The increasing scientific evidence linking anthropogenic warming to extreme weather events, particularly tornadoes, necessitates understanding how these events might respond to climate change. Challenges in studying tornadoes include inhomogeneous reporting, sparse data, and incomplete historical archives. These issues are further complicated in Japan due to limited observational stations and sounding data availability at only 0900 and 2100 JST. To overcome limitations of sparse radiosonde data, this study utilizes pseudo-proximity (grid point) soundings from the high-resolution fifth-generation ECMWF reanalysis (ERA5m), offering improved spatiotemporal resolution compared to traditional observations. While many studies using pseudo-proximity soundings focus on the United States, this approach has expanded to global analyses, including studies of tornadic and severe convective environments worldwide. Studies on severe convective environments’ climatology, trends, and future changes have generally shown increased atmospheric instability and convective potential with warming, but also weakened deep-layer vertical wind shear in mid-latitudes. However, changes vary regionally and seasonally. This study expands upon previous work by providing a comprehensive analysis of tornadic events in Japan using environmental proxies, examining the efficacy of various convective parameters including CAPE, CIN, K-index, LCL, BWD, and SRH, as well as multivariate parameters (EHI, KHI, and STP) to distinguish between weak and significant tornadoes. The study then evaluates the frequency and potential future changes in these environments using reanalysis and a high-resolution large ensemble regional climate experiment.

Previous research on tornado environments has largely focused on the United States, leveraging high-resolution data and advanced techniques. However, global studies are hindered by inconsistent reporting standards and data limitations. While a general understanding suggests increased atmospheric instability and convective potential with warming climates, the impact on deep-layer vertical wind shear is less certain and varies regionally. Existing studies on Japan are limited, often relying on coarse-resolution models or indices derived from US data, which may not be directly applicable to the Japanese context. This study seeks to address these gaps by utilizing high-resolution reanalysis data and tailored tornado environment indices for Japan.

This study uses a dataset of 469 tornadic events in Japan (1979-2021) from the Japan Meteorological Agency's database, categorized by intensity on the Fujita scale. Pseudo-proximity soundings from the ERA5m reanalysis are used to obtain high-resolution environmental data around each tornado event. The analysis examines various thermodynamic (CAPE, CIN, LCL, K-index), kinematic (BWD, SRH), and multivariate (EHI, KHI, STP) parameters to identify those best distinguishing F2+ tornadoes from weaker events. Thresholds are established for each index to define F2+ tornado environments. Long-term trends (1979-2021) are analyzed for F2+ environments in four high-frequency regions (Kanto, Chubu-Kinki, Kyushu, Ryukyu) using the Mann-Kendall test and Theil-Sen slope estimates. Future changes are evaluated using a large ensemble 2-K warming experiment from the d4PDF dataset, focusing on areas where all ensemble members agree on the sign of change. The historical simulation (HIST) is compared against ERA5m to validate the model's ability to reproduce observed environmental conditions. The spatial distribution of parameters and their projected changes under a 2-K warmer climate are assessed.

The study confirms that stronger tornadoes in Japan exhibit higher instability (CAPE, K-index), lower-level moisture (LCL), and stronger SRH, consistent with global findings. Multivariate parameters (EHI, KHI, STP) show better discrimination between tornado intensities than individual parameters. Analysis of long-term trends (1979-2021) reveals statistically significant increasing trends in F2+ environments in some regions, particularly for EHI and STP in Kanto and EHI in Kyushu and Ryukyu. These increases are largely attributed to rising CAPE and K-index, while BWD and SRH show insignificant trends. The increasing trends in atmospheric instability are linked to regional warming. The 2-K warming experiment projects robust increases in F2+ environments across much of Japan, especially in northern regions, where a near doubling is projected based on EHI, KHI, and STP. The primary driver is the projected increase in atmospheric instability due to increased near-surface humidity. The Kanto region shows particularly strong increases in EHI and comparable increases in CAPE. However, the Ryukyu region shows less certainty in projected changes, possibly due to uncertainties in projected tropical cyclone activity. The study notes lower median values for EHI and STP in Japan compared to the US, likely reflecting differences in overall convective environments.

The findings demonstrate a clear link between increasing atmospheric instability and the projected increase in F2+ tornado environments in Japan. The use of high-resolution reanalysis data and multivariate indices provides a more nuanced understanding than previously possible. The observed increasing trends in some regions and the projected increases across much of Japan, especially in densely populated areas, highlight the need for increased public awareness and preparedness for tornadoes. The study's use of a large ensemble climate simulation reduces uncertainty but also reveals regions (like Ryukyu) where projected changes are less certain, potentially due to complexities in the interplay of regional climate factors. The use of high resolution data in this study also highlights the improvements that can be made in improving forecasting and future research.

This study provides a comprehensive analysis of tornadic environments in Japan using high-resolution reanalysis data and a large ensemble climate simulation. The results confirm the importance of multivariate parameters in distinguishing strong tornadoes and highlight the significant projected increases in strong tornado potential across many regions of Japan under a warming climate, primarily due to increased atmospheric instability. Future research should focus on higher-resolution convective-permitting models, refined tornado indices specific to Japan, and continued monitoring to disentangle climate change effects from internal variability.

The study's reliance on damage reports for tornado intensity assessment introduces potential biases and uncertainties. The 2-K warming experiment does not fully capture all aspects of future climate change, including potential changes in tropical cyclone activity and mesoscale processes. The study's focus on environmental proxies does not directly predict the number of tornadoes on the ground but rather the potential for them to occur.