ENSO hindcast skill of the IAP-DecPreS near-term climate prediction system:comparison of full-field and anomaly initialization

Model initialization is a key process of climate predictions using dynamical models. In this study, the authors evaluated the performances of two distinct initialization approaches--anomaly and full-field initializations--in ENSO predictions conducted using the IAP-DecPreS near-term climate prediction system developed by the Institute of Atmospheric Physics （lAP）. IAP-DecPreS is composed of the FGOALS-s2 coupled general circulation model and a newly developed ocean data assimilation scheme called＇ensemble optimal interpolation-incremental analysis update＇ （EnOI-IAU）. It was found that, for IAP-DecPreS, the hindcast runs using the anomaly initialization have higher predictive skills for both conventional ENSO and El Nino Modoki, as compared to using the full-field initialization. The anomaly hindcasts can predict super El Nino/La Nina 10 months in advance and have good skill for most moderate and weak ENSO events about 4-7 months in advance.The predictive skill of the anomaly hindcasts for El Nino Modoki is close to that for conventional ENSO. On the other hand, the anomaly hindcasts at 1- and 4-month lead time can reproduce the major features of large-scale patterns of sea surface temperature, precipitation and atmospheric circulation anomalies during conventional ENSO and El Nino Modoki winter.

Stress initialization methods for dynamic numerical simulation of rock mass with high in-situ stress

In the context of deep rock engineering,the in-situ stress state is of major importance as it plays an important role in rock dynamic response behavior.Thus,stress initialization becomes crucial and is the first step for the dynamic response simulation of rock mass in a high in-situ stress field.In this paper,stress initialization methods,including their principles and operating procedures for reproducing steady in-situ stress state in LS-DYNA,are first introduced.Then the most popular four methods,i.e.,explicit dynamic relaxation(DR)method,implicit-explicit sequence method,Dynain file method and quasi-static method,are exemplified through a case analysis by using the RHT and plastic hardening rock material models to simulate rock blasting under in-situ stress condition.Based on the simulations,it is concluded that the stress initialization results obtained by implicit-explicit sequence method and dynain file method are closely related to the rock material model,and the explicit DR method has an obvious advantage in solution time when compared to other methods.Besides that,it is recommended to adopt two separate analyses for the whole numerical simulation of rock mass under the combined action of in-situ stress and dynamic disturbance.

MJO ensemble prediction in BCC-CSM1.1(m)using different initialization schemes

The Madden–Julian Oscillation（MJO）is a dominant mode of tropical intraseasonal variability（ISV）and has prominent impacts on the climate of the tropics and extratropics.Predicting the MJO using fully coupled climate system models is an interesting and important topic.This paper reports upon a recent progress in MJO ensemble prediction using the climate system model of the Beijing Climate Center,BCC-CSM1.1（m）;specifically,the development of three different initialization schemes in the BCC ISV/MJO prediction system,IMPRESS.Three sets of 10-yr hindcasts were separately conducted with the three initialization schemes.The results showed that the IMPRESS is able to usefully predict the MJO,but is sensitive to the initialization scheme used and becomes better with the initialization of moisture.In addition,a new ensemble approach was developed by averaging the predictions generated from the different initialization schemes,helping to address the uncertainty in the initial values of the MJO.The ensemble-mean MJO prediction showed significant improvement,with a valid prediction length of about 20 days in terms of the different criteria,i.e.,a correlation score beyond 0.5,a RMSE lower than 1.414,or a mean square skill score beyond 0.This study indicates that utilizing the different initialization schemes of this climate model may be an efficient approach when forming ensemble predictions of the MJO.

THE APPLICATION OF TIDAL SIGNAL EXCLUSION SCHEME FROM INITIALIZATION IN A GENERAL CIRCULATION MODEL

In this paper, some corrections was made to the assumption that the forcing is quasi-static, which is the basis of the nonlinear diabatic initialization scheme adopted by a global model T106L19. Thus the tidal signal is expressed and excluded from the initialization scheme. It shows that the new scheme captures the semi-diurnal pressure variation and is much closer to the uninitialized field. Compared with the standard initialization scheme, both the anomaly correlation coefficients and RMS of 500 hPa geopotential height simulated under the new scheme have improved significantly.

NUMERICAL EXPERIMENTS AND ANALYSIS OF DIGITAL FILTER INITIALIZATION FOR WRF MODEL

Initialization and initial imbalance problem were discussed in the context of a three-dimensional variational data assimilation system of the new generation"Weather Research and Forecasting Model". Several options of digital filter initialization have been tested with a rain storm case. It is shown that digital filter initialization, especially diabatic digital filter initialization and twice digital filter initialization, have effectively removed spurious high frequency noise from initial data for numerical weather prediction and produced balanced initial conditions. For six consecutive intermittent data assimilation cycles covering a 3-day period, mean initialization increments and impact on forecast variables are studied. DFI has been demonstrated to provide better adjustment of the hydrometeors and vertical velocity, reduced spin-up time, and improved forecast variables quantity.

Particle filter initialization in non-linear non-Gaussian radar target tracking

When particle filter is applied in radar target tracking, the accuracy of the initial particles greatly effects the results of filtering. For acquiring more accurate initial particles, a new method called “competition strategy algorithm” is presented. In this method, initial measurements give birth to several particle groups around them, regularly. Each of the groups is tested several times, separately, in the beginning periods, and the group that has the most number of efficient particles is selected as the initial particles. For this method, sample initial particles selected are on the basis of several measurements instead of only one first measurement, which surely improves the accuracy of initial particles. The method sacrifices initialization time and computation cost for accuracy of initial particles. Results of simulation show that it greatly improves the accuracy of initial particles, which makes the effect of filtering much better.

A NEW INITIALIZATION METHOD OF CLUSTER CENTERS

The problem of parameters selection for potential function used to initialize cluster centers is discussed, and two formulas are given for determining these parameters. Then a new potential function to initialize cluster centers is also given which is computational effective. Finally, a set of compared experiments is presented to show the effectiveness of the proposed methods.

Evaluation of Two Initialization Schemes for Simulating the Rapid Intensification of Typhoon Lekima (2019)

Two different initialization schemes for tropical cyclone(TC)prediction in numerical models are evaluated based on a case study of Typhoon Lekima(2019).The first is a dynamical initialization(DI)scheme where the axisymmetric TC vortex in the initial conditions is spun up through the 6-h cycle runs before the initial forecast time.The second scheme is a bogussing scheme where the analysis TC vortex is replaced by a synthetic Rankine vortex.Results show that although both initialization schemes can help improve the simulated rapid intensification(RI)of Lekima,the simulation employing the DI scheme(DIS)reproduces better the RI onset and intensification rate than that employing the bogussing scheme(BOG).Further analyses show the cycle runs of DI help establish a realistic TC structure with stronger secondary circulation than those in the control run and BOG,leading to fast vortex spinup and contraction of the radius of maximum wind(RMW).The resultant strong inner-core primary circulation favors precession of the midlevel vortex under the moderate vertical wind shear(VWS)and thus helps vortex alignment,contributing to an earlier RI onset.Afterwards,the decreased vertical shear and the stronger convection inside the RMW support the persistent RI of Lekima in DIS.In contrast,the reduced VWS is not well captured and the inner-core convection is weaker and resides farther away from the TC center in BOG,leading to slower intensification.The results imply that the DI effectively improves the prediction of the inner-core process,which is crucial to the RI forecast.

A STUDY ON THE PREDICT ABILITY OF GRAPES MODEL OVER SOUTH CHINA: COMPARISONS BY TWO INITIALIZATION CONDITIONS BETWEEN ECMWF AND NCEP

This paper aims to assess the performances of different model initialization conditions(ICs)and lateral boundary conditions between two global models(GMs),i.e.,the European Centre for Medium-Range Weather Forecasts(ECMWF)and National Centers for Environmental Prediction(NCEP),on the accuracy of the Global/Regional Assimilation and Prediction System(GRAPES)forecasts for south China.A total of 3-month simulations during the rainy season were examined and a specific case of torrential rain over Guangdong Province was verified.Both ICs exhibited cold biases over south China,as well as a strong dry bias over the Pearl River Delta(PRD).In particular,the ICs from the ECMWF had a stronger cold bias over the PRD region and a more detailed structure than NCEP.In general,the NCEP provided a realistic surface temperature compared to the ECMWF over south China.Moreover,GRAPES initialized by the NCEP had better simulations of both location and intensity of precipitation than by the ECWMF.The results presented in this paper could be used as a general guideline to the operational numerical weather prediction that uses regional models driven by the GMs.

Research on Initialization on EM Algorithm Based on Gaussian Mixture Model

The EM algorithm is a very popular maximum likelihood estimation method, the iterative algorithm for solving the maximum likelihood estimator when the observation data is the incomplete data, but also is very effective algorithm to estimate the finite mixture model parameters. However, EM algorithm can not guarantee to find the global optimal solution, and often easy to fall into local optimal solution, so it is sensitive to the determination of initial value to iteration. Traditional EM algorithm select the initial value at random, we propose an improved method of selection of initial value. First, we use the k-nearest-neighbor method to delete outliers. Second, use the k-means to initialize the EM algorithm. Compare this method with the original random initial value method, numerical experiments show that the parameter estimation effect of the initialization of the EM algorithm is significantly better than the effect of the original EM algorithm.

EXPERIMENTAL STUDY ON A DYNAMIC ASYMMETRICAL TYPHOON INITIALIZATION SCHEME BASED ON 4D-VAR

Axisymmetric bogus vortexes at sea level are usually used in the traditional bogus data assimilation （BDA） scheme. In the traditional scheme, the vortex could not accurately describe the specific characteristics of a typhoon, and the evolving real typhoon is forced to unreasonably adapt to this changeless vortex. For this reason, an asymmetrical typhoon bogus method with information blended from the analysis and the observation is put forward in this paper, in which the impact of the Subtropical High is also taken into consideration. With the fifth-generation Penn State/NCAR Mesoscale Model （MM5） and its adjoint model, a four-dimensional variational data assimilation （4D-Var） technique is employed to build a dynamic asymmetrical BDA scheme to assimilate different asymmetrical bogus vortexes at different time. The track and intensity of six surmner typhoons much influenced by the Subtropical High are simulated and the results are compared. It is shown that the improvement in track simulation in the new scheme is more significant than that in the traditional scheme. Moreover, the periods for which the track cannot be simulated well by the traditional scheme can be improved with the new scheme. The results also reveal that although the simulated typhoon intensity in the new scheme is generally weaker than that in the traditional scheme, this trend enables the new scheme to simulate, in the later period, closer-to-observation intensity than the traditional scheme. However, despite the fact that the observed intensity has been largely weakened, the simulated intensity at later periods of the BDA schemes is still very intensive, resulting in overly development of the typhoon during the simulation. The limitation to the simulation effect of the BDA scheme due to this condition needs to be further studied.

Application of a Typhoon Initialization Scheme Based on the Incremental Analysis Updates Technique in a Rapid Update Cycle System

Initialization of tropical cyclones plays an important role in typhoon numerical prediction. This study applied a typhoon initialization scheme based on the incremental analysis updates (IAU) technique in a rapid refresh system to improve the prediction of Typhoon Lekima (2019). Two numerical sensitivity experiments with or without application of the IAU technique after performing vortex relocation and wind adjustment procedures were conducted for comparison with the control experiment, which did not involve a typhoon initialization scheme. Analysis of the initial fields indicated that the relocation procedure shifted the typhoon circulation to the observed typhoon region, and the wind speeds became closer to the observations following the wind adjustment procedure. Comparison of the results of the sensitivity and control experiments revealed that the vortex relocation and wind adjustment procedures could improve the prediction of typhoon track and intensity in the first 6-h period, and that these improvements were extended throughout the first 12-h period of the prediction by the IAU technique. The new typhoon initialization scheme also improved the simulated typhoon structure in terms of not only the wind speed and warm core prediction but also the organization of the eye of Typhoon Lekima. Diagnosis of the tendencies of variables showed that use of the IAU technique in a typhoon initialization scheme is efficacious in resolving the spurious high-frequency noise problem such that the model is able to reach equilibrium as soon as possible.

Move a Tropical Cyclone with 4D-Var and Vortex Dynamical Initialization in WRF Model

Previous studies showed that 4 D-Var technique used for data assimilation could be modified for weather control. This study demonstrates the ability of 4 D-Var to influence the future path of a tropical cyclone by calculating perturbations in WRF simulation. Given the background error covariance matrix, the initial field is improved by the vortex dynamic initialization technique. Our results show that 4 D-Var can be applied to control the trajectory of simulated tropical cyclones by producing "optimal" perturbations. In the numerical simulation experiment of Typhoon Mitag in 2019, after this kind of weather control similar to data assimilation, the tropical cyclone moved obviously,and the damaging wind over the coastline weakened. The prediction results after the initial field modified by 4 D-Var have a great change, and the position of the tropical cyclone moved about 0.5° southeastward after assimilation,which misses the southeast coast of China. Moreover, the damaging wind is also weakened. Since the 4 D-Var is premised on the assumption that the model is perfect and does not consider the model error, then the research plan to consider model error and introduce new methods is discussed in the paper.

IMPACT OF THE INITIALIZATION ON MESOSCALE MODEL PREDICTION IN SOUTH CHINA

1 INTRODUCTION The initial state of the atmosphere is one of the key factors that affect the result of NWP. With the development of increasingly finer NWP, the quality of initial atmospheric state has been drawing more and more attention . GRAPES 3D- Var （Global and Regional Assimilation and Prediction Enhanced System , a three-dimensional variational data assimilation subsystem developed by the Chinese Academy of Atmospheric Sciences, makes a solution to the issue of NWP data vacancy in China. Owing to it, quantitative application of satellite and radar data in NWP has significant breakthroughs. With the assimilation system of GRAPES 3D-Var and GRAPES regional mesoscale model, this work compares a control and assimilation experiment with regard to a cold air surge affecting south China in late December 2004 and analyzes the sensitivity of mesoscale model forecast on initial values and the effect of initialization on the improvement of forecasting capabilities.

A New Population Initialization of Particle Swarm Optimization Method Based on PCA for Feature Selection

In many fields such as signal processing,machine learning,pattern recognition and data mining,it is common practice to process datasets containing huge numbers of features.In such cases,Feature Selection(FS)is often involved.Meanwhile,owing to their excellent global search ability,evolutionary computation techniques have been widely employed to the FS.So,as a powerful global search method and calculation fast than other EC algorithms,PSO can solve features selection problems well.However,when facing a large number of feature selection,the efficiency of PSO drops significantly.Therefore,plenty of works have been done to improve this situation.Besides,many studies have shown that an appropriate population initialization can effectively help to improve this problem.So,basing on PSO,this paper introduces a new feature selection method with filter-based population.The proposed algorithm uses Principal Component Analysis(PCA)to measure the importance of features first,then based on the sorted feature information,a population initialization method using the threshold selection and the mixed initialization is proposed.The experiments were performed on several datasets and compared to several other related algorithms.Experimental results show that the accuracy of PSO to solve feature selection problems is significantly improved after using proposed method.

Dynamic Analogue Initialization for Ensemble Forecasting

This paper introduces a new approach for the initialization of ensemble numerical forecasting： Dynamic Analogue Initialization （DAI）. DAI assumes that the best model state trajectories for the past provide the initial conditions for the best forecasts in the future. As such, DAI performs the ensemble forecast using the best analogues from a full size ensemble. As a pilot study, the Lorenz63 and Lorenz96 models were used to test DAI＇s effectiveness independently. Results showed that DAI can improve the forecast significantly. Especially in lower-dimensional systems, DAI can reduce the forecast RMSE by ~50% compared to the Monte Carlo forecast （MC）. This improvement is because DAI is able to recognize the direction of the analysis error through the embedding process and therefore selects those good trajectories with reduced initial error. Meanwhile, a potential improvement of DAI is also proposed, and that is to find the optimal range of embedding time based on the error＇s growing speed.

A New Weight Initialization Method Using Cauchy’s Inequality Based on Sensitivity Analysis

In this paper, an efficient weight initialization method is proposed using Cauchy’s inequality based on sensitivity analy- sis to improve the convergence speed in single hidden layer feedforward neural networks. The proposed method ensures that the outputs of hidden neurons are in the active region which increases the rate of convergence. Also the weights are learned by minimizing the sum of squared errors and obtained by solving linear system of equations. The proposed method is simulated on various problems. In all the problems the number of epochs and time required for the proposed method is found to be minimum compared with other weight initialization methods.

CUR Based Initialization Strategy for Non-Negative Matrix Factorization in Application to Hyperspectral Unmixing

Hyperspectral unmixing is a powerful tool for the remote sensing image mining. Nonnegative matrix factorization (NMF) has been adopted to deal with this issue, while the precision of unmixing is closely related with the local minimizers of NMF. We present two novel initialization strategies that is based on CUR decomposition, which is physically meaningful. In the experimental test, NMF with the new initialization method is used to unmix the urban scene which was captured by airborne visible/infrared imaging spectrometer (AVIRIS) in 1997, numerical results show that the initialization methods work well.

Study on the Hardware Initialization of Embedded System

An embedded system generally comprises gffour parts （embedded microprocessor unit, peripheral hardware equipment, embedded operating system, and user application）, and its core function is to complete the control of different equipments as well as necessary monitoring and management measures. Initialization is to set a variable as ＂default value＂, including system initialization, software initialization and hardware initialization. However, these three types of initialization are classified in accordance with different layers. This is studied in this paper.

ON USE OF LHN METHOD TO ASSIMILATE THE INTENSIFIED SURFACE PRECIPITATIONS FOR GRAPES_MESO MODEL INITIALIZATION

The quantitative precipitation forecast(QPF) in very-short range(0-12 hours) has been investigated in this paper by using a convective-scale(3km) GRAPES_Meso model. At first, a latent heat nudging(LHN) scheme to assimilate the hourly intensified surface precipitation data was set up to enhance the initialization of GRAPES_Meso integration. And then based on the LHN scheme, a convective-scale prediction system was built up in considering the initial "triggering"uncertainties by means of multi-scale initial analysis(MSIA), such as the three-dimensional variational data assimilation(3DVAR), the traditional LHN method(VAR0LHN3), the cycling LHN method(CYCLING), the spatial filtering(SS) and the temporal filtering(DFI) LHN methods. Furthermore, the probability matching(PM) method was used to generate the QPF in very-short range by combining the precipitation forecasts of the five runs. The experiments for one month were carried out to validate the MSIA and PM method for QPF in very-short range.The numerical simulation results showed that:(1) in comparison with the control run, the CYCLING run could generate the smaller-scale initial moist increments and was better for reducing the spin-up time and triggering the convection in a very-short time;(2) the DFI runs could generate the initial analysis fields with relatively larger-scale initial increments and trigger the weaker convections at the beginning time(0-3h) of integration, but enhance them at latter time(6-12h);(3) by combining the five runs with different convection triggering features, the PM method could significantly improve the QPF in very-short range in comparison to any single run. Therefore, the QPF with a small size of combining members proposed here is quite prospective in operation for its lower computation cost and better performance.