Zhng Sho-Qing,Yng LIU,M Xio-Hui,Wng Hong-N,Zhng Xu-Fng,Yu Xio-Lin,n Lu LvPhysil Onogrphy Lortory/CIMST,On Univrsity of Chin&Ntionl Lortory for Mrin Sin n Thnology,Qingo,Chin;Ky Lortory of Rsrh on Mrin Hzr Forsting,Ntionl Mrin Environmntl Forsting Cntr,Ministry of Nturl Rsours of th Popl's Rpuli of Chin,Bijing,Chin;Institut of Onology,Chins Amy of Sins&Ntionl Lortory for Mrin Sin n Thnology,Qingo,Chin;Ntionl Mrin Informtion Cntr,Ministry of Nturl Rsours of th Popl's Rpuli of Chin,Tinjin,Chin;Collg of On n Atmosphr,On Univrsity of Chin,Qingo,Chin

ABSTRACT The ‘Two Oceans and One Sea’area(West Pacific,Indian Ocean,and South China Sea;15°S–60°N,39°–178°E)is a core strategic area for the ‘21st Century Maritime Silk Road’project,as well as national defense.With the increasing demand for disaster prevention and mitigation,the importance of 10–30-day extended range prediction,between the conventional short-term(around seven days)and the climate scale(longer than one month),is apparent.However,marine extended range prediction is still a ‘blank point’in China,making the early warning of marine disasters almost impossible.Here,the authors introduce a recently launched Chinese national project on a numerical forecasting system for extended range prediction in the‘Two Oceans and One Sea’area based on a regional ultra-high resolution multi-layer coupled model,including the scientific aims,technical scheme,innovation,and expected achievements.The completion of this prediction system is of considerable significance for the economic development and national security of China.

KEYWORDS Numerical prediction system;ultra-high resolution;multi-layer coupled model;extended range prediction

1.Introduction

The ‘Two Oceans and One Sea’area(West Pacific,Indian Ocean,and South China Sea;15°S–60°N,39°–178°E)is a core strategic area for the ‘21st Century Maritime Silk Road’project,as well as national defense. Improving the forecasting accuracy for this sea area and extending the forecast period is of considerable significance for the economic development and national security of China.

With the increasing demand for disaster prevention and mitigation,the importance of 10–30-day extended range prediction,between the conventional short-term(around seven days) and the climate scale (longer than one month),has emerged(Ding and Liang 2010;He,Liang,and Sun 2013;Ma,Fan,and Li et al.2015;Tonani et al.2015;Yang 2015).However,owing to limitations both theoretical(i.e.,an incomplete understanding of the mechanisms of predictability) and technological (i.e.,modeling and computing power),marine extended range prediction is currently still a ‘blank point’in China,meaning services offering early warning 10–30 days in advance are difficult to achieve.In order to avoid the risks of natural disasters,derivative disasters,and secondary disasters,as well as improve the prediction accuracy of existing numerical platforms,it is necessary for us to study and develop extended range prediction.

Recent advances in scientific research have revealed that within the Earth system the multi-layer coupled(air–land–sea-ice)process is an important mechanism for the development of weather and climate events;plus,it is also a main mechanism for extending the forecast period from a short period of several days to a longer period.In particular,the nesting of high-resolution regional coupled models with lower-resolution global models can optimize the use of computational resources,allowing local-,small-,and medium-scale physical processes and their impact on large-scale circulation to be described effectively.Such an approach can also help us in constructing multi-layer coupled numerical prediction systems,and thus improve the prediction accuracy and extend the forecast period.

National and international studies about modeling and assimilation,which are the two most important aspects involved in numerical forecasting systems for extended range prediction,have mainly focused on the regional coupled modeling of the tropical Atlantic and Pacific(Wen,Chang,and Saravanan 2011;ZuideZuidema et al.2016;Ma et al.2016),regional coupled models and uncoupled assimilation(Stojanovi? et al.2013),regional coupled model development and application(Zou and Zhou 2013),and coupled data assimilation models applied in the Global Climate Prediction System and its quasi-operationalforecasting (Zhang et al. 2007; Chang et al. 2013).For extending and improving weather forecasting skills,the balance and coherence between the initial conditions in different model components is important.Many research centers and operational forecasting agencies throughout the world have begun to study similar coupled data assimilation models(Laloyaux et al.2016;Chatterjee et al.2013;Zheng and Zhu 2015),but have not yet used them in operational application.Besides the initial state,uncertain parameters in the physical schemes employed in coupled models are one of the most important error sources for extended range prediction.A coupled data assimilation scheme with enhanced parameter correction(DAEPC)has been designed to determine how to obtain a signal-dominant state-parameter covariance in order to effectively optimize the coupled model parameters using observations in different system components(Zhang et al.2012).Coupled model parameter estimation that uses observations to adjust the coupled model parameters has shown great potential to reduce model biases and improve the quality of numerical forecasting and prediction(Zhang 2011a,2011b),but it needs to be further examined before it can be applied operationally in global extended range prediction.

To date,operational prediction centers have not yet developed a marine extended range prediction system.Current climate prediction systems based on multi-layer coupled modeling(Table 1)do not include the mechanism of predictability at extended range time scales of 10–30 days,and their sea surface temperature(SST)forecasting results are not as good as those from regional extended range continuous prediction systems for the ‘Two Oceans and One Sea’area(assuming the current state stays the same in the future;Figure 1).

In the above context,this paper introduces a research project in which we will start by investigating the feedback of large-scale atmospheric and oceanic circulation affected by small-and mesoscale air–sea interaction,as well as its impact on extended range prediction,and then focus on downscaling methods for the air,land and sea component models,the coupling( flux exchange)algorithm,and the dynamical-downscaling coupled data assimilation method.Resolving these scientific and technique issues,we intend to develop a numerical forecasting system for extended range prediction in the ‘Two Oceans and One Sea’area based on a regional ultra-high resolution multi-layer coupled model,and provide corresponding forecast products.It is expected that the forecasting skill of this prediction system will attain a level that is similar to equivalent systems already developed in Europe and North America.The research project was launched in July 2017 and will run for three years until December 2020.

2.Scientific aims and technical schemes

2.1 Scientific aims

This project will focus on the effects of meso-and small-scale air–sea interactions on large-scale circulation,multi-component model development and the coupling( flux exchange)algorithm,atmospheric and oceanic data assimilation,dynamical downscaling coupled data assimilation,and prediction system operational application.Within the overall project framework, five sub-projects have been set up;the relationships between the five sub-projects(Figure 2)and the detail are described in the subsections below.

Table 1.The current status of coupled models and coupled assimilation in world modeling and prediction centers applied in quasioperational climate prediction.

Figure 1.The 20-day average SST anomaly correlation coefficients in 2012 between the NCEP’s Climate Forecasting System and satellite observations in the ‘Two Oceans and One Sea’region.

2.1.1 Development of the model

Figure 2.Relationships between the five sub-projects.

Based on recent studies using the Weather Research and Forecasting(WRF)model,the Regional Ocean Modeling System(ROMS),and new theories and methods in multi-scale air–sea interactions(Small,Tomas,and Bryan 2014;Piazza et al.2015;Ma,Fan,and Li et al.2015;Ma et al.2016),along with improved land modules and East Asian land surface data and referring to the coupler design of the Geophysical Fluid Dynamics Laboratory(GFDL)and National Center for Atmospheric Research(NCAR),an ultra-high resolution multi-layer coupled model will be developed for the‘Two Oceans and One Sea’and East Asia region.In the research and development of this ultra-high resolution coupled system,the mass,energy,and momentum conservation will be an important aspect.Based on the characteristics of the regional climate,sensitivity experiments involving key parameters will be carried out to optimize these parameters during the development of the model.Specifically,research on the parameterization schemes of the planetary boundary layer,microphysics,cumulus cloud,radiation,and the land surface will be carried out for the atmospheric component.Furthermore,through combination with data on vegetation type,soil type,leaf area index,and other refined land surface characteristics,the land process modules in the atmospheric model will be refined with two kinds of land surface databases.For the ocean model,the vertically mixed turbulence and bottom friction coefficient parameterization schemes will form the focus.During each coupling step,the mass,energy,and momentum conservation of the coupler will be guaranteed in the prediction system.

2.1.2 Effects of meso-and small-scale air–sea interactions on extended range prediction

Multiple-scale phenomena exist in the variations of the marine environment.Marine environmental variations are not only related to the large-scale air–sea interaction,but also to the mesoscale air–sea coupled processes.In the past,research has mainly focused on large-scale air–sea interaction,but little is known about the features and impacts of mesoscale sea–aircoupled processes.Recently,high-resolution satellite observations of wind fields and SST data have shown that,at the spatial scales of 100–1000 km,there are significant coupled perturbations between the atmosphere and ocean(Xie 2004;Chelton,Schlax,and Freilich et al.2004;Maloney and Chelton 2006;Small,Xie,and O’Neill et al.2008).For example,there are strong positive correlations between the perturbed wind stress amplitude,divergence and curl,and the perturbations of SST,the downwind SST gradient and crosswind SST gradient,respectively(Wei,Zhang,and Wang 2017).Previous studies have shown that mesoscale SST perturbation has a significant impact on the atmosphere(Nonaka and Xie 2003;Xie 2004;Minobe,Kuwano-Yoshida,and Komori et al.2008).On the one hand,SST perturbations can cause wind changes that can give rise to the variations of wind divergence at the bottom boundary layer and air convergence and divergence in the upper troposphere,subsequently affecting cloud formation and rainfall(Minobe,Kuwano-Yoshida,and Komori et al.2008;Putrasahan,Miller,and Seo 2013).On the other hand,the wind perturbation caused by mesoscale SST perturbations can also feed back to the variations in the marine environment.The variability of wind can cause the variations in sea surface momentum flux and heat flux(Nonaka and Xie 2003;Xie 2004;Small,Xie,and O’Neill et al.2008).Through the interactions of multi-scale oceanic processes,local ocean feedback to mesoscale wind field perturbation can affect the large-scale mean and low-frequency marine conditions(Hogg,Dewar,and Berloff et al.2009;Ma et al.2016).Therefore,we should focus on the characteristics of meso-and small-scale air–sea interactions,as well as their impacts on large-scale weather and climate processes.

2.1.3 Data assimilation method

Based on global multi-source atmospheric and oceanic real-time/near-real-time observations,atmospheric and oceanic data assimilation methods adapted to ultrahigh spatiotemporal resolution models will be studied to resolve problems associated with background error covariance localization and multi-scale co-assimilation techniques in dealing with satellite remote sensing and in-situ observations.An ultra-high resolution multi-scale data assimilation model with meso-and small-scale information will be developed,and the effect of the initial field precision on the numerical extended range prediction skill in the regional model will be assessed in detail.

2.1.4 Coupled data assimilation algorithm

Based on ultra-high resolution atmospheric and oceanic assimilation methods,we aim to develop a stable and reliable coupled data assimilation algorithm.Using the ultra-high resolution dynamical downscaling technique,the influence of the boundary conditions on the atmospheric and oceanic data assimilation in the regional coupled model will be studied,and then the coupled data assimilation algorithm of the dynamical downscaling will be developed.This sub-project will also involve studying the balance and coherence of the coupled model initial conditions in the different components,thus helping to develop the regional coupled data assimilation system with the dynamical downscaling technique.

2.1.5 Operational application and demonstration system

The main objective of this final task is to integrate the extended range prediction system and do the test run on operational application.Based on a high-performance computing platform,this sub-project will bring together the achievements form all the other sub-projects,including the coupled modeling and parameterization of meso-and small-scale processes and the coupled data assimilation system,to form a final marine extended range prediction system.This sub-project will also involve studying a series of technical issues within extended range operational forecasting,such as data services and forecast skill evaluation.Based on the extended range numerical prediction results of the operational prediction system,reanalysis data,and the assessment methods and indicators obtained from the above-mentioned studies,the accuracy of the prediction products and forecasting skill will be assessed using a variety of objective and diagnostic methods.While completing the quasi-operational experiments for no less than nine months,we will assess the predictive ability of the extended range prediction system and evaluate the skill in different regions.

2.2 Technical schemes

The WRF model and ROMS are the basic coupled components of the project.Additionally,new research achievements on multi-scale interaction processes(Chelton,Schlax,and Freilich et al.2004;Minobe,Kuwano-Yoshida,and Komori et al.2008;Small,Xie,and O’Neill et al.2008;Bryan et al.2010;Frenger et al.2013)are also utilized.With the newly released soil temperature and moisture data for the East Asian land surface, combined with the land surface module of the WRF model, we will develop a new air–land– sea coupled model. Referring to the coupler from the GFDL and NCAR,we will also improve the flux exchange algorithm to maintain the mass,energy and momentum conservation in this ultra-high resolution multi-layer coupled model.Based on the regional climate characteristics of key parameters in sensitivity experiments,we will achieve optimal selection of these model parameters.

At the same time,an analysis of small-and mesoscale air–sea interaction will be carried out.Using high-resolution satellite observational data,we will be able to study the main features of the meso-and small-scale air–sea interactions in the ‘Two Oceans and One Sea’region.The effects on large-scale oceanic and atmospheric circulations and the parameterization methods should provide an effective numerical solution for the ultra-high coupled model,and thus improve the quality of oceanic environment simulations and extended range prediction accuracy in the ‘Two Oceans and One Sea’region.

Secondly,coupled data assimilation will be studied based on a high accuracy atmospheric and oceanographic data assimilation algorithm.By using the ultrahigh resolution dynamical downscaling technique,we will study the influence of the boundary conditions on atmospheric and oceanographic data assimilation in the coupled model and develop a stable and reliable coupled data assimilation algorithm.We will then establish a regional coupled data assimilation system for the dynamical downscaling environment as the initialization subsystem of the extended range numerical prediction system.

Finally,we will complete the system integration,quasi-operational testing,and evaluation of the ultrahigh resolution and multi-layer coupled extended range prediction system for the ‘Two Oceans and One Sea’region.The ‘big data’processing technology of the realtime extended range prediction product will be studied and a visualization and monitoring system will be constructed(Figure 3).

3.Innovation and expected results

3.1 Scientific innovation

3.1.1 Theoretical innovation for extended range predictability

The innovative aspects in this respect will be our indepth studies on air–land–sea interactions and an evaluation of their impacts on large-scale circulation.Further evaluation on the joint initial–boundary value predictability in the regional coupled prediction system at extended range time scales,as compared to the traditional short-term initial-value problem,will be another key achievement.

3.1.2 Numerical model

A new type of ultra-high resolution multi-layer coupled model(different from the current short-term and climate models)will be constructed.The model will be able to describe the small-and mesoscale air–sea interaction processes and their effects on large-scale circulation.The multi-layer configuration will contain the regional atmosphere,land and ocean,with a horizontal resolution greater than 4 km.The intention is for it to be used as a numerical platform to study the impacts of small-and mesoscale physical processes on regional weather and climate processes.We expect the model to be able to describe small-and mesoscale physical processes effectively,especially the effects of small-and mesoscale air–sea interaction on large-scale circulation feedback and the necessary numerical parameterization schemes.All the above will form the core achievements necessary to solve key scientific problems in the construction of extended range prediction systems.

3.2 Technological innovation

3.2.1 New flux exchange algorithm

Figure 3.Summary of the scientific aims and technical schemes of the project.

The exchange of flux among the components in the ultra-high resolution multi-layer coupled model must maintain the conservation of mass,energy,and momentum.At the interface of the components,grid consistent processing will use the flux exchange technology to maintain the conservation of energy,quality,and momentum in the flux exchange processes.

3.2.2 Dynamical downscaling for the ultra-high resolution system

The boundary condition information of the regional coupled model will be transferred into the interior area by the model dynamics and physics.This will drive the fine-scale physical processes to form a detailed representation of the ocean and atmosphere in our study area.Given the differences between the relatively lower resolution boundary conditions and the relatively higher resolution regional grid structure,we will need to solve how to obtain better boundary condition information for the high-resolution model,as well as stabilize the high-precision model integration and effectively realize the dynamical downscaling in the region.

3.2.3 Dynamical downscaling for coupled data assimilation

Based on the dynamical downscaling effect,a stable and reliable atmospheric oceanographic data assimilation and coupled data assimilation algorithm will be developed in the ultra-high-resolution region.With those achievements,our studies will assess the impact of the balanced and coherent states of the model components on the extended range prediction and improvement in forecast skill owing to the regional coupled data assimilation using the dynamical-downscaling technology.

3.3 Expected achievements

A numerical forecasting system for extended range(forecasting period no less than 30 days)prediction in the ‘Two Oceans and One Sea’area,based on a regional ultra-high resolution(horizontal resolution better than 4 km)multi-layer(air–land–sea)coupled model,will be constructed.The system will provide extended range prediction products for multiple time scales in dynamic environments and complete the goal of pre-operational demonstration for more than nine months.The expectation is that,with this ultra-high-resolution multi-layer coupled model and advanced coupled data assimilation,the forecast skill will attain a level akin to that already achieved internationally.We expect the final achievements to include:

(1)A regionalultra-high resolution multi-layer coupled model for the‘Two Oceans and One Sea’area,with documentation on the main features of the small-and mesoscale air–sea interaction and the physical mechanism of feedback to large-scale circulation,and the provision of a related parametric program in the key area of the‘Two Oceans and One Sea’region.

(2)Advanced data assimilation algorithms for high resolution non-conventional observations.

(3)An advanced regional ultra-high resolution multilayer coupled data assimilation system.

(4)A numerical forecasting system for extended range prediction in the‘Two Oceans and One Sea’area and a forecasting product service platform.

Acknowledgments

We are grateful to the reviewers for their helpful comments.In this study,lots of information that showed the current status of coupled models and coupled assimilation were taken from the following websites:https://www.ecmwf.int/en/forecasts/docu mentation-and-support/extended-range-forecasts;http://www.cpc.ncep.noaa.gov/products/forecasts/;https://www.metoffice.gov.uk/research/modelling-systems/unified-model/weather-fore casting and https://weather.gc.ca/model_forecast/global_e.html.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work is supported by the National Key Research and Development Program of China(Grant Nos.2017YFC1404105,2017YFC1404100,2017YFC1404101,2017YFC1404102,2017 YFC1404103 and 2017YFC1404104).