Sensitivity to climate Delta selection in Pseudo-global warming simulations of mesoscale convective systems: case study of the august 2022 event in northern Japan

The pseudo-global warming (PGW) method is widely used to examine climate-change impacts on heavy rainfall. However, the thermodynamic variables selected as climate deltas in PGW experiments often differ among studies and are determined subjectively. This study investigates how different selections and combinations of climate deltas influence simulated thermodynamic and dynamic fields, using a heavy rainfall event associated with a mesoscale convective system (MCS) as a case study. We conducted PGW sensitivity experiments with the Weather Research and Forecasting (WRF) model by applying climate deltas for air temperature (T), sea surface temperature (SST), and relative humidity (RH), individually and in combination, based on climate projections. MCS precipitation was sensitive to the applied thermodynamic perturbations. The T delta increased atmospheric water vapor, raising basin-averaged accumulated precipitation by a factor of 1.62 relative to the control simulation. The SST delta increased surface heat fluxes and enhanced atmospheric instability, whereas the RH delta caused no substantial thermal changes. The SST and RH deltas shifted the heavy rainfall areas associated with the MCS, producing large changes in basin-scale precipitation. These results indicate that climate-delta selection can modify mesoscale dynamics and represents an important source of uncertainty in PGW experiments.

Mesoscale convective systems (MCSs) drive severe flooding disasters globally (Laing and Fritsch, 2000; Houze, 2004; Schumacher and Rasmussen, 2020; Ding et al., 2024). The frequency and intensity of MCSs are projected to increase due to climate change (Prein et al., 2017; Chan et al., 2023; Kawase et al., 2023; Rehbein and Ambrizzi, 2023; Hiraga et al., 2025a). In Japan, linear-shaped MCSs, known as “Senjo-Kousuitai,” produce devastating floods through persistent and intense precipitation (Kato, 2020). These linear-shaped MCSs are often characterized by band-shaped intense precipitation (>80 mm in 3 h), extending 50–300 km in length and 20–50 km in width (meso- β scale; Orlanski, 1975), and result from the continuous development of multiple convective cell clusters (Kato, 2020; Goto and Satoh, 2022). Evaluating future changes in these linear-shaped MCSs under climate change conditions is critical for flood risk management.

The Pseudo Global Warming (PGW) method is one of the approaches used to estimate future changes in specific heavy rainfall events using numerical models (Schär et al., 1996; Kimura and Kitoh, 2007). PGW applies a “climate delta” as a perturbation of climate change, which is the difference between the average future and present climate values for various thermodynamic and dynamic variables. By adding these perturbations to the initial and boundary conditions of a numerical model, PGW simulations can assess how rainfall events may change under global warming in the future (MCSs: Lackmann, 2013; Hwang et al., 2023; Tropical Cyclones: Patricola and Wehner, 2018; Reed et al., 2020; Atmospheric Rivers: Michaelis et al., 2022; snow fall: Hara et al., 2008; Rasmussen et al., 2011). In PGW experiments, the selection of thermodynamic and dynamic variables for perturbation plays a crucial role. Thermodynamic variables, such as temperature and moisture, are particularly significant due to their relatively robust response to warming compared to dynamic variables (Chou et al., 2009; Shepherd, 2014; Lin et al., 2024). Lin et al. (2024) applied PGW with perturbations in temperature, specific humidity, and sea surface temperature (SST) to study MCSs in China under global warming. Similarly, Hibino et al. (2018) and Tahara et al. (2025) examined future changes in linear-shaped MCS in Japan using PGW with temperature and SST perturbations. Naka and Nakakita (2023) applied climate deltas of water vapor mixing ratio, potential temperature, and SST to simulate the effects of climate change on two heavy rainfall events, including the one analyzed by Hibino et al. (2018). As seen in these studies, the selection of thermodynamic variables in PGW experiments varies among studies and is often determined subjectively. These variations in the choice and combination of thermodynamic perturbations may influence projected changes in heavy rainfall, making it difficult to compare each result.

In particular, the linear-shaped MCSs may exhibit diverse responses to changes in environmental conditions due to their high sensitivity to atmospheric perturbations, in contrast to larger synoptic-scale systems. Hiraga et al. (2025a) analyzed different warming scenarios by considering climate deltas of air temperature alone, SST alone, and their combination in PGW-based extreme precipitation simulations, which showed that the selection of climate deltas significantly altered the location and intensity of precipitation. Kozawa et al. (2022) investigated the impact of incorporating relative humidity perturbations in PGW and found that it reduced the projected precipitation of a linear-shaped MCS in northern Kyushu, Japan. Hiraga and Tahara (2025a) highlighted the strong sensitivity of MCS rainfall intensity and structure to atmospheric moisture content. These findings suggest that PGW-based future projections of heavy rainfall may be subject to considerable uncertainties arising from the selection of variables and their combinations. Assuming this high sensitivity, assessing such uncertainties is essential for improving the reliability of future projections of heavy rainfall and associated flood risks.

Therefore, this study aims to evaluate the sensitivity of projected heavy rainfall associated with linear-shaped MCS to the selection and combination of thermodynamic variables in the PGW method. While previous research has examined variable selection in PGW simulations for tropical cyclones (Nakamura and Mäll, 2021) and explored the effects of dynamic variables (wind, geopotential height, and sea level pressure) in storm simulations over the northern U.S. (Xue et al., 2023), no study has specifically investigated the impact of thermodynamic variable selection on MCS simulations in a PGW framework. Given the relatively small spatial and temporal scales of MCSs, understanding how different thermodynamic perturbations influence PGW results is particularly important to enhance the understanding of uncertainties in PGW-based future projections of heavy rainfall associated with MCSs.

Target event and study area

This study focuses on the heavy rainfall event caused by a linear-shaped MCS in northern Japan, specifically the Tohoku and Hokuriku regions, during August 2022. On August 3, 2022, localized linear-shaped MCS, known as “Senjo-Kousuitai (Kato, 2020),” developed in these areas (Figure 1D). The 24-h accumulated precipitation exceeded 600 mm in the Arakawa River Basin, marking the highest recorded precipitation in its observational history (Japan Meteorological Agency (JMA), 2022a). This study focuses on the Arakawa and Akagawa River Basins (Figure 1C).

The primary meteorological mechanism driving this event was a quasi-stationary front lingering over the Tohoku and Hokuriku regions. A quasi-stationary front with a low-pressure system stagnated over northern Japan, where warm, moist low-level inflow toward the front caused heavy rainfall (Figure 1B; Japan Meteorological Agency (JMA), 2022a; Tokyo Climate Center (TCC), 2022; Zhao et al., 2023). Persistent low-level convergence and continuous moisture transport resulted in deep convection, culminating in the formation of cumulonimbus clusters through back-building and back-and-side-building processes, thereby generating extreme localized precipitation (Tochimoto and Hirockawa, 2024; Hiraga and Tahara, 2025b; Tahara et al., 2025). The frequency of similar MCS-driven rainfall events has increased in recent years and have caused flooding and landslide disasters. These events are also projected to become more frequent under future climate conditions (Kawase et al., 2023). Therefore, assessing projected changes in such events on an event-based scale provides critical insights into future hydrometeorological risks.

Model setup

This study employed the Advanced Research version of the Weather Research and Forecasting Model (WRF-ARW) version 4.1.2 (Skamarock et al., 2019). The model configuration follows that of Tahara et al. (2025), who also conducted simulations and PGW experiments for the same heavy rainfall event. The model configuration is summarized below, following Tahara et al. (2025).

Figure 1A shows the model domain, which employs a two-way nesting approach with horizontal resolutions of 12 km, 4 km, and 1.3 km. The vertical levels comprise 45 layers, extending up to 50 hPa. The model employs the Goddard 4-ice microphysics scheme (Tao et al., 2016) and the Grenier–Bretherton–McCaa planetary boundary layer scheme (Grenier and Bretherton, 2001) as the physical parameterization schemes. Given that the resolution of the outer domain is too coarse to explicitly resolve cumulus convection, the Kain-Fritsch cumulus parameterization (Kain, 2004) is applied solely to the outermost domain. These physical parameterization and domain settings have been successfully used to reproduce MCSs in the target region, as demonstrated by Hiraga and Tahara (2025b).

Initial and boundary conditions are obtained from the NCEP GDAS FNL operational global analysis dataset, which has a horizontal resolution of 0.25°, 34 vertical levels, and a temporal resolution of 6 h (National Centers for Environmental Prediction (NCEP), National Weather Service, NOAA, U.S. Department of Commerce, 2015). The simulation period extends from 00:00 UTC on July 30 to 00:00 UTC on August 7, 2022, including a 72-h spin-up period. The baseline simulation of the heavy rainfall event is referred to as the “CTL run” (Figure 1D).

PGW experiment

In this study, the PGW approach modifies initial and boundary conditions by incorporating monthly mean climate perturbations derived from the MIROC6 model under the SSP2-4.5 scenario. The MIROC6 model, developed collaboratively by the Center for Climate System Research (CCSR) at the University of Tokyo, the Japan Agency for Marine-Earth Science and Technology (JAMSTEC), and the National Institute for Environmental Studies (NIES) (Tatebe et al., 2019), is widely used in climate change research in Japan (Song et al., 2023; Tahara et al., 2025). The climate perturbations were calculated as the differences between the future and present monthly climate means. The present-climate mean was calculated for 2015–2024, corresponding to the first decade of the SSP2-4.5 simulation and representing a recent climate state temporally close to the August 2022 event. The future climate means were calculated for 2041–2060 and 2081–2,100 to represent the mid- and late twenty-first-century climate states, respectively, while reducing the influence of interannual variability. These PGW settings follow Tahara et al. (2025) for the same August 2022 heavy rainfall event, facilitating direct comparison with their assessment of uncertainty arising from GCM selection in the PGW framework. This study does not address uncertainties associated with global climate models (GCMs) or future climate scenarios because it only focuses on evaluating the impact of specific thermodynamic variables within the climate deltas though the importance of such uncertainty is well acknowledged in PGW experiments (Tahara et al., 2025).

This study adopted the different selection and combination of thermodynamic variables for the climate deltas in the PGW experiment. Three thermodynamic variables—air temperature (T), sea surface temperature (SST), and relative humidity (RH)—are modified individually and in combination to generate different PGW scenarios. These include three individual variable perturbations, three two-variable combinations, and a full three-variable perturbation, yielding a total of seven experimental cases. When the T climate delta was applied, relative humidity was retained, causing the water vapor mixing ratio to increase with temperature. Although previous PGW studies often excluded RH climate deltas because projected RH changes were considered insignificant (Pall et al., 2007; Kawase et al., 2009; Takemi et al., 2012), we included the RH climate delta as a sensitivity experiment because localized MCS precipitation may respond to small thermodynamic perturbations through changes in mesoscale circulation and convergence.

The perturbations in the PGW method are primarily introduced through thermodynamic variables because the climate change signals in thermodynamic variables are more robust than those in dynamic changes (Shepherd, 2014). Additionally, Xue et al. (2023) pointed out that spatially averaging the climate deltas in a PGW experiment cannot retain spatial contrasts in warming, such as land-sea differences, which can affect precipitation. To address this, the climate perturbations in this study are applied at each grid level. Because the thermodynamic perturbations were applied without corresponding changes in wind and geopotential height, exact dynamical consistency, including thermal wind balance, was not explicitly maintained in the initial and boundary conditions. The WRF model was integrated with a 72-h spin-up period to allow the atmospheric fields to adjust dynamically to the modified thermodynamic conditions, although complete removal of the introduced imbalance was not guaranteed. The PGW simulations are denoted as “future-period-variable” runs, such as the “2090s-T + SST run”.

Figures 2A,B show vertical profiles of projected change in air temperature and relative humidity based on MIROC6 output. Air temperature increases by approximately 1–2 K in the 2050s and 2–3.5 K in the 2090s, from the surface up to 200 hPa. The upper troposphere exhibits a more pronounced temperature increase compared to the lower levels, consistent with previous studies (Knutson and Tuleya, 2004; Hill and Lackmann, 2011; Delfino et al., 2023). Relative humidity increases by approximately 1%–2% up to 400 hPa. Figures 2C,D illustrate SST changes, which show an increase of approximately 0.5–1.5 K in the 2050s and 1.0–3.0 K in the 2090s. The Japan Sea (northeast of Japan) exhibits a higher rate of warming compared to the Pacific Ocean (south of Japan), consistent with previous studies (Mizuta et al., 2017; Tahara et al., 2025).

This study followed the analysis steps below. First, this study investigated how the selection and combination of thermodynamic variables in the PGW framework affect changes in precipitation. Second, this study investigated how the thermodynamic environment of the relatively large scale (i.e., entire simulation domain) changes under the PGW approach. Third, to determine whether precipitation changes were driven by synoptic-scale modifications or more localized perturbations, this study conducted a multiscale analysis of the thermodynamic and dynamic mechanisms contributing to precipitation changes.

Precipitation change

The selection and combination of thermodynamic variables for climate deltas substantially affected the intensity and spatial distribution of the simulated precipitation. Figures 3, 4 show the 48-h accumulated precipitation distribution, while Figure 5 presents the domain-averaged precipitation over domain 03, the Arakawa River Basin, and the Akagawa River Basin.

Applying only the air temperature climate delta (T) resulted in a marked increase in accumulated precipitation in domain 03. Specifically, in the 2090s-T run, total precipitation over the Arakawa River Basin became 1.62 times that of the CTL run (Figures 4, 5). Similarly, the 2050s-T run exhibits intensified rainfall over the Arakawa River Basin and the southern Akagawa River Basin (Figures 3, 5). In contrast, when only the sea surface temperature (SST) climate delta was applied, accumulated precipitation in domain 03 decreased relative to the CTL run, with total precipitation in both basins remaining comparable to or lower than the CTL run (Figure 5). Applying only the SST climate delta caused the intense rainfall band to shift southward compared to the CTL run, reducing mean precipitation in the basins. Applying only the relative humidity (RH) climate delta did not substantially alter domain-averaged precipitation. However, a northward shift of the intense rainfall band occurred in the 2050s-RH and 2090s-RH runs, leading to increased precipitation over the Akagawa River Basin, while precipitation over the Arakawa River Basin remained comparable to or slightly reduced relative to the CTL run.

When multiple climate deltas were applied simultaneously, the precipitation responses generally retained the characteristic effects of the individual deltas. Precipitation in domain 03 was substantially greater in the T + SST, T + RH, and T + SST + RH runs than in the corresponding runs without the T delta, namely the SST, RH, and SST + RH runs. This result indicates that the precipitation-enhancing effect of the T delta persisted when combined with the SST and RH deltas. Similarly, the rainfall band in the 2090s-T + RH run shifted northward relative to the CTL run, consistent with the response observed in the 2090s-RH run. These results indicate that the effects of individual climate deltas were qualitatively retained in the combined experiments, although the resulting precipitation responses were not necessarily linear or strictly additive.

Thermal environmental changes under the PGW method

To assess how individual climate deltas modified atmospheric thermodynamics in WRF simulation, changes in the PGW-induced thermal environment were examined. Figure 6 shows vertical profiles of air temperature and water vapor mixing ratio differences from the CTL run. Thermodynamic modifications within the WRF simulations correspond directly to the specific climate deltas applied to the model. When applying RH climate deltas alone, air temperature remained nearly unchanged because the RH perturbation did not directly modify temperature. The water vapor mixing ratio changed only slightly because the applied RH delta was small and air temperature remained nearly unchanged at all atmospheric pressure levels. This suggests that the thermodynamic impact of RH climate deltas is relatively small compared to other variables, which is consistent with previous studies (Olschewski and Kunstmann, 2024).

Conversely, when the air temperature climate delta was applied (T, T + SST, T + RH, T + SST + RH runs), air temperature increased throughout the atmosphere, following the vertical warming profiles projected by the GCM. Applying the SST climate delta enhanced air temperature near the surface and increased water vapor mixing ratio, likely due to enhanced heat flux from the ocean. These vertical profiles confirm that the PGW framework effectively integrates future climate signals into the regional model.

Figure 7 shows differences in sensible heat flux and latent heat flux from the CTL run. Focusing on the Japan Sea, in the CTL run, the air temperature in the lowest atmospheric layer was higher than the SST, showing negative sensible heat flux. When air temperature delta was included without the SST delta (T, T + RH runs), the increase in temperature and water vapor mixing ratio in the lowest atmospheric layer were smaller than those in the overlying lower atmosphere (Figures 6B,D). By applying only the T climate delta, the temperature in the lowest atmospheric layer became even higher than the SST compared to the CTL run, causing a decrease in upward sensible heat flux (i.e., an increase in negative sensible heat flux, Figure 7B). On the other hand, when only the SST climate delta was applied, the temperature difference between the lowest atmospheric layer and SST decreased, leading to an increase in upward sensible heat flux compared to the CTL run (Figure 7C). Furthermore, in the CTL run, the upward latent heat flux was positive, but as the relative humidity remained almost unchanged and only the temperature increased, the water vapor content in the lowest atmospheric layer increased. This reduced the specific humidity difference between the atmosphere and the sea surface, resulting in a decrease in upward latent heat flux compared to the CTL run. These results show that changes in lower-atmospheric temperature and water vapor mixing ratio are associated with surface heat flux responses to the applied climate deltas.

It should be noted that applying the SST or air-temperature climate delta in isolation can produce an idealized and potentially unrealistic thermodynamic imbalance between the ocean surface and the overlying atmosphere. These experiments were not intended to reproduce physically complete future climate states, but rather to isolate the effects of individual thermodynamic variables and quantify uncertainty arising from climate-delta selection. Nevertheless, air temperature and SST vary on different timescales, and transiently large air–sea temperature differences can occur over the lifetime of an MCS under actual weather conditions.

Multiscale analysis of thermodynamic and dynamic changes contributing to precipitation change

This section investigates changes in the thermodynamic and dynamic processes at the large-scale and local-scale contributing to precipitation changes.

Climate deltas significantly modify localized convergence zones while exerting minimal influence on large-scale (synoptic-scale) moisture transport pathway (Figure 8). The location of synoptic-scale moisture transport pathway did not change significantly regardless of the applied climate delta, while the precipitable water increased when the air temperature climate delta was applied. On the other hand, different climate deltas shifted the convergence zone, affecting precipitation patterns. Specifically, applying the SST delta shifted the convergence zone southward toward the Arakawa River Basin, while applying the RH delta moved the convergence zone northward over the Akagawa River Basin. These results emphasize that localized dynamic responses to thermodynamic perturbations determine the spatial characteristics of heavy rainfall.

We found that the location of convergence zone was affected by cyclonic circulation stagnating over the Japan Sea. Figure 9 shows the JMA surface weather map and sea level pressure field in the CTL run. Figure 10 shows the relative vorticity and wind vectors at 500 m at 00:00 UTC on August 3. In this event, the small low-pressure system on the front over the Japan Sea moved westward along with the southward shift of the front. This low-pressure system subsequently split, with one part continuing westward and the other remaining over the Japan Sea, where it formed a cyclonic circulation (Figures 9, 10A,E). When the air temperature delta was applied, this cyclonic circulation remained in nearly the same position as in the CTL run (Figures 10B,F). On the other hand, when the sea surface temperature (SST) climate delta was applied, the cyclonic circulation expanded and shifted southward compared to the CTL run, altering the location of the associated convergence zone. In contrast, when the relative humidity (RH) climate delta was applied, the circulation shifted northwestward, resulting in a corresponding northward shift of the convergence zone. The relationship between the location of the low-pressure system stagnating over the Japan Sea in this region and the heavy rainfall area associated with the linear-shaped MCS has also been reported based on statistical analysis (Tahara et al., 2026). These results show that the simulated large-scale circulation differed only slightly among the climate-delta experiments, whereas the location of the localized convergence zone varied substantially, influencing the precipitation patterns associated with the MCS.

The basin-scale precipitation changes align primarily with the intensity and persistence of moisture convergence rather than thermodynamic potential alone. Figures 11, 12 show the time series of basin-mean precipitation, maximum precipitable water, and moisture convergence for the Arakawa and Akagawa River basins. When the air temperature climate delta was applied, the amount of water vapor entering the basin increased compared to the CTL run. The timing of strong low-level convergence generally aligns with periods of heavy precipitation (e.g., Arakawa: 2090s-T run, Akagawa: 2090s-T + RH run). On the other hand, even when precipitable water was high and atmospheric moisture content was abundant, precipitation remained low if strong convergence did not occur (e.g., Arakawa: 2090s-T + SST + RH run, Akagawa: 2090s-T + SST + RH run). This indicates that increased water vapor availability alone was insufficient to explain the precipitation increase, which generally coincided with strong low-level convergence. This correspondence does not establish a one-way causal relationship because low-level convergence can promote convective triggering, while convection can in turn strengthen convergence. The differences in moisture convergence among the PGW experiments represent changes in the mesoscale organization supporting the MCS, although the internal convective structure and near-surface wind intensity were not explicitly evaluated.

Discussion of the findings

This study investigated how the selection and combination of thermodynamic climate deltas affect projected precipitation associated with a linear-shaped MCS to in the PGW framework. While the PGW method is widely used to assess climate change impacts, there are no clear guidelines for selecting and combining climate deltas. We demonstrated that different selections of thermodynamic variables produced distinct thermodynamic and dynamic responses, leading to substantial variations in MCS-associated precipitation. This sensitivity indicates that climate-delta selection can be an important source of uncertainty in PGW-based projections of heavy rainfall.

Previous studies have also investigated the impacts of thermodynamic changes associated with different applications of climate deltas on the precipitation and structure of extreme events. In PGW experiments for a Mediterranean tropical-like cyclone, Koseki et al. (2021) found that warming only the sea surface and skin temperature intensified surface winds, evaporation, and precipitation, whereas warming only the atmosphere weakened the cyclone, reducing SLP, winds, and rainfall. Similarly, Chen et al. (2020) showed that SST warming strengthened tropical cyclones (TCs) by increasing heat flux supplied from a warmer ocean, while air temperature warming reduced their intensity by enhancing atmospheric stability. Consistent with these studies, the climate deltas in our simulations modified convective instability through distinct vertical temperature responses (Figure 13). Air temperature perturbations reduced CAPE by warming the upper troposphere more significantly than the surface levels, whereas SST perturbations increased CAPE by providing additional heat and moisture to the lower atmosphere. Relative humidity produced only minor changes in convective instability. On the other hand, in this study, the magnitude of convective instability did not necessarily correspond to the magnitude of precipitation changes, unlike previous experiments targeting TCs. For example, when the SST climate delta was added, instability increased, whereas when the air temperature climate delta was added, instability decreased. However, the increase in precipitation was larger when the air temperature climate delta was added than when the SST climate delta was added. This is considered to be attributable to the different precipitation processes between TCs and MCSs. In addition, CAPE values were already very high in the CTL simulation for this event, possibly limiting the impact of the increased instability introduced by the SST climate delta on precipitation changes.

The high sensitivity of MCSs to the selection and combination of climate deltas can be attributed to changes in key processes such as moisture inflow, low-pressure system positioning, and convergence zone formation—characteristics also noted in previous studies (Araki et al., 2021). Specifically, this study indicates that applying air temperature climate delta tends to significantly increase precipitation, whereas applying SST and relative humidity climate deltas have a relatively smaller impact on precipitation amounts but do alter the location of the intense rainfall region. These findings suggest that understanding the physical mechanisms of the target event is essential for future projections using the PGW method, as it is crucial to clarify the response to various variables depending on the type of extreme rainfall event.

While the perturbations introduced by climate deltas had limited effects on the synoptic-scale atmospheric configuration, they modified mesoscale structures, leading to substantial modifications in precipitation patterns associated with MCSs. One advantage of the PGW method is that it largely preserves the synoptic-scale conditions while isolating thermodynamic changes linked to climate warming (Dai et al., 2020; Lucas-Picher et al., 2021). However, despite the relatively weak influence of climate deltas on synoptic-scale dynamics, the mesoscale atmospheric responses, particularly in low-level circulation and convergence zone position, play a crucial role in shaping precipitation characteristics. Notably, the positioning of the low-pressure system in this event was a key factor influencing precipitation variability (Japan Meteorological Agency (JMA), 2022b). In PGW experiments focusing on specific events, these scenarios arising from different selections and combinations of climate deltas should be regarded as potential future conditions and as a source of uncertainty. The weak synoptic-scale differences should also be interpreted in the context of the experimental design. The relatively short simulation period may have limited the development of slowly evolving synoptic-scale responses to the thermodynamic perturbations. In addition, the lateral boundary conditions may have constrained the large-scale circulation within the WRF domain, potentially reducing the simulated synoptic-scale differences among the PGW experiments.

Limitations and future directions

This study has some limitations that should be addressed in future research.

This study focused on examining responses to perturbations in thermodynamic variables (air temperature, SST, relative humidity), highlighting the significant sensitivity of MCS to environmental changes. This is because thermodynamic variables are the most commonly used in PGW simulations. However, responses to dynamic variables remain unclear. For instance, Goto and Satoh (2022) demonstrated that, in similar quasi-stationary linear precipitation systems, higher wind speeds, rather than increased water vapor, were the main factor enhancing low-level water vapor flux. The importance of dynamical mechanisms in linear-shaped MCSs was also confirmed in this study, highlighting the need to investigate the influence of dynamical variables in climate change impact assessments using the PGW framework. Also, given the characteristically high CAPE and strong flow instability of the target event in this study, the simulated mesoscale patterns may also be sensitive to infinitesimal perturbations. Although this study focused on sensitivity to climate-delta selection, comparing these responses with the inherent variability generated by an initial-condition ensemble would provide a more robust baseline for distinguishing climate-delta responses from intrinsic flow sensitivity.

To generalize our findings, it is necessary to conduct similar sensitivity experiments on a larger number of MCSs. The response of MCSs to perturbations can be explained by changes in heavy rainfall mechanisms, and their sensitivity to variable selection and combination may vary significantly depending on the structural type of MCS and its initiation mechanism, such as orographically induced convection or low-level convergence-driven MCS development. Given that global warming exhibits spatially heterogeneous signals, it is equally important to investigate how MCSs respond to climate deltas in different geographical regions, such as the United States, Europe, and East Asia. Additionally, comprehensive and systematic sensitivity assessments targeting phenomena with different temporal and spatial scales, such as tropical cyclones (TCs) and atmospheric rivers (ARs), are essential for advancing the pseudo-global warming (PGW) framework.

Furthermore, the PGW approach involves various uncertainties sources beyond variable selection, including the choice of GCMs used to introduce climate change effects, the lack of internal climate variability effects due to the use of multi-decadal climate mean differences (e.g., El Niño, decadal oscillations), and downscaling configuration choices (e.g., model resolution and physical parameterization, domain locations). Tahara et al. (2025) and Hiraga et al. (2025a) demonstrated the significant uncertainty associated with GCM selection in the context of MCSs. Trapp et al. (2021) compared the effects of a traditional decadal-average climate delta with a diurnally varying climate delta, which accounts for the diurnal cycles of temperature, humidity, and winds. Tahara et al. (2025) demonstrated that the projected future precipitation varies significantly depending on the domain location. Comparing the relative magnitude and criticality of these uncertainty sources is essential for establishing a systematic PGW framework.