves, also play a major role in organizing tropical convection and triggering torrential rains and floods (Baranowski et al., 2020; Ferrett et al., 2020; Latos et al., 2021; Lubis & Jacobi, 2015; Lubis & Respati, 2021; Peatman et al., 2021; Sakaeda et al., 2020). In general, equatorial waves organize circulation that favors either active or suppressed convection (e.g., Kiladis et al., 2009; Wheeler and Kiladis, 1999), consequently influencing the frequency and intensity of precipitation events. Lubis and Respati (2021), for example, showed that the convectively active phases of Kelvin waves increased the probability of extreme rains over Java by about 30%–60%. Baranowski et al. (2020) also found evidence that Kelvin waves play a critical role in the majority of flooding events in Sumatra. Similarly, Latos et al. (2021) reported that Kelvin waves and equatorial Rossby waves can double the probability of floods and extreme rain events in Sulawesi. Despite clear evidence from these regional studies, it remains elusive the role of equatorial waves in the major flooding incident in Jakarta in early January 2020.

This study investigates the atmospheric driving mechanisms of the exceptionally extreme rainfall and devastating floods in Jakarta in early January 2020. We focus on the role of large-scale meteorological drivers in modulating large-scale circulations and moisture transport that favor the development of deep convection and local, extreme precipitation event in Jakarta using in situ measurements, satellite data, meteorological radar observations, and reanalysis data.

2 Data

The Integrated Multi-satellite Retrievals (IMERG) data from NASA (Huffman et al., 2020) are used for the period of 1 January 2001–30 December 2020, with an hourly temporal resolution and on a 0.1° × 0.1° grid. We also use the Himawari-8 satellite image of IR13 (infrared channel 13) (Bessho et al., 2016) with a spatial resolution of 4 km and a temporal resolution of an hour to observe the development of large mesoscale convective clouds (see Section 3.1). In addition to satellite data, a Gematronik C-band Doppler radar located in Soekarno-Hatta meteorological station (106.6502°E, 6.1669°S, 30 m above mean sea level, with maximum radius coverage of 250 km) is also used to derive the high resolution local rainfall field for a period of 31 December 2019–1 January 2020 (see Text S1 in Supporting Information S1). The daily rainfall measurements from 5 stations located around Jakarta (namely Kemayoran, Halim Perdana Kusumah (PK), Cengkareng, Tanjung Priok, and Pondok Betung) for the period of 31 December 2019—1 January 2020 operated by the Agency for Meteorology, Climatology and Geophysics of the Republic of Indonesia (BMKG) are also used. Finally, we also use reanalysis products to analyze dynamical fields, moisture, and sea surface temperature in this study (see Text S2 in Supporting Information S1).

3 Methods

3.1 Tracking of Mesoscale Convective System (MCS)

A graph theory based algorithm (GTG) (Whitehall et al., 2015) is used for automated identification and characterization of a large MCS. The GTG algorithm can handle the complex evolution of the MCS, as it allows MCS to have multiple cloud elements simultaneously at one time frame. Tracking of MCS is performed by applying the GTG algorithm to the hourly brightness temperature (TBB) data from the Himawari-8 satellite. In this study, MCS as a convective system is identified based on a set of criteria: an area of contiguous pixels covered more than 2400 km2 with TBB < 243 K, or an area where there exists a convective core containing brightness temperature range of at least 10 K. These criteria must be fulfilled and last continuously for longer than 3 hr (see the detailed procedures in Whitehall et al. (2015)).

3.2 HYSPLIT Model and Backward Trajectories

The mechanisms of moisture transport and the origins of extreme precipitation over Jakarta in early January 2020 are still uncertain. In this study, we employ a backward trajectory analysis to track the moisture source using a Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model version 5.1 (Stein et al., 2015). We run the model every 3 hr starting at 2100 UTC on 31 December 2019 (i.e., the highest peak of precipitation event) to generate 72-hr backward trajectories (from 500 to 2,000 m) above the ground in the 50 target points around Jakarta with a horizontal interval of 0.25° deg (see Figure 2k). We chose this altitude range because water vapor is highly concentrated in the lower troposphere. In addition, we also calculate relative moisture source contributions using a similar algorithm proposed by Nie and Sun (2022) (see Text S3 and Figure S1 in Supporting Information S1 for detailed information).

Details are in the caption following the image

(a) Hourly evolution of CS, cross-equatorial northerly surge (CENS), and southerly flow indices from 1200 UTC 31 Dec 2019 to 0600 UTC 1 January 2020. The red (orange) circle is the period active CENS (CS). The purple circle indicate period of active southerly winds. The first vertical line indicates the period of active CENS (0800 UTC), while the second one indicates the peak of the observed precipitation. (b) Time-latitude cross-sections of the meridional wind component at 925 hPa (shading) and the wind vectors averaged around Jakarta (106.62° and 107.04°E). The white circle indicates the period when CENS intruded into the northern coastline of Java Island (0900 UTC). The region highlighted with the red shading indicates the period of blocking CENS by southerly winds. (c–f) Time evolution of column integrated water vapor transport (shading, kg m−1 s−1) and the corresponding fluxes (vectors). The red (purple) rectangular box is the area used for defining CENS (southerly flow) index. The star indicates the location of Jakarta. (g–j) As in (c–f) but for the evolution of mean sea level pressure (shading) superimposed with 925-hPa winds streamlines (black contour). The climatological mean of December-January 925-hPa wind is shown as white streamlines. (k) The 72-hr backward trajectories of the moisture responsible for the extreme precipitation event. Blue shading indicates the time-integrated integrated water vapor transport for 72-hr prior to the period of maximum precipitation. (l) The relative moisture contributions from the different source regions (see Text S3 in Supporting Information S1 for details).

3.3 Filtering Technique and Wave Analysis

To isolate the equatorial wave signals in precipitation and other dynamical fields, we employ a two-dimensional fast Fourier transformation (2D fast fourier transform) filtering technique (Kiladis et al., 2009; Wheeler & Kiladis, 1999). The wavenumber-frequency domains of each wave mode used in the filtering technique are similar to Kiladis et al. (2009) and Lubis and Jacobi (2015) (Table S1 in Supporting Information S1). Furthermore, to retrieve the local phases and amplitudes of equatorial waves, we use a similar approach proposed by Riley et al. (2011) (see Text S4 in Supporting Information S1).

4 Characteristics of Heavy Rainfall

Figure 1 shows the characteristics of precipitation during the period from 31 December 2019 to 1 January 2020, over the flood zone. During this period, heavy rainfall covered most areas in Jakarta, with higher intensity in the eastern part of the city (Figure 1a). Based on the averaged in situ data from five stations, the first peak reached 20 ± 14 mm/hr at 1000 UTC on 31 December, and the second peak of rainfall reached 39 ± 28 mm/hr at 2100 UTC on 1 January (Figure 1b). These results were confirmed by radar and satellite observations, though there were differences in the amplitude and timing of precipitation (Figures 1b and 1c). In particular, the radar estimates were low compared to the in situ and satellite observations, possibly due to bias in calibration, correction, and quantitative precipitation estimation (e.g., Paski et al., 2020).

Daily rainfall measurements from five meteorological stations in Jakarta indicated that the highest rainfall accumulation of up to 377 mm per day was recorded at Halim PK Air Force Base in East Jakarta (Figure 1d). This event was unprecedented in Jakarta's historical rainfall database since 1866 and is estimated to have a return period of 300 years (Figure 1e and Figure S2 in Supporting Information S1). The return period is much longer compared to the previous major flood event reported in 2015, which is expected to occur once every 139 years (Siswanto et al., 2015). It is also clear from the time series of the highest annual rainfall events that changes in the intensity of maximum daily rainfall in Jakarta have tended to increase by 15.41 mm/decade (Figure 1d and Figure S2 in Supporting Information S1).

5 Large-Scale Atmospheric Drivers of Heavy Rainfall

To understand atmospheric mechanisms responsible for the exceptionally extreme rainfall and devastating flood event in Jakarta in early January 2020, we analyze potential attribution of large-scale meteorological phenomena that drive such event. Given the fact that during this period, the Indian Ocean Dipole was in a positive phase and the El Niurn:x-wiley:00948276:media:grl65075:grl65075-math-0001o-Southern Oscillation (ENSO) was in transition to its neutral phase (not shown), other large-scale phenomena beyond the low-frequency variability must have been responsible for driving the extreme event, as we discuss below.

5.1 The Role of the Cross-Equatorial Northerly Surge (CENS)

Unlike the previous four major flood/extreme rainfall events in Jakarta (i.e., 2 February 2002, 2 February 2007, 17 January 2013, and 10 February 2015), the occurrence of the cross-equatorial northerly surge (CENS) in early January 2020 was unusual. Figure 2a shows the hourly evolution of a cold surge (CS), the CENS, and southerly flow indices during the event. It is evident that the occurrence of the CENS was preceded by the active CS event at 00:00 UTC on 31 December 2019. The CS strengthened the northeasterly winds near the surface with a cold tongue of sea surface temperature (Figure S3 in Supporting Information S1), resulting in the CENS. Interestingly, prior to the arrival of the CENS on Java Island (compare the purple and red lines in Figure 2a), there had been an intrusion of southerly winds into the southwestern part of the island (purple line in Figure 2a). These southerly winds became stronger when the CENS became active and intruded into the island at 09:00 UTC on 31 December 2019. This condition led to the blocking of the CENS, prohibiting its further propagation to the south. The blocking of the CENS can also be clearly seen in Figure 2b. The northerly winds stalled after 09:00 UTC and were somewhat confined over the northern part of Java Island (see the first vertical dashed line in Figure 2b). Only after 21:00 UTC on 31 December 2019 did the northerly winds begin to move further to the south due to a weakening of the southerly winds.

Figures 2c–2f show the temporal and spatial evolutions of moisture flux (vectors) and IVT (color shading) during the event. Prior to the arrival of the CENS, the southerly winds had intruded into the southwestern part of Java Island and transported moist air over the region (Figure 2c). This moisture transported from the south before the arrival of the CENS might partially explain (if not entirely) the precipitation formation prior to 0900 UTC (Figure 1b). Furthermore, as the CENS became active and mature (Figures 2d–2f), a strong transport of moist-rich air induced by the CENS began to intrude into the northern part of the island. The intrusion of moist air induced by the CENS was then blocked by the transport of moist air from the south (see the vector direction), leading to the accumulation of moist air over northwest Java Island, including Jakarta. As a result, a massive amount of water vapor supported the development of a deep convective system characterized by the formation of clouds as high as 15 km over the region (Figure S4 in Supporting Information S1), leading to very intense precipitation. This is consistent with the highest precipitation peak at around 21:00 UTC on 31 December 2019 (Figure 1b).

To understand what drove the pre-existing southerly flow prior to the CENS in early January 2020, we also analyzed the temporal and spatial evolution of wind streamlines (contour) and mean-sea level pressure anomaly (color shading) during the event (Figures 2g and 2h). A cyclonic system emerged over the South Indian Ocean, which developed in time prior to the arrival of the CENS. This low-pressure system strengthened the southerly flow toward the island, causing a strong wind to propagate onshore, which blocked the CENS from migrating further southward. The existence of the southerly flow prior to the arrival of the CENS was unusual, both from a climatological perspective (see the white streamlines in Figures 2g and 2h) and considering the four other previous major flood events recorded (see Figure S5 in Supporting Information S1). Meanwhile, the strong CENS-induced moisture transport during this period is consistent with a large pressure gradient at sea level and the surface temperature gradient between the South China Sea and the North Java Sea (Figure S3 in Supporting Information S1).

The intense moisture transport in the presence of the CENS was also confirmed by the backward Lagrangian trajectory of moisture analysis based on the HYSPLIT model (Figures 2k and 2l). Consistent with the Eulerian transport pathways of IVT analysis earlier (Figures 2c–2f), the moisture source responsible for the extreme rainfall event mainly came from the South China Sea (Figure 2k). Overall, in the presence of the CENS, the South China Sea contributed most of the moisture (up to 59.81%), followed by the southern part of Sumatra Island (26.81%). There was a relatively small (2.76%) contribution of moisture source from the Southern Indian Ocean to the local increase in moisture over Jakarta. As we discussed earlier, the southern flow from the South Indian Ocean was not as a primary source of moisture in this extreme case; instead, it acted as a barrier that trapped the moist-rich air caused by the CENS over Jakarta.

In summary, the extreme precipitation event in early January 2020 was mainly caused by the convergence of moist air due to the blocking of the CENS by the southerly winds. This condition was a crucial factor in driving extremely heavy rain over Jakarta in early January 2020. This has two implications. First, this synoptic condition limited the moist air transport induced by the CENS to migrate further south, resulting in a significant increase in local water vapor content over Jakarta. Second, the effect of converging winds induced by the CENS, and southerly winds led to a mass accumulation that eventually caused vertical movement and the formation of massive heavy-rain clouds.

5.2 The Role of Equatorial Waves and MJO

We found that, in addition to the CENS, the concurrent occurrences of convectively active phases of equatorial waves and MJO partly contributed to the moisture increase and, hence, the development of extreme rainfall in Jakarta (Figure 3 and Figure S6 in Supporting Information S1). On 31 December 2019, three types of equatorial wave modes, including Kelvin waves, TD-type waves, EIG waves, and a convective phase of MJO, were observed concurrently over the region. These concurrent occurrences resulted in significant negative Outgoing Longwave Radiation (OLR) anomalies (enhanced large-scale convection) over Jakarta (Figure 3a). More specifically, the convectively active phase of Kelvin waves was observed over Jakarta on 31 December 2019 (Figures 3a and 3c), while the TD-type waves and EIG waves emerged a few days before the end of 2019 and continued until early January 2020 (Figures 3a, 3e and 3f). Unlike these three types of equatorial waves, the MRG wave was in its dry phase during the period of extreme rainfall (Figure 3g), and the ER wave-convective center had not yet reached Jakarta. Therefore, they did not contribute to the anomalous convective activity during the event (Figures 3a, 3d, and 3g). Meanwhile, the convective phase of MJO over Jakarta began on 27 December 2019 and lasted until 3 January 2020, as it moved slowly eastward and then transitioned to the dry phase (Figures 3a and 3b). The existence of the MJO above the Maritime Continent has been confirmed by other global MJO indices calculated based on univariate Empirical Orthogonal Function analysis of OLR (Figure S7a–S7c in Supporting Information S1), in which the MJO was transitioning from phase 4 to 5, with a relatively weak amplitude. However, its activity over the Maritime Continent was not observed in the multivariate MJO indices, such as Velocity Potential MJO index and Realtime Multivariate Index for tropical Intraseasonal oscillations (i.e., MJO was weak in phase 8, see Figures S7d–S7e in Supporting Information S1). This is because global multivariate MJO indices tend to be more associated with the wind circulation than the convection, while the filtered OLR is much more closely related to the convection (c.f. Gottschalck et al., 2013; Straub, 2013).

Details are in the caption following the image

(a) Time-longitude section of OLR anomalies averaged over 10°S–5°S, with an interval of 10 Wm−2 (color shading). Contour lines show the amplitudes of selected equatorial modes and Madden Julian oscillation (MJO) that have been wavenumber-frequency-filtered (see details in Section 3.3 and Table S1 in Supporting Information S1). The contour line for MJO is −17.12 W/m2 (−2 stddev), for Kelvin waves: −13.48 W/m2 (−2 σ), for ER waves: −17.58 W/m2 (−2 σ), for TD-Type waves: −7.56 W/m2 (−1.5 σ) and for eastward propagating inertio-gravity waves: −5.45 W/m2 (−1.5 σ), where σ is the standard deviation. The vertical line at 107°E marks the location of the city of Jakarta. The horizontal line depicts the period of 31 December 2019. Only the wet phases of the equatorial waves and MJO are shown. (b-g) Local wave phase diagrams of MJO and different types of equatorial waves during the major flood event. The two yellow dots indicate the period of 31 December 2019 and 1 January 2020. The local wave phase diagrams are constructed from the standardized wave-filtered OLR and its tendency centered at 106.5°E (reference longitude). Phases 4–6 (1–2 and 8) are termed as wet (dry) phases throughout the life cycle of waves. (h) Contribution of different types of equatorial waves and MJO on the total daily precipitation and Vertically integrated moisture flux convergence anomalies during the Jakarta Flood in 31 January 2019. The percentage (%) is calculated as a ratio of the each filtered-wave anomaly with respect to the total anomaly (blue bar).

The contribution of each wave mode to the anomalous precipitation during this extreme event is summarized in Figure 3h. Of the total anomalous precipitation (110 mm/day) on 31 December 2019 observed by NASA IMERG, about 23% can be attributed to equatorial waves and MJO (also see Figure S6 in Supporting Information S1 for the total precipitation modulated by these waves and MJO). The Kelvin, TD-type, and EIG waves made a strong positive contribution to the rainfall anomaly (20%), while MJO asserted only a weak positive influence (3%). The positive contributions of these wave modes and MJO to the total anomaly are consistent with the increase in low-level convergence of moisture flux induced by their activities, which enhanced local convection and, hence, increased precipitation over the region (Figure 3h).

Overall, our results indicate that Kelvin waves, TD-type waves, EIG waves and MJO partly contributed to the local increase in moisture over the region by increasing low-level moisture flux convergence. Although they played only a secondary role in increasing precipitation, their occurrence, along with the CENS, can enhance the probability of extreme rainfall over the region. Our statistical analysis showed that only 3.5% of the flooding events in Jakarta occurred in the presence of all these forcing (Figure S8 in Supporting Information S1). Nonetheless, we argue that this is not the sole factor that determined the unusual extreme precipitation event in early January 2020, as they contributed only one-fifth of the total anomaly. Our analysis indicated that the CENS was the main driver of the total moisture source over the region and that its meridional extents are more important than equatorial waves and MJO in triggering this extreme event, which had never been observed previously.

5.3 The Mesoscale Convective System

Occurrences of extreme rainfall and corresponding flooding events are often associated with the MCS, which is more favorable in the presence of the CENS, equatorial waves, and MJO (Latos et al., 2021; Mapes et al., 2006; Schumacher & Johnson, 2008; Wu et al., 2007). This section explains the characteristics of the MCS during the period of the extreme rainfall event.

Figure 4 shows the time evolution of the MCS superimposed with the total precipitation estimated by radar. In general, the spatial distribution between precipitation and the MCS was coherent, indicating MCS-driven precipitation during the event. The MCS began to develop at 0500 UTC on 31 December 2019 from around 7°S over the mountains to the south of Jakarta (Figure 4a). At 0600 UTC, the MCS grew stronger and moved northward toward Jakarta (Figures 4b and 4c). This development and northward propagation were very likely driven by the near-surface convergence due to warm-moist air near the surface (Mori et al., 2018; Wu et al., 2007). At around 0900 UTC, the MCS matured and reached Jakarta (Figure 4d). This maturation stage was associated with increased moisture induced by the CENS (Figure 2a), resulting in a stronger updraft and convection over the region (see the formation of high/deep clouds in Figure S4 in Supporting Information S1).

Details are in the caption following the image

Time evolution of the mesoscale convective system (MCS) calculated based on temperature black body (TBB) retrieved from Himawari satellite (color shading) superimposed with total precipitation (mm) estimated by a C-band Doppler radar (contour lines) from 0500 UTC 31 December 2019 to 0300 UTC 1 January 2020. The color gradation from light to dark blue indicates interior cold cloud with TBB ≤ 221 K.

After 0900 UTC, the MCS became stronger over the Java Sea and the coastline regions (Figures 4e–4h). The MCS then grew locally over Jakarta and persisted until 2300 UTC (Figures 4i and 4j). This local MCS growth and its persistence could have been due to the intrusion of the northerly wind into the MCS, which was blocked by the southerly winds during the active phase of the CENS (Figure 2a) (cf. Wu et al. (2007); Mori et al. (2018)). In this case, the CENS acted as an inflow to the preceding MCS, causing it to grow. Further, the southerly wind blocked its movement, localizing the moisture increase and precipitation over Jakarta. At this stage, the massive rain-producing MCS was formed and brought very heavy rainfall over Jakarta, with the extreme rainfall event peaking at around 2100–2200 UTC. At 0300 UTC, the MCS propagated southward and dissipated.

In summary, we found that the extreme rainfall even in early January 2020 was associated with the formation of a massive rain-producing MCS due to the blocking of the CENS by southerly winds. The development of the MCS was closely related to the evolution of the CENS and its meridional nature. In addition, equatorial waves and MJO, which were present prior to and during the extreme event, may have also supported the development of the MCS over Jakarta during this period (Figure 3 and Figure S5 in Supporting Information S1) (cf. Mapes et al., 2006). Together, the combined effects of these factors provided favorable conditions for the development of the MCS over Jakarta.

6 Conclusions

We investigated the large-scale atmospheric driving mechanisms of the torrential rainfall over Jakarta in early January 2020. The key results of this study can be summarized as follows:
  1. The extreme precipitation event in early January 2020 was mainly caused by the convergence of moisture-rich air due to an unusual blocking of the CENS over Northwest Java by southerly winds induced by a cyclonic flow over the Indian Ocean.

  2. This blocking condition prevented moist air induced by the CENS from migrating further south, resulting in a mass accumulation that eventually caused vertical movement and the formation of massive heavy rains.

  3. The co-occurrence of convectively active phases of equatorial waves (mainly Kelvin, TD-type, and eastward propagating inertia-gravity waves) and MJO during the event also partly contributed to the high local moisture over the region by increasing low-level moisture flux convergence.

  4. Together, these large-scale dynamical drivers provided a conducive convective environment, which fostered the development of a massive rain-producing MCS and, consequently, extreme rainfall over the region.

Mori et al. (2018) showed that the meridional spatiotemporal variation of rainfall over Jakarta is controlled by the meridional extent of the CENS. Our result is consistent with this study, as the present results showed that the meridional extent of the CENS was prevented from migrating further south in the presence of southerly winds, resulting in a significant increase in local water vapor content over Jakarta. The current findings also suggest that the timing of the southerly winds intruding into Jakarta prior to the arrival of CENS was crucial for the development of a massive rain-producing MCS and, hence, extreme rainfall over Jakarta. Unlike the flooding event in February 2007 (Wu et al., 2007), the southerly winds intruded after the arrival of the CENS and were more confined to central Java Island, resulting in less intense precipitation over Jakarta (see Figure S5 in Supporting Information S1). Although the occurrence of the CENS in early January 2022 is not statistically unusual in terms of amplitude and duration (see Figure S9 in Supporting Information S1), the CENS had never before been blocked by southerly winds previously. It is also interesting to note that diurnal land convection can enhance precipitation over the large islands (Mori et al., 2018; Qian, 2008). Understanding the interaction between the CENS and thermally induced diurnal changes in the boundary-layer wind over Jakarta during this period, which can lead to enhanced localized convection, requires sensitivity simulations with a high-resolution mode. This will be the subject of future research.

As climates become warmer, models project an increased risk of more intense, more frequent, and longer-lasting extreme precipitation in the future (Robinson et al., 2021). Therefore, an improved understanding of the dynamical mechanisms responsible for such events may provide some insight into the best way to develop a reliable disaster mitigation system in Jakarta, which is vulnerable to localized flooding due to intense precipitation events (Ward et al., 2013). The results presented here could be leveraged to improve future predictions of extreme weather-driven hazards in Jakarta.

Acknowledgments

This work is supported by the U.S. Department of Energy Office of Science Biological and Environmental Research through the Regional and Global Model Analysis program area. The Pacific Northwest National Laboratory (PNNL) is operated for DOE by Battelle Memorial Institute under contract DE-AC05-76RL01830.

    Data Availability Statement

    All data used in this manuscript are publicly available. The ERA-5 reanalysis and SST datasets are publicly available at https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5 and https://psl.noaa.gov/data/gridded/data.ncep.reanalysis2.html. The NASA GPM data may be obtained from https://doi.org/10.5067/GPM/IMERGDF/DAY/06. Other data including in-situ hourly rainfall, radar reflectivity, blackbody temperature of clouds from Himawari-8 satellite, bandpass filtered data, and HYSPLIT backward trajectory data are available at https://doi.org/10.5281/zenodo.6568356. Daily global MJO indices are available at https://psl.noaa.gov/mjo/mjoindex/.

    Erratum

    In the originally published version of this article, the contributions of authors Risyanto and Siswanto were omitted. Their contributions have since been added, and the present version may be considered the authoritative version of record.