Characteristics of rainfall distribution induced by tropical cyclones using GSMaP data over the Vietnam region

Tropical cyclones (TCs) play a key role in shaping global rainfall patterns (Khouakhi et al. 2017; Ray et al. 2022). While they can replenish water resources and benefit agriculture, they can also cause devastating floods and landslides, causing widespread damage to people and infrastructure. To understand and predict their impact, researchers worldwide employ diverse tools, including sophisticated numerical models, rigorous data analysis, simulations, and direct on-the-ground observations (Walsh et al. 2016; Lavender & McBride 2021; Trinh et al. 2023).

Based on a 25-year data set from 1970 to 2014, consisting of best-track data and daily rainfall records from 18,607 global rain gauges, researchers found that the most significant TC-induced rainfall totals occurred in East Asia (>400 mm/year), northeastern Australia (>200 mm/year), the southeastern United States, and along the coast of the Gulf of Mexico (100–150 mm/year) (Khouakhi et al. 2017). TC rainfall accounted for 30–50% of the average annual rainfall in northwestern Australia, southeastern China, the northern Philippines, Baja California, and Mexico. Seasonal evaluation by Dare et al. (2012) revealed that about 40–50% of TC-induced rain occurs along the western coast of Australia in the islands of the south Indian Ocean in the austral summer, and in East Asia and Mexico in the boreal summer and fall. Regarding extremes, maximum rainfall was found in East Asia, Australia, and North and Central America (Khouakhi et al. 2017). Stream flow data and simulated runoff data from ERA5 reanalysis spanning 1979–2012 reveal significant contributions of TC-induced rainfall to annual floods. TCs account for over 40% of annual floods in coastal East Asia and Australia and over 25% in North America. TCs contribute more than 70% in Eurasia, over 60% in North America, and over 50% in South America and South Africa (Wang et al. 2021). Rainfall stations have demonstrated that TC-induced rainfall in Taiwan, and southeastern China averages around 500 mm/year, constituting 20–40% of the annual rainfall total (Ren et al. 2006). In Australia, during the peak of the TC season (from January to March), the TC rainfall contributes approximately 40% to the rainfall total in coastal regions west of 120°E, while, in December, it comprises approximately 60–70% of the total rainfall in areas west of 115°E (Dare et al. 2012). In the Philippines, ground and satellite observations indicate that TC rainfall accounts for 6% of the annual rainfall in the south and 54% in the north (Bagtasa 2017).

In recent decades, numerous studies worldwide have investigated the characteristics and complexity of TC. Early TC research highlighted the intricate structure of TC precipitation, which is often analyzed from radar images and exhibits variability from case to case (Miller 1958; Frank 1977; Marks 1985; Burpee & Black 1989). TC rainfall is typically organized by spiral bands that converge toward the TC center, referred to as rainbands, which can deform upon landfall (Wexler 1947). The spatial distribution of TC rainfall includes axisymmetric and asymmetric components (Chen & Li 2004), as evidenced by radar images in studies by Marks (1985) for Hurricane Allen (1980) and Burpee & Black (1989) for Hurricanes Alicia (1983) and Elena (1985). These studies observed variations in rainfall patterns, with the maximum rain rate shifting between different quadrants relative to the TC center, and terrain features like coastal lines and mountains significantly influence TC asymmetric rainfall, especially during landfall. The spatial distribution of TC rainfall is significantly influenced by wind velocity, vertical wind shear, moisture supply, and surface characteristics. Faster TC motion concentrates heavy rainfall in narrower bands near the track (Lonfat et al. 2004), while vertical wind shear tilts and asymmetrically enhances precipitation downshear (Shapiro 1983). Furthermore, the moisture supply available to the TC significantly impacts the overall rainfall intensity. TCs over moisture-rich regions, such as warm ocean waters, can continue ingesting abundant moisture and fuel heavy precipitation (Zhang et al. 2013). Conversely, limited moisture supply can constrain rainfall amounts. Finally, surface characteristics, including land–ocean contrasts and topographic features, can modulate the spatial distribution of TC rainfall through processes like frictional convergence and orographic lifting (Lonfat et al. 2007; Yu et al. 2015). For instance, coastal regions and mountainous terrain often experience enhanced rainfall due to these effects.

In Vietnam, TC rainfall data from 138 stations (1979–2019) show varying impact, accounting for 10–20% of total annual rainfall in the northern and central regions, while remaining lower (less than 2%) in the southern region due to fewer TC occurrences (Pham-Thanh et al. 2020). The highest TC rainfall is recorded in the central region in July. Recent observation records (2018–2022) highlight significant TC rainfall contributions during major heavy rain events, ranging from 15 to 37%. TCs frequently interact with East Asian cold surges (CSs) over Vietnam, which dramatically enhances rainfall and causes extreme flooding events (Wang et al. 2015; Trinh et al. 2020; Gori et al. 2022). The complex interplay between TCs and CSs results in a significant modulation of moisture transport and convergence, leading to heavy precipitation over coastal Vietnam. Recent studies indicate that extreme rainfall events in Central Vietnam involve TC–CS interactions (Yokoi & Matsumoto 2008; Do et al. 2020; Wolf et al. 2020). As climate variability increases, improved forecasting of these compound events is crucial for disaster risk reduction in this vulnerable region.

Satellite-based precipitation estimates have proven invaluable for understanding TC structure and evolution, especially in data-sparse oceanic regions. The Tropical Rainfall Measuring Mission (TRMM) satellite's precipitation radar has been leveraged to reveal fine-scale structural details related to TC eyes, eyewalls (known as the ring of intense thunderstorms surrounding the eye of a TC), and rainbands. This reveals how vertical profiles vary by storm quadrant and environmental wind shear (Hence & Houze 2011). Chen et al. (2006) found that vertical wind shear over 5 m/s significantly impacted TC precipitation asymmetry, while TC motion was important in low-shear environments. Similarly, Gao et al. (2017) found that incorporating total precipitable water content, in addition to TC intensity, substantially improved model landfall precipitation forecasts. Analysis of global TC core precipitation generally shows agreement with a Gaussian distribution pattern, where precipitation intensity is highest at the center and decreases radially outward (Huang & Chen 2019). However, significant variability exists across different ocean basins due to factors such as environmental conditions, geographic features, and landmass configurations surrounding each basin. Despite the Gaussian model serving as a reference, the observed deviations highlight the complexity and diversity of TC precipitation structures globally. Other studies have utilized the high-resolution precipitation estimates from the Global Satellite Mapping of Precipitation (GSMaP) product (Kubota et al. 2020). Richardson et al. (2022) analyzed GSMaP data for North Pacific TCs, finding a good agreement with ground observations and revealing asymmetric outer rainband structures modulated by vertical wind shear. Examining typhoons in 2004, Lonfat et al. (2004) showed that rain area and volumetric rain content scaled with storm intensity, providing new metrics for monitoring TC strength. For landfalling TCs, blended GSMaP with gauge data improve quantitative precipitation forecasting (Richardson et al. 2022). Huang & Chen (2019) used GSMaP to study precipitation structures of recurring TCs in the Western North Pacific, detecting increased rainfall in the outer rainbands as storms recurved poleward. Beyond TCs, Aonashi et al. (2009) described the full GSMaP processing system that merges passive microwave and infrared satellite observations to produce high-resolution global precipitation maps. Evaluation studies by Miyato et al. (2021) demonstrated GSMaP's overall skill in capturing precipitation distribution and timing compared to ground radar and rain gauge networks across Japan. The fusion of TRMM's detailed structural observations and GSMaP's high-resolution rainfall estimates highlights the value of satellite platforms for both advancing TC's physical understanding and improving operational forecasts.

TCs and floods account for over 90% of natural disaster damages in Vietnam (MONRE 2021; Takagi et al. 2023). As climate conditions in Vietnam are projected to worsen by the end of the 21st century (Trinh-Tuan et al. 2019a; Tran et al. 2022; Tran-Anh et al. 2022, 2023), studying changes in TC rainfall distributions over Vietnam will become increasingly critical. Accurately forecasting TC rainfall remains challenging, especially as TCs approach landfall, due to complex interactions with terrain and weather systems that alter rainfall patterns (Cheung et al. 2018; Ray et al. 2022). Utilizing satellite data can help reconcile these discrepancies and improve process understanding. Satellite rainfall data provide extensive spatial coverage and improved accuracy to enhance understanding of TC rainfall structures over Vietnam (Trinh-Tuan et al. 2019b; Pham & Vu 2020).

This study leverages high-resolution precipitation estimates from the GSMaP data set to conduct a comprehensive investigation of TC rainfall distributions over the Vietnam East Sea (VES) region. A key focus is characterizing the spatial patterns and asymmetries in TC rainfall fields by categorizing them based on influential factors such as TC intensity, storm motion direction, landing location along the Vietnamese coastline, interactions with CS systems, and the evolution of rainfall distribution before and after landfall. The primary objectives are to systematically characterize the variability in TC rainfall features and their asymmetric structure across these influential factors, deepen our understanding of the mesoscale precipitation processes governing the evolution and distribution of rainfall in landfalling TCs over the complex coastal environment of Vietnam, and utilize the observational insights to enhance operational forecasting capabilities for more accurate predictions of TC rainfall distributions and associated hazards that can severely impact Vietnam. The insights gained from this multi-faceted analysis will provide valuable information to support TC and flood forecasting efforts in the VES, as well as to assess the potential rainfall impacts and hazards associated with landfalling systems affecting Vietnam.

The best-track data of TCs are used from the Regional Specialized Meteorological (RSMC) of Tokyo – Typhoon Center during the period 2000–2020 (Kunitsugu 2012). The historical TC records provide information on the position of the TC center (LAT, LONG), maximum sustained winds (V_max), minimum central pressure (P_min), the radius of maximum winds, TC track name, and 6-hourly temporal resolutions (0000UTC, 0600UTC, 1200UTC, and 1800UTC). This study focuses on TCs that originated in the central Pacific Ocean and tracked westward toward the VES region during the period 2000–2020. Specifically, we consider TCs that passed through the domain bounded by 5–25°N latitude and 145–120°E longitude. This domain encompasses the entire westward journey of the TCs from their formation in the central Pacific until they reached the VES and coastal areas of Vietnam. Within this domain, a total of 177 TCs were identified and tracked during the 2000–2020 period. These 177 TCs provided 5,044 instantaneous precipitation observations associated with their passage through the domain.

TC intensity is typically classified internationally using the Saffir–Simpson hurricane wind scale, with storms categorized into five levels (CAT 1–5) based on their maximum sustained wind speeds. However, in Vietnam, TC intensity is traditionally divided into levels 8–12 and greater than 12 on the Beaufort wind scale. To ensure consistency with previous studies (Camargo & Sobel 2005; Wang & Zhou 2008) and facilitate comparison with international data, we categorized TC intensity into three classes based on maximum sustained wind speeds: Tropical Storm (TS) (34–48 knots), Severe Tropical Storm (STS) (48–63 knots), and Typhoon (TYP) (over 63 knots), aligning with the international Saffir–Simpson scale.

The GSMaP is a global rain rate data set that provides hourly measurements with a resolution of 0.1° × 0.1°, encompassing the expansive region between 60°N and 60°S. It is created through the collaboration of multiple satellites, including Global Precipitation Measurement satellites and passive microwave and infrared radiometers on geostationary satellites, developed by Japanese scientists under the sponsorship of the Japan Science and Technology Agency (JST) and the Japan Aerospace Exploration Agency (JAXA) (Kubota et al. 2007; Aonashi et al. 2009; Ushio et al. 2009). In this study, we utilized the standard Version 6 of GSMaP data, known as GSMaP_MKV, spanning the period 2000–2020. The GSMaP_MKV data set is a bias-corrected product that employs a Kalman filter model to refine precipitation rate estimates (Ushio et al. 2009). This enhancement is achieved by leveraging atmospheric moving vectors derived from successive infrared images. While this global data set offers rainfall data for any location worldwide, the GSMaP_MKV data are limited to the period from March 2000 to the present. It provides a maximum spatial resolution of 0.1 ° ×0.1° (approximately 10 km × 10 km) in its highest resolution product. The effectiveness of GSMaP in capturing rainfall patterns has been validated in Vietnam and its coastal areas (Nodzu et al. 2019; Trinh-Tuan et al. 2019b). GSMaP enables detailed tracking and analysis of rainfall patterns, trends, and climate variations due to its fine resolution and coverage. The GSMaP data can be freely downloaded from the website of JAXA: http://sharaku.eorc.jaxa.jp/GSMaP/.

ERA5 represents the fifth major global atmospheric reanalysis developed by the European Center for Medium-Range Weather Forecasts (ECMWF) (Hersbach et al. 2020). With 0.25° spatial resolution and hourly outputs, ERA5 incorporates observations, models, and data assimilation to generate consistent estimates of historical weather variables like rainfall. Here, ERA5 wind data were utilized over Vietnam from 2001 to 2020 to complement the GSMaP rainfall analysis in cases of CS–TC interaction.

In this study, we categorized TC rainfall into three intensity types using specific thresholds: light rain (>0.5 mm/h), moderate rain (1–5 mm/h), and heavy rain (>5 mm/h). The light rain threshold was chosen to capture even minimal precipitation, ensuring comprehensive coverage detectable by sensors. Moderate rain represents the predominant intensity range observed during TC events in our study area, encompassing moderate rainfall rates commonly associated with TC activity. Meanwhile, the heavy rain threshold, exceeding the 95th percentile level in the VES during TC passages, highlights the most extreme precipitation events, indicating the potential for hazardous rainfall conditions. Comparing the high-resolution patterns seen in GSMaP observations and robust climatological trends from the ERA5 reanalysis facilitates deeper insights into TC precipitation variability.

The CS refers to a periodic southward advancement of cold, midlatitude air along the eastern flank of the Tibetan Plateau in the lower troposphere, primarily occurring during the November–March period. Numerous studies have shed light on the spatio-temporal patterns of CSs. The CS northerly wind signal first appears near 40°N in northern China associated with a trough–ridge system aloft (Wu & Chan 1997; Chen et al. 2015; Perşoiu et al. 2019), following the strengthening of the Siberian High (Zhang et al. 1997). Subsequently, the CS signal progresses southward into the tropics within a few days (Wu & Chan 1995; Compo et al. 1999). The CS's arrival is characterized by a pronounced increase in surface pressure, robust northerly winds, and a decrease in temperature (Perşoiu et al. 2019). After reaching the VES, the CS enhances convective activity over the VES, Borneo Island, and at times, the Maritime Continent in the Southern Hemisphere (Chang et al. 2005). Consequently, the CS also impacts the dynamic structure of TCs as well as the distribution of rainfall induced by TCs (Cheung et al. 2018).

To investigate the influence of CS on TC rainfall distributions, this study adopts specific criteria to identify TCs interacting with CS events, as defined by Chang et al. (2005) and Yokoi & Matsumoto (2008). The center of the TC must fall within a 500 km radius of the longitudinal line representing the cold air mass boundary. Two thresholds are used: (1) CS15 based on the 15°N line (3-day mean) and (2) CS20 on the 20°N line (1-day mean). Specifically, CS15 requires a 925-hPa meridional wind anomaly of less than −3.1 m/s averaged from 110°E to 120°E at 20°N. CS20 requires 925-hPa meridional winds to be less than −8 m/s averaged from 110°E to 117.5°E at 15°N. The CS15 threshold delineates a more robust CS, while CS20 captures all cold air masses. TCs that do not meet either criterion are considered neutral cases, not influenced by CSs. This integrated framework facilitates the examination of rainfall distribution when TC interacts with cold air masses.

Analysis of the full radial distribution of rainfall (Figure 3(b)) reveals peak intensities within 10 km of the eyewall for all TCs, representing the zone of strongest convection and inflow (Houze 2010). Moving radially outward from the eyewall, rainfall intensity gradually diminishes between 10 and 50 km before exhibiting secondary maxima around 60–70 km from the storm center. These secondary maxima highlight the presence of outer rainband convection, a significant structural feature of TC precipitation distributions. Outer rainbands are organized, spiraling bands of deep convection that extend outward from the TC's inner core region, typically observed between 100 and 300 km from the center. While the eyewall remains the primary precipitation feature, outer rainbands can substantially contribute to the TC's total rainfall output and often produce intense precipitation rates. The formation and maintenance of outer rainband structures result from several dynamical mechanisms, including the upscale growth of convective elements through vortex stretching and vorticity organization (Qin et al. 2021), interactions between the TC's vortex and environmental factors like vertical wind shear and upper-level troughs inducing asymmetries (Ying & Zhang 2012), and frictional convergence near the TC's outer circulation, enhancing convection (Martinez et al. 2020). Distinctions between TS, STS, and TYP are most apparent within 300 km, reflecting varying precipitation patterns across TC intensities. As the TC intensity increases, so do the rain rates in the critical eyewall and rainband regions, highlighting the connection between precipitation processes and storm dynamics (Hence & Houze 2011). However, the peak outer band location seems insensitive to intensity, contrasting some studies (Didlake & Houze 2013).

Despite slightly lower fractions of intense rain rates compared to R1/R2, Figure 4(b) makes it clear that the extent of extreme rainfall occurring during TC events is greater in the southern R4 region versus northern areas (R1–R3). This disparity is likely due to higher atmospheric moisture content from proximity to warm ocean waters, weaker vertical wind shear favoring better-organized convection, and flatter coastal terrain in southern regions compared to the north. Southern Vietnam's location allows for longer TC overland trajectories, sustained moisture uptake, and potential enhancement from interactions with moisture-laden monsoonal flows, all contributing to higher rainfall accumulations during TC passages toward lower latitudes. Overall, northern Vietnam faces higher cyclone occurrence frequency, but southern regions experience greater extreme precipitation magnitudes.

In CS15 cases, rainfall concentrates in the first/second quadrants, while CS20 and neutral TCs peak in the second quadrant. CS15 shows the most intense pattern, particularly within the inner core. The enhanced convection likely arises from the upward motion and stretching of the storm system due to the positive rotation of winds brought about by the CS (Heo et al. 2018; Tan et al. 2023). Tighter meridional gradients in CS15 versus CS20 (Chang et al. 2005; Yokoi & Matsumoto 2008) support stronger interactions. The CS appears to redirect and intensify rainfall within the inner core of TCs, as depicted in Figure 7(b).

Overall, it is evident that TC rainfall is significantly influenced by interactions with environmental air masses like CSs. The structural changes visible in high-resolution GSMaP data provide new observational evidence elucidating the underlying dynamics modulating TC morphology.

This study analyzed TC rainfall distributions over the Vietnam region from 2000 to 2020 using GSMaP precipitation and ERA5 reanalysis data. Rainfall features were characterized across categorizing by TC intensities, motion vectors, landfall regions, and interactions with CS systems.

Key findings highlight the scaling of total rainfall coverage with TC intensity, while peak rain rates increasingly concentrate around the eyewall region at higher intensities. Rainfall distribution shows asymmetry, with higher precipitation observed in the front-right quadrant due to the combined effects of cyclonic flow and storm motion vectors, enhancing uplift and moisture convergence. However, distinct regional variations are observed, which are tied to geographical influences and climatological factors. Northern Vietnam experiences more frequent TC occurrences, but southern regions exhibit heavier extreme rainfall events. This disparity is attributed to higher atmospheric moisture content, weaker vertical wind shear favoring better-organized convection, and flatter coastal terrain in southern areas compared to the mountainous north. Furthermore, southern Vietnam's location allows for longer overland TC trajectories, sustained moisture uptake, and potential enhancement from interactions with moisture-laden monsoonal flows, collectively contributing to higher rainfall accumulations during TC passages toward lower latitudes.

Rainfall distributions undergo evolution significantly before and after landfall. Pre-landfall, rainfall concentrates in the front-right quadrant, resembling patterns over the ocean due to vertical wind shear and motion vector interactions. Post-landfall, moderate rain rates become scattered across quadrants until dissipating as the storm decays inland. However, even 24 h after landfall, rainfall exceeding 1 mm/h still covers a substantial area, highlighting the role of Vietnam's mountainous terrain in sustaining convection through orographic lifting despite dissipating dynamics. Notably, interactions with CS air masses substantially intensify convection within the TC inner core. GSMaP observations reveal that strong CS cases (CS15) exhibit enhanced rainfall rates and a concentration of precipitation in the first/second quadrants, while weaker CS cases (CS20) and neutral TCs peak in the second quadrant. The structural changes visible at fine scales in GSMaP data provide new observational evidence elucidating the dynamics modulating TC morphology due to synoptic-scale interactions.

This integrated analysis combining high-resolution satellite rainfall estimates and reanalysis climatology provides an improved understanding of mesoscale precipitation processes in TCs making landfall in complex coastal environments like Vietnam. Characterizing structural sensitivities and regional variations aids in forecasting these high-impact events. The findings offer insights into mesoscale processes governing TC precipitation, such as outer rainband formation and evolution, which can be incorporated into operational forecast models to enhance rainfall predictions. Additionally, the analysis can inform tailored forecast products, risk assessments, and hazard mapping specific to the region, ultimately improving disaster preparedness and resilience against TC-related rainfall hazards. While the methodology offers a framework applicable to other TC regions worldwide, acknowledging potential limitations is crucial. Factors such as environmental conditions, geographical influences, data availability, climatological differences, and temporal changes may require adjustments. Region-specific validations and local expertise incorporation are recommended for appropriate application in new TC regions.

The authors acknowledge JAXA for providing GSMaP data. TC data were sourced from RSMC Tokyo – Typhoon Center, and ERA5 data were provided by the ECMWF. The authors thank the editor and three reviewers for their valuable comments and suggestions, which significantly enhanced the quality of this manuscript.

N.P.T.T. and L.T.-T. conceptualized the research. L.T.-T., Q.T.-A., T.D.T., and N.P.-Q. processed the data and conducted the analysis. All authors discussed the results and contributed to the writing of the manuscript.

This study was conducted with full support from Vietnam's National Project 562 (Grant Number ĐTĐL.CN-59/21), which is funded by the Vietnam Ministry of Science and Technology.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.