Slower-decaying tropical cyclones produce heavier precipitation over China

Tropical cyclones (TCs) cause severe impacts through heavy precipitation, storm surges, and strong winds. After landfall, TC intensity usually decays rapidly due to diminished moisture supply and increased surface friction; thus, impacts are often confined to a few days and limited inland areas. With a warming climate, higher atmospheric moisture content can allow stronger TCs that decay more slowly, potentially increasing damages as losses scale with intensity. Slower-decaying TCs can maintain destructive winds and rainfall for longer and penetrate farther inland, elevating risks. Recent studies reported significant slowdowns in decay of landfalling TCs in the North Atlantic and western North Pacific, but large uncertainties remain due to sensitivity to region, period, dataset, and methods. Proposed drivers include warmer sea surface temperatures (SSTs) increasing storm moisture content versus the role of track-dependent moisture supply near landfall, as well as environmental factors like soil moisture, low-level vorticity, and upper-level divergence. While TC-related changes such as higher rainfall rates and slower translation speeds are linked to elevated flood risk, the relationship between TC decay rate and TC-induced precipitation has received less attention. Prior regional evidence (e.g., Guangxi, China) suggested more stations with rainstorms during slower-decay events, but broader observational evidence remains limited. This study evaluates long-term trends and temporal evolution of post-landfall decay timescale (τ) for 141 landfalling TCs over China (1967–2018), using satellite-era best tracks to reduce inhomogeneity. We assess relationships between τ and SSTs as well as TC characteristics (landfall location, intensity, translation speed), investigate regional differences between eastern (EC) and southern (SC) mainland China, and quantify impacts of changing decay on TC-induced precipitation using 3-hourly data (1979–2018).

Prior work indicates a slowdown of post-landfall decay in the North Atlantic and WNP, though estimates vary by region, period, dataset, and methodology, underscoring uncertainty (e.g., Chan et al.). Proposed mechanisms differ: Li and Chakraborty attributed slower decay to higher storm moisture content linked to warmer SSTs, whereas Chan et al. emphasized the effective moisture supply area governed by storm track upon landfall. Song et al. highlighted roles of soil moisture, low-level vorticity, and upper-level divergence in altering decay timescales. Regarding precipitation, studies documented increased TC rain rates and slower translation speeds contributing to flood risk; Lu et al. found slower-decay TCs impacted more rainstorm stations in Guangxi (1981–2001). However, comprehensive observational evidence directly linking decay-rate changes to precipitation changes remained limited, motivating this study.

Data and period: TC best-track data from IBTrACS v4.0 (CMA 2-minute maximum sustained winds; 6-hourly for 1967–2018 and 3-hourly for 1979–2018). ERA5 reanalysis (0.25°; monthly and hourly) for environmental fields (SST, winds, humidity), with vertical wind shear defined as 200–850 hPa wind difference and 500-hPa winds as steering flow. Gridded precipitation: CMFD 0.1° × 0.1°, 3-hourly, 1979–2018. Event selection for decay analysis: Landfalling TCs during 1967–2018 meeting four criteria: (1) intensity at last ocean point before landfall ≥24.5 m s−1; (2) survived >24 h after landfall (≥4 consecutive 6-hourly inland records); (3) no intensity increase after landfall; (4) no extratropical transition during first day after landfall. Initially 148 events; 7 outliers with τ > mean + 2σ were excluded, leaving 141 events. Decay timescale τ: For each TC, post-landfall intensity I(t) over 24 h modeled as exponential decay I(t) = I(0) e^(−t/τ). Larger τ indicates slower decay. Time series of annual mean τ smoothed with double 3-year smoothing; trends by OLS; uncertainties shown as ±1 s.e.m. Sensitivity checks included unsmoothed series and inclusion of outliers. Predictors and regions: Examined associations between τ and June–September SSTs in three oceanic regions (genesis, development, pre-landfall), landfall longitude/latitude, landfall intensity, translation speed over first four inland positions (v_t), and coastline-perpendicular component (v_t sinα). China’s coast divided into eastern mainland China (EC) and southern mainland China (SC) for regional contrasts. Genesis and environmental diagnostics: Analyzed trends in TC genesis position and correlations with landfall location. Composite differences between 10 years with largest versus smallest landfall-center longitudes assessed dynamic/thermodynamic environments (850-hPa vorticity, 500-hPa vertical motion, 600-hPa RH, vertical wind shear, SST, integrated water vapor and vapor flux, soil moisture). Genesis Potential Index (GPI) computed (Emanuel & Nolan) using η (850-hPa absolute vorticity), 600-hPa RH, potential intensity V_pot (Bister & Emanuel), and vertical wind shear. TC-induced precipitation identification: Employed Objective Synoptic Analysis Technique (OSAT) to isolate TC-induced precipitation at 3-hourly steps within 48 h post-landfall for each event (1979–2018). For each timestep, computed P_mean and P_max over affected grids; aggregated to P_mean24, P_max24, P_mean48, P_max48. Change rates computed as 100% × (OLS slope / mean). Total amounts PT_mean24, PT_max24, PT_mean48, PT_max48 computed as grid-averaged and maximum accumulated precipitation within 24/48 h over all affected grids. Restricted-area metrics (re-PT_mean24, re-PT_max24) required precipitation at all timesteps within 24 h. Attribution of τ increase: Relative contributions to τ increase between 1967–1992 and 1993–2018 estimated by holding either regional proportions (EC/SC) or regional τ values constant to separate effects of SST-driven τ changes versus shifts in landfall location, with residual labeled as “other factors” (e.g., landfall intensity).

- Significant slowdown of post-landfall decay: Annual mean τ increased by 1.6 h per decade (p < 0.01), amounting to a 45% increase from 1967 to 2018. Using unsmoothed series or including outliers, trends remained significant (1.9 and 2.3 h per decade, respectively). The fraction of landfall intensity retained after 24 h rose from 34% (1967) to 47% (2018). - Regional contrast (EC vs SC): τ is consistently larger in EC than SC. Mean τ (1967–2018): EC 34.04 h vs SC 25.90 h (significant). Subperiods: 1967–1992 EC 33.29 h vs SC 25.53 h; 1993–2018 EC 34.58 h vs SC 26.27 h (significant). The share of EC landfalls increased from 46.15% to 53.95%. - Associations with environmental and track factors: • SST: τ correlates positively with June–September SSTs, strongest over the pre-landfall region (r = 0.38, p < 0.01); SST trend there is +0.13 K/decade (1967–2018). SST, integrated water vapor, and τ are significantly correlated. • Landfall location: Longitude centroid shows an increasing trend (+0.23°/decade, p < 0.05). τ correlates with landfall longitude (r = 0.34, p < 0.05); latitude trend and correlation with τ are not significant. • Landfall intensity: Weak positive relation with τ (r = 0.27, p = 0.06) and an increasing trend in landfall intensity (+0.40 m s−1 per decade, p < 0.05). • Translation speeds: No significant relationship between τ and v_t (r = 0.02, p = 0.86) or v_t sinα (r = 0.27, p = 0.06); no significant trends in these speeds. - Drivers of increased τ (1993–2018 vs 1967–1992): Estimated relative contributions—SST warming 60.6%, eastward shift of landfall locations 37.0%, other factors (incl. intensity) 2.4%. - Genesis and moisture context: Genesis longitude shifted eastward (+0.60°/decade). Landfall longitude and latitude correlate with genesis longitude (r = 0.40 and 0.43). EC-landfalling TCs originate farther northeast (mean 138.5°E, 15.9°N) than SC (131.5°E, 14.1°N), have slightly longer over-sea duration (181.6 h vs 170.9 h), and higher 500-hPa specific humidity at landfall (significant). EC cases also exhibit higher coastline-perpendicular speed, but landfall intensities do not differ significantly between EC and SC. - Favorable environments during years with more easterly landfall: Positive low-level vorticity anomalies, enhanced ascent (negative 500-hPa ω), weaker vertical wind shear, warmer SSTs, higher midlevel humidity and integrated water vapor, and higher soil moisture in EC—all conducive to sustaining storm structure after landfall. GPI anomalies shift genesis east/north, with steering flows favoring EC landfalls. - Precipitation impacts of slower decay (1979–2018; 104 events): • Change rates of P_mean48 and P_max48 increase significantly with τ percentile (r = 0.30 and 0.36; p < 0.01). PDFs show slower-decay TCs (>70th τ) have significantly higher probability of positive change rates than faster-decay TCs (<30th τ), especially within 48 h. • Total TC-induced precipitation shows significant upward trends: PT_mean24 +2.0 mm/decade (p < 0.01); PT_max24 +23.0 mm/decade (p < 0.01); PT_mean48 +1.9 mm/decade (p < 0.01); PT_max48 +24.2 mm/decade (p < 0.05). All correlate positively with τ (r = 0.64, 0.76, 0.67, 0.70; p < 0.01). • Restricted-area totals also increase (re-PT_mean24 +13.1 mm/decade; re-PT_max24 +16.7 mm/decade) and correlate strongly with τ (r = 0.69, 0.82; p < 0.01). - Implications: Slower decay extends storm duration and intensity over land, elevating flood risks in China’s coastal regions.

The study addresses whether landfalling TCs in China are decaying more slowly and how this affects precipitation. Robust statistical evidence indicates a significant slowdown in post-landfall decay since 1967, with results consistent across smoothing choices and after outlier tests. The observed increase in τ is linked primarily to warmer SSTs in adjacent oceanic regions and to an eastward shift in landfall locations that steer more storms into EC, where environmental conditions (enhanced vorticity, ascent, humidity, weaker shear, higher soil moisture) favor persistence after landfall. Eastward shifts in genesis locations and steering flows further support these track changes. The positive association between τ and multiple precipitation metrics demonstrates that slower-decaying TCs more often maintain or amplify rainfall intensity after landfall and yield larger total precipitation within 24–48 h. This mechanistic link aligns with enhanced storm moisture availability due to warmer SSTs and longer over-sea durations prior to landfall, translating into increased flood risk. While translation speed changes did not significantly explain τ trends over China during the study period, interactions with slower decay could exacerbate rainfall totals when and where translation slows. The study also situates its findings within broader uncertainties surrounding regional variability, dataset inhomogeneities, and natural climate variability (e.g., ENSO, PDO, monsoon), which can modulate interannual changes in decay and tracks.

This work provides regional-scale observational evidence that landfalling tropical cyclones over China have decayed more slowly since 1967, with τ increasing by 45%. The slowdown is primarily driven by warming SSTs near pre-landfall regions and an eastward shift of landfall locations that increase the share of storms striking eastern China, where environmental conditions favor post-landfall persistence. Slower decay is associated with higher likelihood of sustained or increasing rainfall intensity within 48 h of landfall and with significant increases in total TC-induced precipitation, implying heightened flood risk for coastal regions. These findings underscore the need to incorporate slower post-landfall decay into hazard assessments and to strengthen flood risk management and adaptation strategies along China’s coasts. Future research should: (1) disentangle contributions of anthropogenic forcing versus natural variability to decay trends; (2) investigate interactions among SST patterns, landfall location shifts, and storm intensity; (3) leverage high-resolution reanalyses and modeling to reduce uncertainties in best-track intensities and to elucidate physical mechanisms; and (4) assess compound effects with storm surge, antecedent soil moisture, and urbanization on flood outcomes.

- Data uncertainties: Best-track intensity estimates can vary over time due to evolving observation techniques; weak/short-lived TCs over open ocean may be underdetected. The study mitigates this by focusing on landfalling TCs, imposing a minimum landfall intensity (≥24.5 m s−1), requiring ≥24 h survival post-landfall, and excluding outliers, but residual uncertainty remains. - Sample size and sensitivity: Despite relatively large regional sample (141 events), results can still be sensitive to event selection and processing choices; individual storm variability is high. - Model form and window: The exponential decay model is applied only to the first 24 h post-landfall; behavior beyond 24 h may not be captured by a single-parameter decay. - Precipitation analysis period: TC-induced precipitation analyses are limited to 1979–2018 due to data availability; identification via OSAT, while objective, carries methodological assumptions. - Attribution: The study does not attribute the detected slowdown to anthropogenic forcing versus natural variability; further dedicated attribution work is needed. - Translation speed effects: Although not significant for τ trends here, potential interactions with translation speed could modulate rainfall impacts and warrant further study.