Global increase in tropical cyclone rain rate

The study addresses whether and how tropical cyclone (TC) rainfall rates have changed globally in recent decades, motivated by theoretical expectations that TC potential intensity and associated rainfall should increase as global mean sea surface temperatures rise. While TC-related precipitation poses major hazards (flooding and landslides), robust detection of global trends has been limited by data availability and homogeneity. Numerical models generally project increased TC rainfall under warming due to enhanced evaporation and atmospheric moisture, but observational validation at global scale has been lacking. This work aims to quantify global and basin-specific trends in TC rain rates over 1998–2016 using long-term, high temporal (3-hourly) and spatial (0.25°) resolution satellite precipitation records, and to relate observed trends to environmental changes in sea surface temperature (SST) and total precipitable water (TPW).

Prior theoretical and modeling studies (e.g., Emanuel 1987; Holland 1997) suggest increased TC potential intensity with warming SSTs. There is debate over changes in TC frequency and intensity due to observational limitations, though some analyses indicate increasing destructiveness. Modeling studies and assessments (IPCC AR5; Knutson et al. 2020) project TC rain rate increases on the order of 5–20%, with an aggregate estimate of ~14% for +2°C warming and lower confidence in Southern Hemisphere basins. Observational trend detection has been challenging: Lau and Zhou (2012) reported basin-dependent trends in total lifetime TC rain using coarse-resolution GPCP data, but acknowledged that such integral measures and low resolution are suboptimal for rain rate trends. More recent observational work links TC rainfall structure to both intensity and environmental conditions, with rain outside the core more sensitive to large-scale moisture and SST.

Data and period: The analysis covers 1998–2016 and includes all major TC basins except the South Atlantic (insufficient samples). A total of 81,242 satellite observations were analyzed, with ~3171–5315 samples per year (avg ~4232). Precipitation data are from TRMM 3B42 V7 (Jan 1998–Sep 2014) and the transitional GPM-based product (Sep 2014–Dec 2016), at 3-hourly, 0.25°×0.25° resolution. TC track, timing, and intensity information are from IBTRACS v4, prioritizing JTWC records for Pacific basins. Only cases where the 3B42 domain fully covers the storm area are included; best-track points beyond 44°N/S are excluded to avoid edge effects. Environmental variables (SST and TPW, t=0) are from the SHIPS developmental database (Reynolds SST and TPW averaged 0–500 km from the TC center). Defining TC rainfall regions: The total TC raining area is defined using a tropical cyclone precipitation feature (TCPF) framework. Precipitation features (PFs) are contiguous 3B42 pixels with rain rate >0.1 mm h−1; a PF is considered a TCPF if its centroid lies within 500 km of the TC center. This approach is contrasted with a simple 500 km radius truncation but adopted due to better TC size representation. Inner-core region is defined using the radius of maximum azimuthal rain rate (RMR), computed from azimuthal means (wavenumber-0 Fourier component). Sensitivity tests of inner-core definitions (100 km, RMR, 1.5×RMR, 2×RMR) showed similar qualitative results; 2×RMR is used to encompass the eyewall and inner convection. The rainband region is defined from 2×RMR to the TCPF outer boundary (~500 km). Only rain pixels >0.1 mm h−1 are used for averaging; zeros and missing data are excluded. Yearly averages are computed globally, by basin, and by intensity category (TD, TS, CAT1–CAT5). Time series and statistics: Trends are quantified with linear regression (time as predictor) and non-parametric Mann–Kendall tests for monotonic trends. Reported are regression slopes and R², Kendall’s tau and p-values, and Sen’s slope with 95% confidence bounds to assess robustness to outliers. A 5-year moving average is shown; for endpoints (1998–1999, 2015–2016) a moving-average smoothing model with q=2 is used to estimate tails. RMR yearly series are similarly analyzed. Basin delineation: ATL, ECPA, NWP, NIO, SIO, and SPA basins are analyzed separately. Intensity categories follow wind thresholds: TD (10–33 kt), TS (34–63 kt), and CAT1–CAT5 per Saffir–Simpson.

- Global TC rain rate increased significantly over 1998–2016: slope ≈ 0.027 mm h−1 yr−1 (≈ +1.3% yr−1), R²=0.90; Mann–Kendall indicates a trend at 99.9% confidence (Kendall’s tau=0.80). Sen’s slope 95% bounds: ~0.022–0.031 mm h−1 yr−1. - Basin-specific trends (mm h−1 yr−1; linear fit R²): NWP ≈ 0.037 (R²=0.84), ATL ≈ 0.0405 (R²=0.75), NIO ≈ 0.0317 (R²=0.51), ECPA ≈ 0.0178 (R²=0.46), SIO ≈ 0.019 (R²=0.59), SPA ≈ 0.0177 (R²=0.44). Trends are strongest in NWP and ATL; weaker in Southern Hemisphere basins. - By intensity category (global): positive trends across all categories (~0.015–0.043 mm h−1 yr−1). Strongest increases for TD, TS, CAT4, and CAT5 (CAT4 ~0.03; CAT5 ~0.04 mm h−1 yr−1). Smallest for CAT2 (~0.015 mm h−1 yr−1). - Regional differentiation within storms: Inner-core rain rate decreased while rainband rain rate increased. Inner-core decrease for CAT1–CAT5 ranges ~−0.011 to −0.096 mm h−1 yr−1 (~−1.5% yr−1 for CAT1–CAT5). Rainband increase ≈ +0.035 mm h−1 yr−1 (~+1.4% yr−1). Rainband trends are more homogeneous and statistically robust. - Possible inflection around 2003–2004 in inner-core and rainband time series suggests influence of decadal variability. - Environmental linkage: In the Northern Hemisphere, SST and TPW within 0–500 km increased significantly over the study period (robust regressions and Kendall tests). Correlations: rain rate vs SST r≈0.22; rain rate vs TPW r≈0.50. In Southern Hemisphere basins, SST and TPW show no significant trends (Kendall trends approach zero). - RMR shows no global trend overall; slight outward expansion in CAT1–CAT5 with low R² and confidence. - Over the 19-year period, the observed global increase in TC rain rate (~21% total) is larger than model-aggregated expectations scaled to the small observed SST change (~0.21°C), suggesting a dominant contribution from rainband rainfall increases associated with greater atmospheric moisture availability. - Storm intensity (maximum sustained wind) shows no significant trend over the study period, indicating environmental moisture/temperature, rather than intensity changes, primarily drives the rain rate increase.

Findings support the hypothesis that a warmer and moister environment enhances TC rain rates, with increases concentrated in the rainband region where environmental control (SST and TPW) is stronger than direct intensity control. The more pronounced increases in the Northern Hemisphere (especially NWP and ATL) align with observed increases in local SST and TPW, whereas weaker or negligible increases in the Southern Hemisphere are consistent with minimal environmental trends. The decrease in inner-core rain rate for hurricanes (CAT1–CAT5) contrasts with some model projections of increased inner-core rainfall within 100 km; this discrepancy may reflect structural changes not captured in models, dataset/definition differences, or multidecadal variability and warrants further investigation. The lack of significant trends in TC intensity suggests that environmental moisture and temperature increases can enhance TC rainfall independent of intensity changes, reinforcing the importance of large-scale conditions in modulating TC rainbands. Comparisons with prior studies highlight that coarse-resolution rainfall datasets are inadequate for rain rate trend detection, validating the need for high-resolution, long-duration satellite data. The increased TC rainfall rates imply heightened freshwater flooding risk, particularly in NWP and ATL, underscoring the need for improved hazard assessments and for evaluating climate model projections against observed structural (core vs rainband) responses.

Using nearly two decades of high-resolution satellite rainfall observations linked with best-track and environmental analyses, this study detects a significant global increase (~1.3% per year) in tropical cyclone rain rates from 1998 to 2016. The increase is strongest in the Northwestern Pacific and North Atlantic, occurs across all intensity categories, and is driven primarily by enhanced rainband rainfall, while inner-core rain rates decreased for hurricanes. Observed increases in local SST and TPW, especially in the Northern Hemisphere, are associated with the rainfall trends, whereas Southern Hemisphere basins show weaker trends consistent with minimal environmental changes. These results provide observational support for model-based expectations of increased TC rainfall under warming, identify rainband-dominated mechanisms, and offer benchmarks for evaluating climate models. Future work should investigate the causes of decreasing inner-core rain rates, potential roles of decadal variability, refine structural metrics (e.g., RMW vs RMR), and extend analyses with longer records and additional sensors.

- Temporal scope is limited to 19 years (1998–2016), potentially influenced by decadal variability (e.g., an apparent inflection around 2003–2004). - Inner-core definition uses RMR as a proxy for RMW due to limited global RMW data; RMR trends for stronger storms show low confidence. - Rain detection threshold (>0.1 mm h−1) chosen to mitigate retrieval contamination may affect light-rain representation. - Spatial coverage constraints exclude cases beyond 44°N/S and require full 3B42 domain coverage; South Atlantic excluded due to small sample size. - Potential inter-agency differences in best-track data, though JTWC prioritization was applied for Pacific basins. - Endpoint smoothing for 5-year moving averages uses a model-based approach, introducing minor methodological assumptions. - Environmental analyses indicate weak trends in the Southern Hemisphere, limiting detection power there; causality is inferred from correlations and trend co-variations rather than formal attribution.