Numerous small island states are confronting substantial, potentially existential, threats due to climate change. These islands are often not represented in Global Climate Models (GCMs) because of their coarse resolution and exclusion from current CORDEX domains, resulting in their absence from regional climate projection ensembles. Consequently, climate change projections for many small islands are derived from GCM projections for the open ocean, which do not accurately reflect the island climates. Beyond sea level rise, these islands are affected by changes in various climate extremes (e.g., tropical cyclones, extreme rainfall, extreme heat, drought). However, future changes in these hazards over islands are characterized by low confidence, as indicated in the recent IPCC report. This project aims to generate actionable climate change projections for a region encompassing numerous Pacific island states and territories. To capture critical island climate processes, these projections must be high resolution (kilometer scale). Additionally, to quantify confidence levels and uncertainties, a multi-model ensemble approach is necessary. The project will also incorporate empirical-statistical downscaling methods, including those based on machine learning techniques, tailored to island climates. This presentation outlines the project plan and the current status of IC-Pac.

This study examines the ability of convection-permitting downscaling at 8km and 2km resolutions to improve the representation of precipitation over the Western Maritime Continent (WMC) relative to the driving global climate model. We focus on sub-daily time scales, which has received relatively little attention in prior work. In addition to mean precipitation, detailed characteristics including the diurnal cycle, intensity distribution, and spatial and temporal coherence statistics are evaluated. The evaluation is performed on both coarse (2 degree) and fine (0.25 degree) spatial scales over the WMC and on a 2km grid over Singapore, making use of Singapore observations as well as the IMERG and PERSIANN-CCS-CDR datasets. In terms of mean precipitation, the downscaled simulations broadly reflect the spatial features from the driving global models, including the spatial differences in mean precipitation, with noticeable differences in finer scale structures, especially over orography and coastlines. In contrast, detailed precipitation characteristics such as the diurnal cycle, intensity distribution, spatial and temporal coherence appear to be strongly shaped by the convection-permitting model. Nevertheless, the precipitation characteristics in the downscaled models are largely in agreement with observations and provide confidence in their usage in a wide range of applications.

To analyze future climate change at the regional scale, dynamical downscaling from global climate data through regional climate models(RCM) is essential. However, as can be seen in the ESGF node, long-term global data such as CMIP, due to limited storage capacity, are provided only for some standard pressure levels, which are mainly used for climate analysis. Therefore, there is a limit to dynamical downscaling, where the quality of the results depends on the performance of the input data. Vertical incremental interpolation(INC) is a technique that supplements information by interpolating the predictions of a general circulation model(GCM) into the forcing used as input data for RCM. In this study, we used GloSea6 data as forcing data, arbitrarily restricting the number of levels and time intervals, and compared the results of dynamical downscaled precipitation. The models used in the study are GMP, a global model of the Global/Regional Integrated Model system (GRIMs), and RMP, a regional model. The dynamical downscaling experiment was conducted over the East Asian region centered on Korea Peninsula for the summer period from June to August at a horizontal resolution of 25 km. First, when the precipitation results were compared with the observations, it was confirmed that the results obtained through dynamical downscaling had a significantly reduced error in intensity compared to the precipitation of the global model used as the forcing. The experiment with limited forcing could not properly simulate the summer precipitation and diurnal cycle, but after applying the incremental interpolation method, it was shown that the precipitation results using the entire levels and the diurnal cycle were reproduced. By interpolating the increments, the atmospheric field and diurnal variation of the GCM prediction were supplemented, so that sufficient downscaling results could be obtained with less information.※ This work was funded by the Korea Meteorological Administration Research and Development Program under Grant(RS-2024-00403386)

Regional Climate Models (RCMs) are dependent on boundary conditions provided by Global Climate Models (GCMs). A significant challenge in regional climate modelling is the "Garbage in – garbage out" problem. Specifically, if the input boundary conditions from a GCM are unrealistic, the RCM cannot rectify this and will consequently produce inaccurate results. While we can avoid using unrealistic GCMs, this issue is critical as all GCMs, even the best performing, exhibit biases. Past work has shown bias correction of GCM boundary conditions can mitigate this problem and enhance RCM simulations in a number of ways including the mean climate, extremes, compound events, and synoptic systems. Calibration/validation experiments show improvements in the simulated climate persist outside the bias correction calibration period. Here we outline a new methodology for applying the bias correction to future simulations and explore the impact of the bias correction on future regional climate changes.

Tropical cyclones (TCs), a major natural phenomenon that causes substantial socio-economic damage, occur approximately 25 times annually in the western North Pacific (WNP). Among these, about 3 to 4 TCs directly or indirectly impact Korea each year. Despite their relatively small number, the potential damage they inflict can be exceptionally severe. To enhance preparedness and response measures for TCs, it is crucial to improve the predictability of sub-seasonal to seasonal (S2S) forecasts, which cover time frames ranging from two weeks to two months. Accordingly, identifying the causes of sub-seasonal forecasting errors in the Korea Meteorological Administration’s Global Seasonal Forecast System version 6 (GloSea6) is essential for minimizing TC-related damage through proactive strategies. GloSea6 underestimated TC activity and intensity over the WNP from June to September. During periods of low performance in sub-seasonal TC simulations, a characteristic deficiency was observed in reproducing the variability of the western North Pacific Subtropical High. To enhance the forecast skill of sub-seasonal TC predictions, GloSea6 was applied dynamically downscaling using the Weather Research and Forecasting (WRF) Model.  In the GloSea6, the WNPSH rapidly weakened while weakly simulating a circum-global teleconnection (CGT) from the north of India, but it is shown that the predictability of sub-seasonal TC was improved by realistically simulating the CGT through dynamical downscaling. Dynamic downscaling improved the prediction of sub-seasonal TC activity; however, it had been observed that TC intensity was excessively simulated. To address this issue, the Ocean Mixed Model was coupled with the WRF model. As a result, the previously overestimated TC intensity was more realistically simulated.

Regional climate downscaling is essential for enhancing coarse-resolution global climate model outputs for high-impact weather assessments and regional climate applications. Given the increasing significance of air-sea interactions in regional climate projections, this study evaluates the performance of the Coupled Ocean-Atmosphere-Wave-Sediment Transport (COAWST) model in downscaling the Coordinated Regional Climate Downscaling Experiment for East Asia. Specifically, we compare COAWST with the uncoupled Weather Research and Forecasting (WRF) model to assess the effects of two-way air-sea interaction.We utilize the fifth-generation ECMWF reanalysis (ERA5, 0.25°, 6-hourly) and the Global Ocean Physics Reanalysis (GLORYS, 0.083°, daily) datasets as initial and boundary conditions. The COAWST model consists of WRF v4.2.2 for the atmosphere and the Regional Ocean Modeling System (ROMS) v3.9 for the ocean, with a horizontal resolution of 12 km. We first perform sensitivity experiments to physical parameterizations for a short period to identify the optimal configuration, then conduct a single long-term simulation for the entire period, focusing on precipitation, temperature, and other high-impact weather phenomena.This study aims to assess whether the coupled model offers more reliable climate representations than the atmosphere-only model. Since air-sea interaction strongly influences regional climate processes, improving downscaling performance could have significant implications for future climate projections. Provisional results will be presented, demonstrating the extent to which atmosphere-ocean coupling enhances the performance of regional climate downscaling, particularly in high-impact weather phenomena. This study provides valuable insights into the feasibility of incorporating ocean coupling in regional downscaling efforts for East Asia.

Simulating winter temperatures in East Asia is vital for predicting extreme weather events, managing energy demands, supporting agriculture, protecting public health, and understanding climate change trends. Land-ocean interactions involve the exchange of heat, moisture, and momentum between the ocean and atmosphere, influencing winter monsoons and long-term climate variability. Since oceanic conditions highly influence winter temperatures in the East Asian region, accurate simulations with air-sea coupling are necessary for effective disaster mitigation, energy planning, and climate adaptation strategies. This study employed the coupled ocean-atmosphere-wave-sediment transport (COAWST) modeling system and a standalone atmospheric model (WRF) to simulate high-resolution regional climate information during extremely strong East Asian winter monsoon year. The performance of the coupled (CPL) and uncoupled (UNCPL) simulations was evaluated by comparing atmospheric and oceanic variables. The CPL outputs showed notable cold biases relative to ERA5, particularly over the Manchurian (MC) region, although the CPL simulations captured the overall spatial pattern well, with a high pattern correlation coefficient. However, the cold biases in CPL experiments were reduced by approximately 1°C compared to those in the UNCPL runs. Furthermore, while using SSTs that account for atmosphere-ocean interactions instead of SSTs from reanalysis data led to increased SST simulation errors, this study was more valuable as the coupled atmosphere-ocean model captured the atmospheric energy balance and realistic oceanic flow dynamics.

High-resolution, gridded, self-consistent, continental-scale climate data are essential for understanding historical and future climate variability and risks,  as well as for providing actional insights to stakeholders and policymakers. We utilize the 4-km Weather Research and Forecasting (WRF) model to dynamically download ERA5 reanalysis data, creating an extensive hydroclimate reanalysis datasets spanning over 50 years (1969-2023) along with future climate projections for the data-sparse Southeast Asia (SEA) region. This effort is executed in coordinating with the WCRP CORDEX-SEA and My Climate Risk Regional Urban Risk Hub activities, aiming to mitigate the impacts of increasing weather and climate vulnerabilities in the densely populated coastal SEA megacities. This paper describes our general approach, optimized spectral nudging method, coupled with an enhanced land-surface/hydrology and urbanization modelling framework,  ensuring a refined representation of local and regional climate processes. We evaluate the performance of this fine-scale reanalysis in capturing key climatic phenomena, including tropical cyclones, extreme heat events and heavy precipitation in urban areas, as well as the evolution of water cycle components and drought conditions. The results underscore the dataset's ability in providing detailed insights into the dynamics of extreme weather events and their implications for urban risk management and climate adaptation strategies in Southeast Asia.

This study investigates the impact of climate change on extreme climate indices in Türkiye using a high-resolution dynamically downscaled simulation. The MPI-ESM1–2-HR global climate model is downscaled to 3 km resolution with the Weather Research and Forecasting (WRF) model using a nest configuration. The temperature and precipitation outputs are adjusted using the dynamically downscaled 3 km ERA5 reanalysis dataset that are obtained using the WRF model for the period 1980–2014. Quantile Delta Mapping is employed to adjust systematic biases, and future projections are analyzed under SSP2-4.5 and SSP5-8.5 scenarios. Results indicate significant regional variations in precipitation and temperature extremes. The total annual precipitation (PRCPTOT) shows little change under SSP2-4.5 but decreases 5% under SSP5-8.5, with a notable reduction up to 30% in the Mediterranean region, while eastern Türkiye and the Black Sea coast exhibit increases of around 10%. Consecutive dry days (CDD) increase from a historical mean of 65 days to 74 and 73 days under SSP2-4.5 and SSP5-8.5, respectively, with significant increases along the Mediterranean and Aegean coasts. The maximum one-day precipitation (Rx1day) increases nationwide, with a more significant 10% rise under SSP5-8.5. Similarly, contribution to total precipitation from very wet days (R95pTOT) increases from 20% to 22% and 26% under SSP2-4.5 and SSP5-8.5, respectively, exceeding 40% in some coastal regions under SSP5-8.5. Temperature extremes exhibit more spatially uniform trends, with the annual maximum temperature (TXx) increasing by 2.5°C and 5°C under SSP2-4.5 and SSP5-8.5, respectively, with localized increases exceeding 7°C in inland regions. These findings highlight substantial shifts in Türkiye’s climate extremes, emphasizing increased drought risks and intensified extreme precipitation events, particularly under high-emission scenarios. In addition, these findings demonstrate the importance of high-resolution simulations in capturing fine-scale variations in extreme climate events, which are often underestimated in coarser-resolution models.

Dynamical downscaling provides a physics-based approach to generating high-resolution climate data, which is essential for understanding regional and local climate variability. The coarse spatial resolution of large-scale climate datasets limits their ability to represent fine-scale climate features, particularly in regions with complex topography. To address these limitations, the nest configuration of Weather Research and Forecasting (WRF) model is used to downscale the ERA5 reanalysis dataset, produced by the European Centre for Medium-Range Weather Forecasts (ECMWF), from its spatial resolution of 0.25° (≃25 km) to 9 km and 3 km resolution for the period 1980–2023. The added value of high-resolution downscaling is evaluated through statistical comparisons with 235 station observations, using daily spatial correlation and bias metrics, along with the climate indices related to precipitation and temperature. Spatial correlation analysis highlights the improvements in high-resolution simulations. ERA5 shows moderate spatial correlations, with precipitation and temperature correlations of 0.81 and 0.92, respectively. WRF-9 km improves these to 0.85 and 0.95, while WRF-3 km achieves the highest agreement at 0.94 and 0.97. While ERA5 overestimates precipitation, WRF-9km and WRF-3km simulations reduce the overestimation but introduce regional variations. Results further show that ERA5 has a wet bias, underestimating consecutive dry days (CDD) at 48 days, while observations average 64 days. In contrast, WRF-9 km and WRF-3 km simulate 75 and 73 days, respectively. For consecutive wet days (CWD), ERA5 overestimates at 9 days, while WRF-9 km and WRF-3 km simulations align with the observed 6 days. The findings highlight both the benefits and limitations of high-resolution simulations, offering insights into their ability to capture fine-scale climate patterns and extremes. The results contribute to a better understanding of how dynamical downscaling can refine regional climate assessments and serve as a valuable framework for future climate projections.

As global warming intensifies, the frequency and intensity of extremely high temperatures are expected to increase. This will impact the production of photovoltaics (PVs), which are increasingly adopted as an effective alternative to replace fossil fuel–based energy sources and reduce CO₂ emissions. Furthermore, days account for a considerable proportion of days with high PV power potential (PVpot). Therefore, this study investigates changes in the PVpot on future extremely high temperature days over East Asia, a region with high greenhouse gas emissions and vulnerability to extreme climate. The analysis revealed that the PVpot for extremely high temperature days has recently increased in Korea, central China, South China, and Japan, while no considerable change was observed in PV hotspot areas. For PVpot projections, East Asia–averaged PVpot for extremely high temperature days was projected to decrease across all scenarios and future periods. The PVpot was predicted to decrease more substantially toward the late 21^st century for both summer mean and extremely high temperature days, with a larger magnitude of decrease expected under the high–carbon emissions scenario. By the mid–and late 21^st century, the PVpot for extremely high temperature days was projected to decrease in the PV hotspot areas, particularly in the regions of northern China and southern Mongolia, by up to -7.2%. Near–surface air temperature has been identified as a key driver of future changes in the magnitude of PVpot’s decrease for extremely high temperature conditions over East Asia. As the negative contribution of near–surface air temperature was expected to increase toward the late 21^st century, and become more pronounced under the high–carbon emission scenario, Based on the findings, this study is expected to provide new insights for the development of renewable energy policies in a future where extremely high temperatures are projected to increase.

The Philippines has consistently been in the World Risk Reports’ top ten most vulnerable countries in recent years, in part due to its exposure to extreme rainfall events. This is of particular concern to its rapidly urbanizing capital, Metro Manila (MM), where the mechanisms behind rainfall are still poorly understood. This research aims to identify key synoptic conditions and local factors that are linked to rainfall occurrence and to shed light on drivers behind observed long-term rainfall trends over this region. PCA was used to identify recurring synoptic patterns over the Philippines between 1991 to 2021. For each identified pattern, annual precipitation amount (PA), frequency (PF), and intensity (PI) values were calculated from available station data. The Mann-Kendall test was used to identify long-term trends in PA, PF, and PI for each synoptic pattern, and correlation analysis was then used to see if these trends were linked to MM urbanization in recent decades. Finally, idealized simulations using the WRF model were used to clarify the mechanisms behind the influence of urbanization on rainfall. Results reveal that rainfall over MM is primarily driven by synoptic conditions, particularly the southwest monsoon (SWM) season and passing low-pressure systems. However, rainfall occurs more frequently in the days leading up to the SWM season and when a high-pressure system is present over the northern region of the Philippines. Correlation analysis and the WRF model simulations partially link these trends to the influence of the expanding MM urban surface. The city seems to provide ideal conditions for convective activity, though the relative importance of the dynamic and thermodynamic effects varies depending on synoptic conditions. The results of this study should inform the decisions of policymakers and key stakeholders involved in disaster risk management and urban planning for the Philippines.

In recent years, global climate change and rapid urbanization have increased the risk of various environmental problems in cities. Urban greening, a type of green infrastructure, has two environmental-improving effects: A retention effect that reduces runoff and mitigates flood damage and a cooling effect that improves urban weather by lowering temperature. It is, therefore, attracting attention as a climate change adaptation measure. In this study, we targeted cities in East Asia, where urbanization is progressing rapidly, and used a simple land surface model, SMBM, to reproduce urban greening in the target cities. We then conducted sensitivity experiments for various greening conditions, such as climate, soil thickness, soil quality, and irrigation, to conceptually evaluate the water retention and cooling effects of urban greening. The SMBM model reproduced results from a demonstration experiment using rooftop greenery. The model adequately reproduced the water balance of planting beds in typical urban greening. This study showed that of the two environmental improvement effects of urban greening, the water retention effect is in dry areas.
In contrast, the cooling effect is dominant in humid areas. It was also suggested that the retention and cooling effects depend on the climate and are in a trade-off relationship. It was also confirmed that the greater the thickness of the soil layer in the planting bed, such as in park green spaces, the greater the effectiveness of both. Introducing these may improve the system's effectiveness.

In this study, we introduce the global urban geometric parameter dataset for use in global urban climate analysis. This dataset is designed to meet the input requirements of Single layer urban canopy model (Kusaka et al., 2001) and has a spatial resolution of 30 arcseconds in latitude and 45 arcseconds in longitude, including five building parameters (average building height, maximum building height, standard deviation of building height, plane area index, and frontal area index) as well as aerodynamic parameters (zero-plane displacement and aerodynamic roughness length) calculated according to Macdonald et al. (1998) and Kanda et al. (2013).To address the previous issue of being unable to construct an extensive dataset due to the lack of detailed building data, this study used JAXA Advanced Land Observing Satellite World-3D-30 m (ALOS AW3D30), a DSM with a spatial resolution of 30 m and some global building footprints datasets (e.g., Microsoft GlobalMLBuildingFootprints, OpenStreetMap, Google Open Buildings), to estimate urban geometric parameters in areas where detailed building height data is not available. Specifically, a 30 m resolution Normalized Digital Surface Model (nDSM) was generated from AW3D30 using the neighborhood analysis with similar operations in Huang et al. (2020), and building heights extracted nDSM were assigned to individual buildings based on building footprint datasets to calculate urban geometric parameters. The results of accuracy validation conducted for New York State in the United States showed that the RMSE values for average building height, roughness length, and displacement height were 7.35 m, 1.79 m, and 3.80 m, respectively. The details of the dataset will be presented in the conference.

 Urbanization driven by industrialization has significantly intensified the urban thermal environment. Among various mitigation strategies, cool roofs, which increase rooftop albedo, have been recognized as an effective countermeasure to reduce solar heating on the city and lower ambient temperatures. Previous studies using mesoscale weather simulation with uniform or categorical urban geometric parameters have demonstrated that cool roofs can reduce near-surface temperatures by approximately 0.3 to 3K (Imran et al., 2018). However, the effectiveness of cool roofs is inherently influenced by urban morphology, particularly building density. To accurately assess their effect, it is essential to incorporate both realistic urban morphological parameters and rooftop albedo into the simulations. We conducted one-month mesoscale weather simulations for August in Japan, integrating realistic roof albedo distributions and urban geometric parameters to evaluate the effect of cool roofs. We simulated two cases: case A increasing the rooftop albedo from 0.2 to 0.85, and case B increasing the actual rooftop albedo to 0.85 Whole Tokyo 23 ward are selected as cool roof area for both cases and the average of actual rooftop albedo over the area is 0.177.
 A month simulation yielded unexpected result that case A had slightly larger reduction of daily average near-surface temperature by 0.07 K despite the lower albedo change from the baseline simulation: the change in case A was 0.65 while that in case B was 0.67. This may arise from the chaotic nature of meteorological simulations. To clarify the pure cool roof impact without chaotic influences in temperature reduction, we have conducted ensemble simulations for the cases. At AOGS, we will present the realistic cool roof impact on Tokyo 23 ward, examining the underlying mechanism as well as spatiotemporal extent of their influence within and beyond the surrounding atmosphere.

Water security is always an issue in Pacific Island countries, where the El Niño/La Niña oscillation and its modulation have a significant impact on rainfall variability. The Republic of Kiribati, located in the equatorial Pacific, is also in a fragile water environment and relies almost entirely on rainwater for domestic use. There is therefore a need to identify future changes in precipitation due to global warming in these regions, but observational data are extremely scarce in these island nations. The present study, therefore, applies the bias correction (BC) method in the limited observational data and projects future precipitation using the statistical downscaling (SD) method, considering the sea surface temperature and atmospheric circulation changes derived from Global Climate Models (GCMs), as in a previous paper (AOGS2023:AS36-001) analyzed in Java Island, Indonesia. It should be noted that regional climate model simulations are not available for these regions, so we use the GCMs, which generally have poor reproducibility for precipitation changes. Preliminary analysis showed that the reproducibility of GCM-derived atmospheric fields, including the trade wind circulation in the central equatorial Pacific, is relatively good. On the other hand, the seasonal changes in precipitation at the Betio station (in South Tarawa Island, capital of the Republic of Kiribati) differ between the GCMs output and the observed values. Therefore, as before, different BC and SD methods should be applied. The analysis has been expanded to include Fiji and the Marshall Islands, where data is relatively easy to obtain.

High-resolution climate data is crucial for assessing heat stress risks, particularly in regions increasingly vulnerable to rising temperatures and humidity. This study employs a hybrid statistical downscaling method—Bias Correction Constructed Analogues with Quantile Mapping Reordering (BCCAQ)—to refine projections from five advanced CMIP6 global climate models (GCMs): EC-Earth3-Veg, EC-Earth3, NorESM2-MM, MRI-ESM2-0, and IPSL-CM6A-LR. These models were selected based on the DN22 performance ranking to ensure reliable climate projections. Our study area spans Vietnam and Cambodia, covering the geographical extent from 102°E to 110°E longitude and 8°N to 16°N latitude. We used daily observational data from the Multi-Source Weather (MSWX)-GloH2O dataset (0.1° resolution, 1981–2014) as our reference to evaluate model performance. Results indicate that BCCAQ substantially enhances the simulation of minimum, average, and maximum temperatures as well as relative humidity compared to the raw outputs from the GCMs. In particular, minimum temperature shows an exceptionally strong spatial correlation (greater than 0.95) across the region, while maximum temperature is more accurately captured along Vietnam’s central coastal areas. Furthermore, relative humidity biases are effectively reduced to near zero during the JJASO season, highlighting an improved representation of humid conditions. As a key application, the improved downscaled data are utilized to analyze heat stress conditions by jointly considering temperature and relative humidity. This enhanced climate information is vital for managing heat stress, mitigating extreme events, and informing local adaptation strategies in the densely populated and highly vulnerable regions of Southeast Asia.

As climate change influences the global atmospheric patterns, understanding its effects on renewable energy resources becomes much more crucial. In particular, wind and solar energies are highly sensitive to regional climate variations. The coarse-resolution models often fail to resolve the local-scale dynamics, which causes uncertainty in infrastructure adaptation and energy planning. Therefore, the localized effects should be captured by using high-resolution atmospheric simulations. This study uses the Weather Research and Forecasting (WRF) model to downscale the coarse CMIP6 MPI-ESM1-2-HR global climate projections (~100 km) to fine-scaled simulations with much higher resolution (~3 km) over Türkiye, climatological hotspot. The research investigates the significance of high-resolution climate modeling for producing accurate projections of renewable energy sources. The possible shifts in the renewable energy potential from the year 1950 to the end of 2100, are assessed under SSP2-4.5 and SSP5-8.5 future scenarios. These high-resolution simulations are conducted to examine the changes in wind, solar radiation, and the overall renewable energy production potential. This is particularly important for the areas with topography and variable conditions. The study's findings deepen our understanding of the possible effects of climate change on renewable energy sources and highlight the need for high-resolution simulations to make well-informed decisions during upcoming energy transitions.

Convection-permitting regional climate simulations over the Maritime Continent are carried out using the Singapore Regional Climate Model (SINGV-RCM) and the Weather Research and Forecasting (WRF) model at 8km and 2km spatial resolutions. Both RCMs are forced with ERA5 reanalysis data for 36 years (1979-2014) with same boundary forcing and similar domain setup at 8km resolution over the Maritime Continent (79E-160E; 16S-24N) including regular updates of the sea surface temperature at 6-hour intervals from ERA5. The 8 km simulation is subsequently used to force a nested 2 km simulation over the Western Maritime Continent (93°E–110°E; 7.2°S–9.9°N) for 20 years (1995–2014). Rainfall characteristics including the diurnal cycle and daily extremes over land and ocean from the two model simulations and the two domains are evaluated against the high-resolution IMERG V6B precipitation data available from 2001 to 2014. Both SINGV-RCM and WRF capture mean and extremes of rainfall well, however, with noticeable differences in the rainfall magnitude over both land and ocean. These findings highlight the importance of high-resolution and convection-permitting simulations for capturing the localized regional features. Key similarities and differences between the simulations from the two RCMs will be presented.

Mesoscale convective systems (MCSs) are a key driver of extreme precipitation across East Asia. However, the long-term variability of these systems remains insufficiently investigated due to observational limitations. Here, we assess the capacity of the Weather Research & Forecasting Model (WRF) convection-permitting model (CPM) to simulate MCS characteristics and trends over East Asia from 2001 to 2023 (June–September) by comparing simulations with high-resolution observational datasets for validation. To identify and track the MCS, we use precipitation and brightness temperature data. The model effectively captures fundamental MCS properties, including duration, total rainfall volume, and MCS movement speed. Additionally, the model reasonably represents the key evolutionary stages of MCSs, from initiation to maturity and dissipation. However, systematic biases exist: the model underestimates the size of MCSs and the frequency of meso-α systems, while overestimating meso-β MCS occurrences, as well as both mean and max precipitation intensities. The model successfully reproduces increasing (decreasing) trends in precipitation and MCS-associated precipitation over Manchuria and eastern China (Taiwan), but fails to capture observed trends over northern China and the Korean Peninsula. These discrepancies are attributable to the model's inability to simulate recent changes in mid-latitude moisture transport from the western North Pacific and the associated reduction in low-level moisture over land, as inferred from vertically integrated moisture flux and 850 hPa specific humidity trends. The mitigating of these biases will enhance the robustness of future convection-permitting studies on MCS variability, its sensitivity to natural variability and anthropogenic forcing, and its projected shifts under future climate change.