The Clausius-Clapeyron (C-C) relation predicts a 7% per °C increase in atmospheric moisture-holding capacity, which should also apply to extreme precipitation intensity. However, observational studies have shown that short-duration extreme precipitation often increases beyond the C-C rate (super C-C scaling), particularly in convective events. This study addresses a gap in understanding how synoptic patterns influence this scaling relationship, especially for long-duration, large-scale precipitation events common in mid-latitude coastal regions like Japan. These events, frequently associated with atmospheric rivers (ARs) and monsoonal influences, cause widespread flooding and landslides. Previous research using daily precipitation analysis can be skewed for long-duration events. Event-based analysis, focusing on individual storm life cycles, is better suited to understand these complex interactions. This study uses event-based analysis of high-resolution precipitation data from Japan (646 stations over 25 years) to compare scaling rates and synoptic weather patterns for short- and long-duration events, focusing on the warm season (May-September) to minimize seasonal effects. The study uses the Japanese 55-year Reanalysis (JRA-55) to analyze associated atmospheric conditions.

Prior research has explored the relationship between extreme precipitation and temperature, focusing primarily on short-duration events. Studies have demonstrated super C-C scaling in short-duration extreme precipitation, particularly for convective events. Different methodologies have been used to separate convective and large-scale precipitation, including cloud type observation, lightning detection, large-scale circulation patterns, and event duration. The consensus points to the dominant role of convective precipitation in super C-C scaling. However, synoptic-scale storms, notably those associated with ARs, contribute substantially to extreme precipitation, especially in mid-latitudes, leading to widespread impacts. While studies have noted unexpectedly higher scaling rates for longer-duration (daily) precipitation, the link between scaling and synoptic situations for persistent events remains understudied. The limitations of fixed-interval (e.g., daily) analyses for long-duration events, where a single event can span multiple days, are acknowledged, highlighting the need for event-based statistics that capture individual storm characteristics and meteorological context.

The study employed an event-based analysis of 10-minute resolution precipitation data from 646 AMeDAS stations across Japan, covering 25 years (1994-2018). Data were carefully quality-controlled, focusing on the warm season (May-September) to minimize seasonal pattern variability. The JRA-55 reanalysis data provided the synoptic context. Precipitation events were defined as periods separated by at least 3 hours of no precipitation, with a total event accumulation >1.0 mm. Events influenced by tropical cyclones were excluded. Three key event properties were extracted: duration, peak hourly intensity, and total accumulation. Events were classified into short-duration (<5h, afternoon peak) and long-duration (>10h) categories based on diurnal variations in peak frequency, reflecting different physical mechanisms. Synoptic patterns were analyzed by compositing JRA-55 data for days exceeding the 99th percentile of peak intensity. A binning technique with 2°C wide bins and 1°C steps was used to examine the scaling relationship between temperature and precipitation (both peak intensity and total accumulation). The 99th percentile of each precipitation property was calculated for each temperature bin, with 95% confidence intervals determined through bootstrapping. Exponential regression was used to assess the scaling rates (ΔP% °C⁻¹), and the median temperatures of each bin were used in the analysis.

The study found that long-duration events, typically occurring overnight and associated with AR-like circulation patterns (indicated by integrated moisture flux and convergence), show a sharper increase in peak intensity with temperature than shorter-duration events. The scaling rate for peak intensity was 11.1% °C⁻¹ for long-duration events, compared to 9.8% °C⁻¹ for short-duration events and 9.7% °C⁻¹ for all events. Both peak intensity and total accumulated precipitation in long-duration events exhibit super C-C scaling, particularly at temperatures between 18°C and 24°C. For long-duration events, the scaling rates for median total precipitation accumulation were 11.8% °C⁻¹ and 11.4% °C⁻¹ based on different thresholds (99th percentile of peak intensity or total accumulation, respectively). This super C-C scaling is more apparent in events exceeding the 99th percentile of total accumulation and lasting longer than 24 hours. Below 18°C, the relationship weakens, possibly due to a larger contribution from non-convective precipitation masking the convective effects. Above 24°C, both peak intensity and accumulated precipitation show a decrease with temperature, possibly linked to reduced moisture availability. Long-duration extreme events were most frequent during the Baiu/Meiyu season (May-July), consistent with the known influence of southwesterly flow and the formation of mesoscale convective systems near the quasi-stationary Baiu front.

The findings demonstrate that long-duration precipitation events are considerably more sensitive to warming than shorter events, particularly at higher temperatures. This super C-C scaling for both peak intensity and total accumulation, especially in long-duration events resembling ARs, has significant implications for future flood risk in mid-latitude coastal regions. The dominance of long-duration events during the Baiu/Meiyu season suggests the importance of monsoonal dynamics and convective systems embedded within larger synoptic patterns. The observed relationship between local temperature and precipitation intensity suggests the utility of local temperature as a diagnostic tool for convective activity within synoptic-scale systems. While the exact mechanisms require further investigation, the results highlight the critical role of convective intensification within larger-scale systems. The peak-shaped scaling relationship, observed in some studies, may not represent the true upper limit of future precipitation extremes, as climate models suggest increases in both peak intensity and the temperature at which the peak occurs.

This study demonstrates a significant super C-C scaling of long-duration, large-scale precipitation events in Japan, particularly at higher temperatures. The increase in both peak intensity and total accumulation highlights a heightened risk of widespread flooding and landslides under future warming scenarios. The association of these events with ARs and monsoonal influence has broad implications for regional climate projections. Future research should focus on improving the representation of these large-scale, long-duration events in climate models, possibly using convection-permitting models.

The study focuses on Japan, limiting the generalizability of the findings to other regions. The analysis relies on observed data and does not explore the underlying physical mechanisms driving the super C-C scaling in detail. Uncertainty remains in the warmest temperature regimes due to limited event samples. The exclusion of tropical cyclone events could potentially influence the overall scaling results, warranting further investigation of their independent contribution.