Japan Meteorological Research Articles

A framework for transformation to nearshore wave from global wave data using machine learning techniques: Validation at the Port of Hitachinaka, Japan

The present study introduces a framework for predicting nearshore waves using two machine learning techniques of Group Method of Data Handling (GMDH) and Artificial Neural Network (ANN), trained with three global wave datasets of Japan Meteorological Agency (JMA), National Oceanic and Atmospheric Administration (NOAA), and European Centre for Medium-Range Weather Forecasts (ECMWF). Prior to our ultimate goal of forecasting nearshore waves up to one week in advance, the current study challenges to hindcast nearshore wave heights and periods for a target year at the Port of Hitachinaka, Japan using the framework compounding GMDH and ANN trained with the initially forecasted (0 h) and reanalyzed two datasets. It was found that the GMDH-based wave model, trained with NOAA and ECMWF, well predicted observed significant wave heights, while a combination of JMA and ECMWF for training gave the best performance for significant wave periods. The same tendency was found when using ANN. Since the present framework successfully transforms global waves into local nearshore waves, it can be said that the framework for the nearshore wave prediction is able to support the one week ahead wave prediction and to be implemented at a particular location, where the nearshore wave observations are available.

An inter-comparison performance assessment of a Brazilian global sub-seasonal prediction model against four sub-seasonal to seasonal (S2S) prediction project models

This paper presents an inter-comparison performance assessment of the newly developed Centre for Weather Forecast and Climate Studies (CPTEC) model (the Brazilian Atmospheric Model version 1.2, BAM-1.2) against four sub-seasonal to seasonal (S2S) prediction project models from: Japan Meteorological Agency (JMA), Environmental and Climate Change Canada (ECCC), European Centre for Medium-range Weather Forecasts (ECMWF) and Australian Bureau of Meteorology (BoM). The inter-comparison was performed using hindcasts of weekly precipitation anomalies and the daily evolution of Madden–Julian Oscillation (MJO) for 12 extended austral summers (November–March, 1999/2000–2010/2011), leading to a verification sample of 120 hindcasts. The deterministic assessment of the prediction of precipitation anomalies revealed ECMWF as the model presenting the highest (smallest) correlation (root mean squared error, RMSE) values among all examined models. JMA ranked as the second best performing model, followed by ECCC, CPTEC and BoM. The probabilistic assessment for the event “positive precipitation anomaly” revealed that ECMWF presented better discrimination, reliability and resolution when compared to CPTEC and BoM. However, these three models produced overconfident probabilistic predictions. For MJO predictions, CPTEC crosses the 0.5 bivariate correlation threshold at around 19 days when using the mean of 4 ensemble members, presenting similar performance to BoM, JMA and ECCC. Overall, CPTEC proved to be competitive compared to the S2S models investigated, but with respect to ECMWF there is scope to improve the prediction system, likely by a combination of including coupling to an interactive ocean, improving resolution and model parameterization schemes, and better methods for ensemble generation.

Does air ionization by radon cause low-frequency atmospheric electromagnetic earthquake precursors?

The aim of this work is to study the relationship between the pre-earthquake emissions of radon and ULF/ELF (1–30 Hz) atmospheric electromagnetic radiation. The problem is considered on the example of the 2011 Tohoku earthquake. Radon, ionizing air, creates ions—centers of condensation of water vapor. As a result of condensation, heat is generated. It results in growth of air temperature and decrease in its humidity. This phenomenon serves as an indicator of air ionization. We used data from 20 Japan Meteorological Agency (JMA) weather stations located on Honshu Island to estimate any changes in temperature and humidity over ± 20 days from the date of the main shock. At the same time, we monitored the intensity and location of the source of ULF/ELF radiation using three induction magnetometers belonging to Chubu University. We compared the times and locations of observed signs of ionization and electromagnetic radiation to find out their relationship. It turned out that they are independent, since their dates and localizations do not match. In addition, we found intense ionization of the air after March 11 over a large area of Honshu Island, caused by radiative radiation from the nuclear disaster at the Fukushima Daiichi nuclear power plant. However, this phenomenon did not cause low-frequency atmospheric electromagnetic radiation either. These suggest that there is no direct relationship between air ionization and ULF/ELF radiation. This is true at least for this case, given the island nature of the land and oceanic EQs.

Variational Data Assimilation System for Operational Regional Models at Japan Meteorological Agency

Variational Data Assimilation System for Operational Regional Models at Japan Meteorological Agency

Improvements in tropical precipitation and sea surface air temperature fields in a coupled atmosphere–ocean data assimilation system

AbstractA coupled atmosphere–ocean data assimilation system, the Meteorological Research Institute‐Coupled Data Assimilation System Version 1 (MRI‐CDA1), was developed based on the coupled atmosphere–ocean general circulation model and separate atmosphere and ocean analysis routines operated by the Japan Meteorological Agency (JMA). To implement coupled atmosphere–ocean data assimilation, 6‐hr data assimilation cycles with the coupled model in the outer loop are adopted in atmospheric data assimilation, whereas incremental analysis updates with 10‐day data assimilation cycles are adopted in ocean data assimilation. A coupled data assimilation (reanalysis) experiment (CDA‐Exp) is conducted using MRI‐CDA1 along with an uncoupled reanalysis experiment (UCPL‐Exp) in which the same atmospheric component of the coupled model is forced by prescribed sea surface temperature (SST) similarly to the JMA reanalysis, JRA‐55. Climatological precipitation and variations in precipitation and sea surface air temperature (SAT) are better represented in CDA‐Exp than in JRA‐55, which is brought about by the modification of the atmosphere model physics of the coupled model to better represent climate states. In CDA‐Exp, the SST adjustment to the atmosphere amplifies the lead/lag correlations between subseasonal variations of SST and precipitation. The atmosphere–ocean coupling generates SST variations associated with tropical instability waves, and the SAT field responds to SST variations. The SST–precipitation and SST–SAT relationships on the weather timescale are also recovered in CDA‐Exp, although they are hardly seen in UCPL‐Exp. The coupled model physics generates weather‐timescale SST variations consistent with the atmospheric state, and the atmospheric parameters respond to the SST variations through the coupled model physics. This study suggests that the benefits of coupled data assimilation would be more evident if the ability of MRI‐CDA1 to represent SST variations and its interaction with the atmosphere are improved.

Sea surface temperature (SST) anomalies of the Yellow and East China Seas in July of 2020

With the global warming, the long term variations of sea surface temperature and its anomalies in the Yellow and East China Seas, especially for the July of 2020 due to the abnormally torrential rain along Changjiang/Yangtze River Valley, have been investigated based on the Merged Satellite and In-situ Data Global Daily sea surface temperature (MGDSST) provided by the Japan Meteorological Agency (JMA) using the methods of composite and correlation analyses. The results suggest, contrary to warming anomalies in the North-western Pacifica Ocean, the sea surface temperature in the East China Sea is cooler around 0.5°C and that in the East China Sea is cooler around 1.3°C than the normal values. The sea surface temperatures approach the extreme low value in the Yellow Sea and East China Sea in July, and warm up to the normal year in August. In addition, the south-westerly summer monsoon over this region, is proposed to contribute the transport of Kuroshio and its pathway. The obvious westerly wind anomalies, correspond to the lower sea surface temperature over the Yellow and East China Seas in July of 2020, leads to a clear less heat advection from Kuroshio to this region. Further, the low sea surface temperature, leading a downward air motion with a convergence at near sea surface level, is helpful for the enhance of the westerly wind anomalies until the strong surface heat flux in August. This study suggests that the local horizontal circulation advection and net heat flux are also dominated on the heat content of the East China waters. Further quantitative studies are worth conducting.

Focal mechanisms and the stress field in the aftershock area of the 2018 Hokkaido Eastern Iburi earthquake (MJMA\u2009=\u20096.7)

The tectonic stress field was investigated in and around the aftershock area of the Hokkaido Eastern Iburi earthquake (MJMA = 6.7) occurred on 6 September 2018. We deployed 26 temporary seismic stations in the aftershock area for approximately 2 months and located 1785 aftershocks precisely. Among these aftershocks, 894 focal mechanism solutions were determined using the first-motion polarity of P wave from the temporary observation and the permanent seismic networks of Hokkaido University, Japan Meteorological Agency (JMA), and High Sensitivity Seismograph Network Japan (Hi-net). We found that (1) the reverse faulting and the strike-slip faulting are dominant in the aftershock area, (2) the average trend of P- and T-axes is 78° ± 33° and 352° ± 51°, respectively, and (3) the average plunge of P- and T-axes is 25° ± 16° and 44° ± 20°, respectively: the P-axis is close to be horizontal and the T-axis is more vertical than the average of the P-axes. We applied a stress inversion method to the focal mechanism solutions to estimate a stress field in the aftershock area. As a result, we found that the reverse fault type stress field is dominant in the aftershock area. An axis of the maximum principal stress (σ1) has the trend of 72° ± 7° and the dipping eastward of 19° ± 4° and an axis of the intermediate principal stress (σ2) has the trend of 131° ± 73° and the dipping southward of 10° ± 9°, indicating that both of σ1- and σ2-axes are close to be horizontal. An axis of the minimum principal stress (σ3) has the dipping westward of 67° ± 6° that is close to be vertical. The results strongly suggest that the reverse-fault-type stress field is predominant as an average over the aftershock area which is in the western boundary of the Hidaka Collision Zone. The average of the stress ratio R = (σ1 − σ2)/(σ1 − σ3) is 0.61 ± 0.13 in the whole aftershock area. Although not statistically significant, we suggest that R decreases systematically as the depth is getting deep, which is modeled by a quadratic polynomial of depth.

Ambient Temperature and Cardiorenal Connection in Elderly Patients with Stable Heart Failure.

Heart failure increases among the elderly; however, the influence of ambient temperature on cardiorenal function has not been well investigated. Patients (n = 110, mean age 82.9 years, 43 males) with stable heart failure and creatinine < 3.0 mg/dl were studied. Medical records, such as ejection fraction, B-type natriuretic peptide (BNP), and estimated glomerular filtration rate (eGFR) at each visit every 1-3 months were collected by the end-point for death, additional prescription to treat heart failure, or heart failure hospitalization. The ambient temperatures at each visit were obtained from the Japan Meteorological Agency. During the follow-up period (median 399 days and 7 visits), follow-up BNP showed a trend toward a positive correlation with the diurnal temperature range. After dividing into two groups by median baseline eGFR, follow-up BNP was positively correlated with minimum temperature (p = 0.039) and the diurnal temperature range (p = 0.007) in the Low-eGFR group but not in the High-eGFR group. Follow-up eGFR was negatively correlated with the ambient day temperature in both groups (p ≤ 0.002). Follow-up BNP was positively correlated with follow-up eGFR (p < 0.0001) only in the Low-eGFR group and not in the High-eGFR group, suggesting that BNP and eGFR increase in winter and BNP and eGFR decrease in summer in the Low-eGFR group. In conclusions, heart failure may be worsened by larger diurnal temperature range or in winter in patients with renal impairment. This population should be carefully managed in the clinic according to the ambient temperature.

Heavy Winter Precipitation Events with Extratropical Cyclone Diagnosed by GPM Products and Trajectory Analysis

Heavy precipitation events were identified during the cold seasons of 2014–2019 using two-day accumulated precipitation data at 137 stations of the Japan Meteorological Agency. The mechanisms for producing heavy precipitation regarding the structure of an occluding extratropical cyclone were analyzed using the products of the Dual-frequency Precipitation Radar (DPR) onboard the Global Precipitation Measurement (GPM) core satellite and trajectory analysis on the European Centre for Medium-range Weather Forecasts atmospheric reanalysis data. Upper-ranked events with heavy precipitation were predominantly caused by extratropical cyclones and were in mature stages. In the top 50 ranked events, three south-coast cyclones were nominated, and the relationships between developing the mesoscale precipitation system and airstreams were intensively diagnosed. Hourly precipitation changes at stations that recorded heavy precipitation were primarily affected by a combination of the warm conveyor belt (WCB), cold conveyor belt (CCB), and dry intrusion (DI). Wide-ranging stratiform precipitation east of the cyclone center was composed of low-level WCB over the CCB and upper WCB, and convective clouds around the cyclone center were associated with the upper DI over the WCB that provided an extreme precipitation rate at the surface, including the formation of a band-shaped precipitation system. Convective cloud activities also contributed to moist air advection over the stationary stratiform precipitation areas recognized as the upper WCB. DPR products also identified deep stratiform precipitation in the cloud-head area behind the cyclone center with mid-level (near-surface) latent heat release (absorption) with increased potential vorticity along the CCB, making feedback intensification of the cyclone possible.

One‐dimensional maximum‐likelihood estimation for spaceborne precipitation radar data assimilation

AbstractSpaceborne precipitation radar such as Global Precipitation Measurement (GPM)/dual‐frequency precipitation radar (DPR) provides valuable observations of precipitation systems in three dimensions. Assimilation of GPM/DPR data is becoming an important technique for improving the accuracy of forecasting to complement scarce ground‐based observations. This study presents a new, one‐dimensional maximum‐likelihood estimation (1D‐MLE) method developed by the authors that enables the estimation of relative humidity profiles according to a non‐Gaussian probability density function. By assimilating the estimated relative humidity profiles using a four‐dimensional variational (4D‐Var) method, mesoscale precipitation forecasts by the Japan Meteorological Agency (JMA) have been considerably improved. The displacement of forecast precipitation during a severe weather event, in particular, is improved significantly. It was found that forecasting accuracy was maintained for a narrow GPM/DPR swath and low revisit frequency by repeating the assimilation–forecast cycle. Since the effectiveness was confirmed, the JMA began assimilating GPM/DPR data using the 1D‐MLE approach from March 2016.

Data Validation and Mesoscale Assimilation of Himawari-8 Optimal Cloud Analysis Products

AbstractHimawari-8 optimal cloud analysis (OCA), which employs all 16 channels of the Advanced Himawari Imager, provides cloud properties such as cloud phase, top pressure, optical thickness, effective radius, and water path. By using OCA, the water vapor distribution can be inferred with high spatiotemporal resolution and with a wide coverage, including over the ocean, which can be useful for improving initial states for prediction of the torrential rainfalls that occur frequently in Japan. OCA products were first evaluated by comparing them with different kinds of datasets (surface, sonde, and ceilometer observations) and with model outputs, to determine their data characteristics. Overall, OCA data were consistent with observations of water clouds with moderate optical thicknesses at low to midlevels. Next, pseudorelative humidity data were derived from the OCA products, and utilized in assimilation experiments of a few heavy rainfall cases, conducted with the Japan Meteorological Agency’s nonhydrostatic model–based Variational Data Assimilation System. Assimilation of OCA pseudorelative humidities caused there to be significant differences in the initial conditions of water vapor fields compared to the control, especially where OCA clouds were detected, and their influence lasted relatively long in terms of forecast hours. Impacts of assimilation on other variables, such as wind speed, were also seen. When the OCA data successfully represented low-level inflows from over the ocean, they positively impacted precipitation forecasts at extended forecast times.

Himawari-8 Aerosol Optical Depth (AOD) Retrieval Using a Deep Neural Network Trained Using AERONET Observations

Spectral aerosol optical depth (AOD) estimation from satellite-measured top of atmosphere (TOA) reflectances is challenging because of the complicated TOA-AOD relationship and a nexus of land surface and atmospheric state variations. This task is usually undertaken using a physical model to provide a first estimate of the TOA reflectances which are then optimized by comparison with the satellite data. Recently developed deep neural network (DNN) models provide a powerful tool to represent the complicated relationship statistically. This study presents a methodology based on DNN to estimate AOD using Himawari-8 Advanced Himawari Imager (AHI) TOA observations. A year (2017) of AHI TOA observations over the Himawari-8 full disk collocated in space and time with Aerosol Robotic Network (AERONET) AOD data were used to derive a total of 14,154 training and validation samples. The TOA reflectance in all six AHI solar bands, three TOA reflectance ratios derived based on the dark-target assumptions, sun-sensor geometry, and auxiliary data are used as predictors to estimate AOD at 500 nm. The DNN AOD is validated by separating training and validation samples using random k-fold cross-validation and using AERONET site-specific leave-one-station-out validation, and is compared with a random forest regression estimator and Japan Meteorological Agency (JMA) AOD. The DNN AOD shows high accuracy: (1) RMSE = 0.094, R2 = 0.915 for k-fold cross-validation, and (2) RMSE = 0.172, R2 = 0.730 for leave-one-station-out validation. The k-fold cross-validation overestimates the DNN accuracy as the training and validation samples may come from the same AHI pixel location. The leave-one-station-out validation reflects the accuracy for large-area applications where there are no training samples for the pixel location to be estimated. The DNN AOD has better accuracy than the random forest AOD and JMA AOD. In addition, the contribution of the dark-target derived TOA ratio predictors is examined and confirmed, and the sensitivity to the DNN structure is discussed.

Statistical Analysis of Building Damage from the 2013 Super Typhoon Haiyan and its Storm Surge in the Philippines

In November 2013, Super Typhoon Haiyan (Yolanda) hit the Philippines. It caused heavy loss of lives and extensive damages to buildings and infrastructure. When collapsed buildings are focused on, it is interesting to find that these buildings did not collapse for the same reasons after the landfall of the typhoon and storm surge. The objective of this study is to develop a statistical model for building damage due to Super Typhoon Haiyan and its storm surge. The data were collected in collaboration with Tanauan Municipality, the Philippines. The data for the inundation map were obtained by field surveys conducted on-site to determine the cause of the damages inferred from satellite data. The maximum wind speed was derived from the Holland parametric hurricane model based on the Japan Meteorological Agency (JMA) typhoon track data and the inundation depth of storm surge was calculated using the MIKE model. Multinomial logistic regression was used to develop a model to identify the significant factors influencing the damage to buildings. The result of this work is expected to be used to prepare urban plans for preventing damage from future storms.

Application of the WRF model to the coastal area at Ise Bay, Japan: evaluation of model output sensitivity to input data

ABSTRACT WRF simulations were conducted for Ise Bay, Japan for January and July 2016 to evaluate sensitivity of model input and output above sea surface to the replacement of three default input datasets with region-specific input datasets. For atmospheric input data, a final analysis created by the National Centers for Environmental Prediction (NCEP-FNL) was replaced with a mesoscale analysis created by the Japan Meteorological Agency (JMA). Topography and land use dataset released by the US Geological Survey were replaced with dataset released by the GeoSpatial Information authority of Japan. For sea surface temperature (SST) data, NCEP-FNL was replaced with an analysis created by JMA. Of the three region-specific datasets, replacement of atmospheric data results in the largest improvements in the accuracy of simulated wind speeds in January and July and of temperature in July 2016. Improvements in model output accuracy over sea surface can be seen near the coastline by replacing topography and land use data. Replacement of SST data results in the largest improvements in simulated temperature accuracy in January 2016. Replacing all three default input datasets results in the largest improvement, and expands on results from previous studies that focused on the effects of replacing only one input data.

Decadal-scale variability of the North Pacific subtropical mode water and its influence on the pycnocline observed along 137\xb0E

Decadal-scale variability of the North Pacific subtropical mode water (STMW) and its influence on the pycnocline are examined by analyzing Japan Meteorological Agency (JMA) repeat hydrographic observations along the 137°E meridian from 1972 to 2019, with a particular focus on the summer season when the seasonal upper pycnocline develops above the STMW. The STMW appears between 20° and 32°N at 137°E, with the thickness varying on decadal timescales of approximately 9–15 years. Argo float observations suggest that the observed change in the STMW thickness originates in the wintertime mixed layer south of the Kuroshio Extension in the preceding year. The STMW has a substantial impact on the pycnocline. The presence of thick STMW shoals the upper pycnocline, occasionally concurrent with the deepening of the lower main pycnocline. The change is robust in the upper pycnocline, where the heaving of isopycnal surfaces occurs with density anomalies up near the surface. The subtropical front (STF) at subsurface depths, which is associated with a northward shoaling of the upper pycnocline and is maintained by the STMW in the climatology, also changes on decadal timescales. A thick STMW increases the northward shoaling of the upper pycnocline and intensifies the STF. On decadal timescales, the STF variations are accounted for by the STMW-induced change in the upper pycnocline slope. The change in the STF due to mode waters is consistent with previous findings from numerical models.

PUFF Model Prediction of Volcanic Ash Plume Dispersal for Sakurajima Using MP Radar Observation

In this study, a real-time volcanic ash plume prediction by the PUFF system was applied to the Sakurajima volcano (which erupted at 17:24 Japan Standard Time (JST) on 8 November 2019), using the direct observation of the multi-parameter (MP) radar data installed at the Sakurajima Volcano Research Center. The MP radar showed a plume height of 5500 m a.s.l. around the volcano. The height was higher than the 4000 m by the PUFF system, but was lower than the observational report of 6500 m by the Japan Meteorological Agency in Kagoshima. In this study, ash particles by the MP radar observation were assimilated to the running PUFF system operated by the real-time emission rate and plume height, since the radar provides accurate plume height. According to the simulation results, the model prediction has been improved in the shape of the ash cloud with accurate plume top by the new MP radar observation. The plume top is corrected from 4000 m to 5500 m a.s.l., and the three-dimensional (3D) ash dispersal agrees with the observation. It was demonstrated by this study that the direct observation of MP radar obviously improved the model prediction, and enhanced the reliability of the prediction model.

Continuity of Earthquake and Tsunami Monitoring by Japan Meteorological Agency under Critical Conditions

Abstract The Japan Meteorological Agency (JMA) is a governmental organization that has responsibilities for mitigation of natural disasters. JMA issues warnings and information about natural disasters, in addition to daily weather forecasts. When an earthquake occurs, JMA analyzes seismic data to issue an earthquake early warning and to warn of possible tsunamis when a tsunami is expected to strike coastal areas of Japan. During tsunami warning in effect, JMA monitors tsunami meters and updates the warning. JMA also provides several types of macroseismic information. To fulfill these responsibilities, JMA collects data from 4400 seismic intensity meters, 1800 seismometers, 400 tsunami meters, and 39 strainmeters. Monitoring must be continued even under difficult situations such as times following great earthquakes, volcanic eruptions, severe weather conditions, and pandemics. JMA has dual operations centers located in Tokyo and Osaka. When one loses functionality due to a disaster or infection, the other continues 24/7 operations including warnings and issuing other information. Disastrous situations often cause power and communication failures and insufficient numbers of technical specialists. Following the 2011 Tohoku-oki earthquake, JMA enhanced power and communication capabilities by adding large capacity batteries at each station and satellite communication links. During the coronavirus disease 2019 (COVID-19) pandemic, JMA has taken several measures to prevent technical specialists’ infection to continue the full range of functions for issuing of warnings and conveying needed information.

Tropical cyclone movement features in the western North Pacific region based on multi‐dimensional statistical analyses

AbstractIn order to protect and mitigate the disaster caused by tropical cyclone (TC) in the western North Pacific (WNP) region, further understandings of the characteristics of historical TCs are important. In this study, multi‐dimensional statistical analyses are carried out to reveal the TC movement features in the WNP. Two representative TC datasets provided by China Meteorological Administration (CMA) and Japan Meteorological Agency (JMA) are used to investigate the historical TC properties in the WNP covering a time duration from 1951 to 2018. TC movement features generally show the great consistency with respect to these two datasets. Based on the regional (zero‐dimensional analysis) and zonally averaged (one‐dimensional analysis) results, it is found that TCs move northwest with a small velocity at low latitudes, whereas they tend to drift northeast with a large velocity at high latitudes. The recurvature occurs near 25°N where the direction of zonal drift velocity reverses. Applying the bivariate normal distribution method (quasi‐two‐dimensional analysis), the randomness of the zonal component is found to be generally greater than that of the meridional component. Accordingly, relevant attentions, rather than universal ones, should be paid to forecast the TC movement in different regions. Regarding the complete two‐dimensional analysis, it is found that compared with other areas, TCs have a smaller magnitude of drift velocity but stronger intensity with higher passage frequencies in the area enclosed by 10–25°N and 110–140°E, where significant TC‐induced disasters are prone to occur, resulting in a serious threat to the safety of human beings and local societies.

Possibility of Wave Prediction by Deep Learning

Numerical wave prediction models require a large amount of computational power to timely complete the required calculations. Artificial Neural Networks (ANN) have been introduced to perform predictions at a lesser computational cost and increased processing speed. Deep learning and specifically Convolutional Neural Networks (CNN) have become accepted for various image recognition applications. Motivation for the examination of wave prediction by deep learning came from the success of CNN in vision applications and the similarity of meteorological weather grid data to visual images. This study investigates a deep learning technique using the Japan Meteorological Agency’s Grid Point Value Mesoscale Model to predict wave height and period. In particular, this study uses the Xception deep learning architecture with depthwise separable convolution to obtain improved wave height and period prediction over artificial neural networks, and gets overall success results.

The regional model‐based Mesoscale Ensemble Prediction System, MEPS, at the Japan Meteorological Agency

AbstractThe regional model‐based Mesoscale Ensemble Prediction System (MEPS) has been operational since June 2019 at the Japan Meteorological Agency (JMA). The primary objective of the newly operational MEPS is to provide uncertainty information for JMA's operational regional model, Mesoscale Model (MSM), which provides information to support disaster prevention and aviation safety. This article describes MEPS in detail and discusses issues to be addressed in the future. For effective evaluation of uncertainties in MSM, the forecast model in MEPS is configured in the same way as that in MSM, except for the initial and lateral boundary conditions. Initial perturbations for all 20 ensemble runs are generated by a linear combination of singular vectors (SVs) with three different spatial and temporal resolutions, with the aim of capturing multi‐scale uncertainties in the initial conditions simultaneously. The SVs from a global model are also used as lateral boundary perturbations to ensure consistency between the initial and boundary conditions of each ensemble member. The verification results showed that MEPS achieved the expected performance of an ensemble prediction system: the ensemble mean outperformed the control forecast with a good spread–skill relationship; moreover, the skill scores of probabilistic precipitation forecasts were evaluated as valid for rainfall of up to 30 mm·(3 hr)−1. In an additional experiment conducted without using the two smaller‐scale initial perturbations, the skill was substantially reduced compared with that of the original MEPS, especially for larger precipitation thresholds. Therefore, the smaller‐scale perturbations were essential to capture uncertainties associated with local heavy rainfall events.