Unveiling the dynamics of sequential extreme precipitation-heatwave compounds in China

Global warming has increased the frequency and severity of climate extremes worldwide, notably heatwaves and extreme precipitation, which cause significant human and ecological impacts. China has recently suffered large-scale flood losses and widespread heatwave impacts. The IPCC defines compound events as combinations of multiple drivers or hazards in space and/or time whose impacts can exceed those of single events. Sequential extreme precipitation–heatwave compound events (SEPHCE) are of growing concern because of their cascading impacts. Previous studies have emphasized such compound events in hotspots like the United States and Japan, but rapid urbanization and climate change in China since the 1970s necessitate a comprehensive assessment of SEPHCE to inform disaster risk reduction and sustainable development. Beyond standard indicators (frequency, duration, intensity), understanding the seasonal timing (start and end dates) of SEPHCE is crucial because earlier onsets and delayed cessations can heighten societal risks. This study identifies SEPHCE across China from 1975 to 2020 using station observations, analyzing their frequency, duration, timing, spatial patterns, and the joint intensity distribution of extreme precipitation and heatwaves, to support targeted risk management and climate adaptation.

The compound-event concept, introduced in IPCC AR5 and expanded in AR6, includes antecedent, multivariate, temporal, and spatial compounds whose impacts often exceed those of single hazards. Studies report global increases in heatwaves and extreme precipitation since the 1950s and a fivefold rise in climate change-induced hazards in recent decades. Case studies include the 2018 western Japan flood–heatwave sequence and frequent flood events following heatwaves in the central United States, with implications for food security. Research has examined changes in the probability and characteristics of compound heatwave–flood events, noting a tendency for heavy rainfall to occur shortly after shorter, hotter heatwaves. In China, rapid urbanization has exacerbated extremes, and recent work reports detectable increases in sequential flood–heatwave events, particularly in southeastern and western China. Heatwave metrics have been standardized in various regions (e.g., 90th percentile exceedance for multiple consecutive days), and numerous indices capture frequency, duration, and intensity. Given observations of earlier arrivals of extreme weather and record-setting events, the timing of onset and cessation is increasingly recognized as critical for risk management.

Data: Daily maximum temperature, minimum temperature, and precipitation for 1975–2020 were obtained from the National Meteorological Information Center (CMA), initially covering 2479 stations. Quality control included checks for extremes, internal/temporal/spatial consistency; missing values were interpolated and constrained to <5% per year. The analysis focuses on extended summer (May–September). After QC, 1929 stations remained. Definitions: Heatwaves were defined following Chen et al. and Della-Marta et al. as periods of at least 3 consecutive days where daily Tmax and/or Tmin exceed the 90th percentile threshold computed within a 15-day moving window, using a 30-year reference (1975–2004). Mean thresholds across stations were 34.3 °C (Tmax) and 23.5 °C (Tmin). Extreme precipitation events were defined by a relative threshold: daily precipitation exceeding the 95th percentile of all rain days during the same reference period, with an average threshold of 44.5 mm day−1. SEPHCE were defined when an extreme precipitation event is followed by a heatwave within a specified interval (7, 5, 3, or 1 days). To avoid double counting, if multiple events occur in sequence, only the first extreme precipitation and the first subsequent heatwave were paired. Metrics and spatiotemporal analysis: SEPHCE frequency is the total annual count across stations; duration is the cumulative duration of the extreme precipitation and heatwave components within SEPHCE. Start (earliest onset) and end (latest cessation) times per year were computed at each station. Annual station metrics were interpolated to a 0.1° × 0.1° grid to derive spatial patterns and long-term totals. Trends were estimated using the Theil–Sen median slope and significance assessed with the Mann–Kendall test. Joint intensity and return periods: Intensities were quantified as follows—heatwave intensity: exceedance of Tmax/Tmin above thresholds; extreme precipitation intensity: cumulative precipitation during the extreme-precipitation component within each compound event. Five candidate univariate distributions were fitted to each margin; the generalized extreme value (GEV) distribution best fit extreme precipitation intensity and the logarithmic distribution best fit heatwave intensity (by maximum likelihood). Four copulas (Gaussian, Gumbel, Clayton, Frank) were tested; the Gaussian copula was selected (OLS test), with estimated correlation parameter rhohat = −0.049. Joint return periods for events where precipitation intensity ≥ p and heatwave intensity ≥ h were derived from the joint CDF using the selected marginals and copula, with the return period defined as the reciprocal of joint exceedance probability multiplied by the average interarrival time estimated from observations.

- National frequency of SEPHCE increased significantly from 1975–2020, with overall rates of about 17, 67, 122, and 176 occurrences per decade for 1-, 3-, 5-, and 7-day intervals, respectively. A pronounced post-1993 acceleration yielded higher growth rates of 30, 107, 179, and 258 occurrences per decade for the same intervals, with peak annual counts reaching 149 (1-day), 494 (3-day), 826 (5-day), and 1175 (7-day). - Durations increased rapidly nationwide: 113, 424, 765, and 1117 days per decade for 1-, 3-, 5-, and 7-day intervals, respectively. The 7-day interval duration peaked at 7791 days cumulatively in 2018. - Spatial patterns show higher SEPHCE frequency and longer durations concentrated in southwestern and southeastern coastal China, and lower values in northern regions. Trends in frequency and duration increased faster in the southwest and southern coasts. Statistically significant increasing trends are widespread at 7- and 5-day intervals; trends at 3- and 1-day intervals are generally weaker or not significant. - Start times advanced and end times were delayed across intervals, lengthening the SEPHCE season. Overall, starts advanced by about 4.4 days/decade and ends delayed by about 3.3 days/decade. By interval, start advances were ~9.3, 4.2, 2.5, and 1.7 days/decade (1-, 3-, 5-, 7-day), while end delays were ~6.2, 3.7, 1.8, and 1.6 days/decade, respectively. On average, SEPHCE began near May 3 and ended near September 25, with widening seasonality over time. - Joint intensity analysis using a Gaussian copula indicates a weak negative dependence between extreme precipitation intensity and heatwave intensity (rhohat ≈ −0.049). Joint return periods increase with increases in either or both intensities; contour plots show increasingly sparse isopleths at higher intensities, consistent with rarer high-intensity joint events.

The study demonstrates that SEPHCE in China have become more frequent, longer, and temporally extended (earlier onset and later cessation) since 1975, with a marked acceleration after 1993. The burden is uneven geographically, with hotspots in southwestern and southeastern coastal regions—areas with complex terrain, fragile ecosystems, and, in places, less economic resilience—implicating equity concerns in climate risk. Short-interval SEPHCE (e.g., within 1–3 days), though rarer and spatially more limited, show rapid increases in frequency and duration over recent decades, suggesting escalating risk from closely spaced hazard sequences. The weak negative correlation between precipitation and heatwave intensities implies limited linear dependence, yet both intensities contribute to longer joint return periods when elevated. Physical processes potentially linking sequential flood–heatwave hazards in southern China may involve tropical intraseasonal variability, subtropical flow meanders, and typhoon-related diabatic heating, but these connections require further investigation. The findings support prioritizing integrated planning, early warning, and adaptation that considers compounding sequences rather than isolated hazards.

This work provides the first comprehensive, nationwide assessment of sequential extreme precipitation–heatwave compound events (SEPHCE) in China over 1975–2020, quantifying increases in frequency, duration, temporal extent, and joint intensity characteristics. Results highlight substantial post-1993 accelerations and distinct regional hotspots in the southwest and southeastern coasts. The methodological framework—combining station-based detection, timing metrics, robust trend testing, and copula-based joint return period estimation—can be generalized to other regions given suitable long-term datasets. Policymakers should integrate SEPHCE risks into disaster risk reduction, urban planning, infrastructure design, and public health preparedness, with particular attention to vulnerable regions and the growing prominence of short-interval sequences. Future research should improve understanding of the physical linkages driving SEPHCE, extend analyses to other seasons and regions, refine joint modeling of intensities and dependencies, and assess impacts on critical sectors (e.g., agriculture, water resources, health) to inform targeted adaptation.

- Geographic scope is limited to mainland China and the extended summer season (May–September), which may limit generalizability to other regions or seasons. - The last 10-year ratio analysis uses only 2015–2020 due to data availability, potentially affecting comparability with earlier decades. - Station coverage reduced from 2479 to 1929 after quality control; residual missing data were interpolated (<5% per year), which may introduce uncertainties. - Trends at shorter intervals (1- and 3-day) in frequency and duration are weaker or not statistically significant in many areas. - Suggested physical mechanisms (e.g., intraseasonal variability, typhoon effects) are hypothesized but not conclusively established; further process-based studies are needed.