As a preliminary attempt to cope with the low predictability of heavy rainfall over Southern China in the presummer rainy season, an experimental convection‐permitting ensemble prediction system (GM‐CPEPS) based on the Global/Regional Assimilation and Prediction System (GRAPES) is developed. GM‐CPEPS produces 12‐h forecasts at 0.03° horizontal resolution based on 16 perturbed members. Perturbations from downscaling, ensemble of data assimilation, time‐lagged scheme, and topography are combined to generate the initial perturbations. Sea surface temperature is perturbed and a combination of downscaling and balanced random perturbations is used to perturb the lateral boundary conditions. Stochastically perturbed parameterization tendencies, multi‐physics, and perturbed parameters are all implemented. In this study, GM‐CPEPS was verified over a 15‐day period during the Southern China Monsoon Rainfall Experiment (SCMREX) in May 2014. It was indicated that GM‐CPEPS provided estimates of forecast uncertainty that are comparable to some international peers. Compared with the control forecasts (DET), some deterministic guidance, including the forecast distribution with 90^th percentile, probability‐matched mean, and linear combination of both (NPM), showed advantages in forecasting moderate and heavy rainfall; and the optimal‐member technique was superior in reducing bias. Probabilistic guidance demonstrated the value over DET in detecting potential threats of severe weather, with both the optimal‐probability and neighborhood‐probability technique leading to improvements in predicting lighter rainfall. Two cases were used to display the deterministic and probabilistic guidance intuitively and to illustrate corresponding advantages and drawbacks.

The multiscale characteristics of initial conditions (ICs), model physics (MO), and lateral boundary conditions (LBCs) perturbations, as well as their interactions, are investigated for an experimental convection-permitting ensemble prediction system used in southern China, with a focus on 32 12-h forecasts collected over a half-month period of the Southern China Monsoon Rainfall Experiment (SCMREX) in May 2014. Forecast perturbations differed from each other in terms of magnitude, evolution, and vertical distribution between components at different scales. The meso-β-scale MO perturbations show faster growth and saturation, as well as larger magnitude than the meso-α-scale ones, especially in the presence of moist convection. For IC perturbations, damping was present for nonprecipitation variables, while rapid growth and saturation occurred for precipitation. Adding LBCs perturbations to IC or MO perturbations caused linear impacts, which caused consistent, although small, perturbation increments. Nonlinear impacts of adding MO perturbations to IC perturbations were evident and were intrinsic to the predictability limits of IC and MO perturbations, which were closely related to moist dynamics. These nonlinear impacts had the most significant effects on meso-β-scale precipitation perturbations and can effectively improve precipitation prediction. These results highlight the importance of developing multiscale and multisource perturbation methods that reasonably include the interactions among different-scale perturbations from different sources, especially the nonlinear impacts of adding MO perturbations to IC perturbations.

To improve short‐range tropical cyclone (TC) forecasting in China from both deterministic and probabilistic standpoints, a mesoscale ensemble prediction system (TREPS) based on the Global/Regional Assimilation and Prediction System (GRAPES) is constructed as a planned operational system.TREPScomprises one unperturbed and 30 perturbed forecasts ranging up to 60 h at 0.09° horizontal resolution. Downscaling perturbations from the European Centre for Medium‐range Forecasts' (ECMWF) ensemble prediction system (ENS) are used to generate initial perturbations by blending them with balanced random perturbations implemented around theTCand to generate perturbations for lateral boundary conditions. The surface temperature around theTCis also perturbed, with a combination of multi‐physics and stochastically perturbed parametrization tendencies considered to represent uncertainties in the model physics. These perturbation methods contribute a full‐area, multi‐scale and multi‐variable mutual complement to increasing forecast perturbations.TREPSis evaluated for 19TCsmaking landfall in China during 2014–2016. For deterministic guidance,TREPSsignificantly improvesTCforecasting for intensity, rainfall and wind compared toECMWFdeterministic and ensemble forecasts.TREPSdemonstrates significant advantages over its deterministic forecasts in predicting track, intensity, (extremely) heavy rainfall and wind. Furthermore,TREPSeffectively reduces the degree of underdispersion in intensity forecasting compared toENS. For probabilistic forecasting,TREPSoutperformsENSin accurately predicting extremely heavy rainfall and (extremely) strong wind owing to its better discriminating ability, but performs worse in predicting modest wind, light and heavy rainfall. Two post‐processing techniques – optimal member and optimal probability – to improve, respectively, deterministic and probabilistic guidance ofTREPSin rainfall and wind forecasting are proposed. The former is shown to be effective only for light rainfall, while the latter demonstrates consistent (partial) improvements in rainfall (wind). The superiorities and inferiorities ofTREPS, as well as possible reasons for these, are intuitively illustrated in a case‐study forRammasun.

This study investigates the impacts of different perturbation methods on the multiscale interactions between different‐source perturbations and precipitation forecasting in convection‐permitting short‐term forecasts. Initial Condition (IC) and MOdel physics (MO) perturbations are applied to a convection‐permitting ensemble prediction system (CPEPS) in southern China for the Southern China Monsoon Rainfall Experiment (SCMREX) in May 2014. The analysis presented in this study is based on 32 12‐hr forecasts and builds upon perturbation methods explored in the previous work. Different perturbation methods show different multiscale characteristics especially for four factors: the meso‐β‐scale saturation rates and saturation values for MO perturbations and the forecast‐perturbation magnitudes and changes in convective activities relative to the control member for IC perturbations. All of these factors are closely associated with the occurrence of dispersion reduction due to adding MO perturbations to IC perturbations. Such factors can be changed to different degrees by modifying the initial magnitudes of IC perturbations and by combining various types of perturbation methods. The faster meso‐β‐scale saturations and smaller meso‐β‐scale saturation values for MO perturbations, as well as the larger forecast‐perturbation magnitudes and larger enhancements of convective activities relative to the control member for IC perturbations boost the dispersion reduction. To improve precipitation forecasting, it is instructive to apply dispersion reduction to the design of perturbation methods for CPEPSs, where forecast errors have been overestimated by IC perturbations especially at smaller scales.

This study examines the case dependence of the multiscale characteristics of initial condition (IC) and model physics (MO) perturbations and their interactions in a convection-permitting ensemble prediction system (CPEPS), focusing on the 12-h forecasts of precipitation perturbation energy. The case dependence of forecast performances of various ensemble configurations is also examined to gain guidance for CPEPS design. Heavy-rainfall cases over Southern China during the Southern China Monsoon Rainfall Experiment (SCMREX) in May 2014 were discriminated between the strongly and weakly forced events in terms of synoptic-scale forcing, each of which included 10 cases. In the cases with weaker forcing, MO perturbations showed larger influences while the enhancements of convective activities relative to the control member due to IC perturbations were less evident, leading to smaller dispersion reduction due to adding MO perturbations to IC perturbations. Such dispersion reduction was more sensitive to IC and MO perturbation methods in the weakly and strongly forced cases, respectively. The dispersion reduction improved the probabilistic forecasts of precipitation, with more evident improvements in the cases with weaker forcing. To improve the benefits of dispersion reduction in forecasts, it is instructive to elaborately consider the case dependence of dispersion reduction, especially the various sensitivities of dispersion reduction to different-source perturbation methods in various cases, in CPEPS design.

To improve the ensemble prediction system of the tropical regional atmosphere model for the South China Sea (TREPS) in predicting landfalling tropical cyclones (TCs), the impacts of three new implementing strategies for surface and model physics perturbations in TREPS were evaluated for 19 TCs making landfall in China during 2014–16. For sea surface temperature (SST) perturbations, spatially uncorrelated random perturbations were replaced with spatially correlated ones. The multiplier f, which is used to form perturbed tendency in the Stochastically Perturbed Parameterization Tendency (SPPT) scheme, was inflated in regions with evident convective activity (f-inflated SPPT). Lastly, the Stochastically Perturbed Parameterization (SPP) scheme with 14 perturbed parameters selected from the planetary boundary layer, surface layer, microphysics, and cumulus convection parameterizations was added. Overall, all these methods improved forecasts more significantly for non-intensifying than intensifying TCs. Compared with f-inflated SPPT, the spatially correlated SST perturbations generally showed comparable performance but were more (less) skillful for intensifying (non-intensifying) TCs. The advantages of the spatially correlated SST perturbations and f-inflated SPPT were mainly present in the deterministic guidance for both TC track and wind and in the probabilistic guidance for reliability of wind. For intensifying TCs, adding SPP led to mixed impacts with significant improvements in probability-matched mean of modest winds and in probabilistic forecasts of rainfall; while for non-intensifying TCs, adding SPP frequently led to positive impacts on the deterministic guidance for track, intensity, strong winds, and moderate rainfall and on the probabilistic guidance for wind and discrimination of rainfall.