A digital lightning prototype system: DLPS/HUST

Abstract

Digital lightning is an important thematic branch of the digital earth. Lightning data are complex, momentary, discrete, dynamic and spatio-temporal in nature. Hence, making great progress in their management, analysis and service is challenging. Many studies have been directed towards this field all over the world. Some different relevant systems are currently available. However, most of them focus on a single or a few aspects of lightning applications, rather than taking an overall account of lightning detection, management, analysis and service from the strategic perspective of the digital earth. Lightning disaster also demands a better developed spatial information technology in this field than at present. The idea of the digital earth offers a reasonable scheme to address these issues. Based on reviews of the concept, development and progress of the digital earth, an integrated technical framework of a digital lightning prototype system (DLPS) is proposed, including its data, functions and implementation of subsystems, which aims at the facilitation, management, accessing, visualisation, application, release and sharing of lightning data. Some key technologies are also carefully discussed to facilitate the construction of the DLPS. This framework could provide a helpful guide to future lightning research and hazard mitigation, as well as enrich connotation on the digital earth. So far, there are still many difficulties in multi-level grid construction that need to be solved.

Keywords:

• digital lightning
• digital earth
• prototype system
• lightning spatial database
• lightning geographic information system
• lightning spatial information grid

1. Introduction

The digital earth endows human beings with a new angle of viewing the world, which can be deemed as a great leap for mankind in understanding the Earth. The appearance and construction of the digital earth is a profound revolution not only in science and technology but also in the quality of life (Chen and van Genderen 2008, Goodchild 2008).

The concept of the digital earth has been first proposed by Al Gore (1998). Its original aim is to integrate a tremendous amount of resources from the real world into a virtual earth. In theory, the digital earth is a high-level integration of earth science, space science and information science (Li 2004, Guo et al. 2009). Technically, it is an effective combination and application of several modern technologies, such as remote sensing (RS), geographic information science (GIS), global positioning system (GPS), database, wireless and wired networks, as well as virtual simulation. In other words, the digital earth can be regarded as an informational earth, containing the digitalisation, visualisation, networking and intellectualisation of physical earth data. In 1999, the first International Symposium on Digital Earth (ISDE) was held in Beijing. More than 500 delegates from 27 countries attended this symposium, which was a milestone in the globalisation advancement of the digital earth. The sixth ISDE held in Beijing in September 2009 was another milestone in the history of the digital earth. Guo et al. have described the 10 years of development of the digital earth (Guo et al. 2010). In August 2011, the seventh ISDE was convened at Perth in Australia. With the continuously increasing influences and applications of the digital earth, there have been great progresses in wide aspects of human life during this decade. Such progresses include those involving the sustainable development of societies and economies, environment and natural resources, precision agriculture, human life and intelligent traffic. The digital earth is not only a magnificent potential instrument for fusing different kinds of geospatial data but also a stepping stone towards building a comprehensive information home of mankind.

Currently, some digital earth branches have appeared, such as digital ocean (Zhang et al. 2011), digital city, digital power, digital river basin and so on. Digital lightning is a new research field of the digital earth. It is born with building an ‘information home’ and providing us a better understanding of lightning. The core idea of digital lightning is to apply the theories, methods and techniques used in digital earth to address today's large-scale lightning problems. The purpose is to study the spatio-temporal patterns of lightning and reduce the damage it causes to the minimum.

Lightning is one of the most powerful and spectacular natural phenomena that mankind has ever encountered. It is not only breathtakingly beautiful, but also treacherous and menacing at the same time. Every year, about three billion lightning flashes occur throughout the world. There are numerous documents showing that lightning is a threat to humans. In the UK, about three people are annually killed by lightning, and the number of people injured by lightning is about 50 (Elsom 2000). In the USA, there were about 2566 deaths and 6720 injuries caused by lightning from 1959 to 1985 (Duclos and Sanderson 1990). In South Africa, about 1.5 per million deaths per year in urban regions are caused by lightning. The rate is about 9 per million in rural areas (Meel 2007).

These tragic incidences have rendered lightning as a hot research topic in meteorology, disaster and information science. To reduce the danger and damage caused by lightning and make communities safer, worldwide research efforts have been directed toward lightning activity, particularly its detection and physical mechanism (Ramirez-Vazquez et al. 2004, Rakov et al. 2005). Many different relevant systems are currently available, such as the National Lightning Detection Network of the USA (Cummins et al. 1998) and the Boltek website (http://www.boltek.com/) that offers lightning application software. All these systems have achieved considerable and valuable progress. However, most of them only focus on a single or a few aspects of lightning applications, such as detection efficiency, data service or visualisation, rather than taking an overall account from the strategic view of the digital earth. In recent years, spatial information technology, which is widely used in such fields as environmental and disaster management, has rapidly developed. Some researchers have tried to apply the GIS technology to study the relationship between lightning activity and climatology (Stallins and Bentley 2006). Many useful products in lightning data analysis have been made, although they cannot yet meet the increasing demand for the information development of lightning. Based on the idea of the digital earth, the digital lightning prototype system (DLPS) is put forward. The DLPS tries to combine spatial database, GIS, spatial information grid, three-dimensional (3D) visualisation and computer networking with other spatial information technologies. The purpose is to provide a new scientific and reasonable way to deal with the acquisition, intelligent storage, integrated management, copious spatial and statistical analyses of lightning spatial data, as well as address various problems in the service and application of lightning data.

Based on the analyses of the characteristics and classification of lightning data, the present article discusses in detail the concept, aims and architecture of digital lightning. According to the concept of the digital earth, a corresponding DLPS designed and developed by the Huazhong University of Science and Technology (HUST), named the DLPS/HUST, is proposed. Some key technologies used to facilitate the construction of the prototype system are also described.

2. Lightning data

2.1. Characteristics of lightning data

Lightning events dynamically occur. The characteristics of lightning data are as follows.

2.1.1. Complex data types and large amounts of data

Lightning data contain the fundamental geographical data of the area covered by lightning monitoring network, the aerial remotely sensed image, the lightning thematic data, as well as various graphical data and reports generated from spatio-temporal analyses and operations. All these data may come from different sources. The fundamental geographical data are obtained from the National Geomatics Center of China (NGCC) as well as national surveying and mapping agencies. The lightning thematic data are acquired from field detecting stations as well as several professional processing and location algorithms about lightning. They are either the raster data stored in pixels, or the vector data represented by geometric figures and the attribute data that describe the corresponding spatial features. Both the coordinate system and the scale size of these data are not strictly uniform, resulting in the complexity of lightning data. The lightning thematic data are also updated in real time. Hence, the total amount of lightning data is enormous.

2.1.2. Highly dynamic and time sensitive

Lightning is not stationary; it changes dynamically over time. To ensure the accurate and effective detection of lightning, the status information of detecting sub-stations, original lightning records and lightning location data need to be fed back into the lightning spatial database and updated real time. Indeed, the dynamic effectiveness and time sensitivity of lightning data is very obvious.

2.2. Classification of lightning data

Lightning data with different formats and sources are mainly generalised into three key types in the proposed system: (1) lightning thematic data; (2) general geographical data that contain infrastructural data such as railways, roads and bridges, as well as administration distributions; and (3) other data relevant to lightning, such as the digital elevation model (DEM) and aerial orthographic images. In the following subsections, the data structure and contents of every type are illustrated in more detail.

2.2.1. Lightning thematic data

According to their contents, lightning thematic data can be classified into raw data, ripe data, detection status and spatial distribution of detecting stations. Raw data are a series of records detected by lightning detecting stations and obtained via network transmission. These data are un-located and temporarily stored as tables with lots of property fields for the data receiver of lightning (see Table 1a). With the aid of the positioning calculation of the lightning location system, the ripe data, a series of points with detailed positional information (such as longitude and latitude), are obtained. Ripe data are indispensable to lightning analysis and such applications as lightning early warning and super-short forecast. Ripe data are also stored as tables but have some different fields from those of raw data (see Table 1b).

Table 1. Schematic attribute tables for lightning thematic data.

(a) Lightning raw data
Frame Tag	Frame Size ID	Detector ID	Year	Month	Day	Hour	Minute	Second	Data Type	…
Frame code	Frame code	Detector code	yyyy	MM	dd	hh	mm	ss	0 or 1
(b) Lightning ripe data
Object ID	Date	Datetime	Polarity	Lightning type	Longitude	Latitude	Amplitude	Slope	…
ID code		hh:mm:ss	±	CC or CG
(c) Detector status
Detector ID	Year	Month	Day	Hour	Minute	Second	Self test result	Threshold	GPS state	AD error	…
Detector code	Yyyy	MM	dd	hh	mm	ss	0 or 1
(d) Feature class of the spatial distribution of the detector
Detector ID	Name	Longitude	Latitude	Latest state
Detector code				0 or 1

Notes: The raw data of lightning and detector status data are automatically received frame by frame. The property fields of raw data are different from those of ripe data. Lightning can be mainly classified into two types: cloud-to-cloud (CC) and cloud-to-ground (CG). CG can be further classified into positive (+) and negative (−).

To obtain lightning information, detector information should be contained in the database. There are two types of information about detectors, namely, non-spatial tabular data with several properties about the detector status information and the feature layer about the spatial distribution of the detector. One of the most important properties of detector status is the ‘result of self test’, which has two values, 0 and 1. ‘1’ represents a normal state and ‘0’ is abnormal (see Table 1c). The feature layer of the spatial distribution of the detector (see Table 1d) is added to the map when lightning activities are monitored. The layer consists of several detector features and has an attribute table with many property fields. The ‘latest state’ field needs updating in accordance with the detector status table. If the field value is ‘1’, the detector feature is highlighted in green. If the field value is ‘0’, the highlighting is red.

2.2.2. General geographical data

General geographical data mainly contain fundamental geographical and thematic information relevant to the serving objects. The former involves basic geographical features, including traffic lines (e.g. road, railroad, high-speed road), key river systems, administrations that contain administrative boundaries, local government and residential areas, as well as other public infrastructures. The latter is concentrated on the construction of different professional spatial attribute tables (e.g. forest and transmission lines) for thematic meteorological businesses.

2.2.3. Other data related to lightning

To further broaden the application of lightning spatial data, other vector and raster datasets can be added to the study. Lightning occurrence is usually influenced by the terrain. Hence, elevation data should be included, and these data can either be digitalised contour or DEM and triangulation irregular network format data, because they can be constructed from each other. Otherwise, additional data such as aerial orthographic, satellite and radar echo images should also be included in the proposed system. Multi-source geospatial data used in the DLPS are demonstrated in Figure 1.

Figure 1. Multi-source geospatial data relevant to the DLPS.

3. DLPS framework

3.1. Concept and aims of the DLPS

Digital lightning is an important branch of the digital earth. Based on the digital earth theory, digital lightning is an integrated application of several information technologies. It is also an information system that aims to fulfil the digitalisation of lightning detection, acquisition, storage, management, analysis, application and service based on the characteristics of lightning. This kind of system will be very valuable to different lightning users and meet the demands of society for lightning information. The overall objective of the DLPS is to build such an integrated system that can address issues in lightning information detection, management, comprehensive analysis and sharing service. The specific aims of the proposed system are as follows.

1. To achieve a lightning spatial database management system (DBMS) covering multi-source and heterogeneous data. The database system should be able to solve the problem of centralised data storage and management, ensuring data consistency, completeness, sharing and safety, as well as realising scientific and standardised data management. Consequently, the management level and efficiency of lightning is improved, and more scientific as well as reliable data services for other lightning thematic business systems are provided.

2. To construct a visual lightning analysis and application platform oriented to many thematic operational systems. The DLPS should provide two-dimensional (2D) and 3D displaying platforms for lightning essential and thematic data. Based on this, visual management, comprehensive analysis and multiple thematic applications are realised.

3. To implement a lightning spatial information grid (LSIG) system. Many different kinds of grids, such as multi-grain, regular and irregular, as well as many lightning-related subjects, should be constructed. For meteorology engineering and other important lightning protection areas, a window analysis operation could be performed. A new service pattern on lightning information products could also be provided.

4. To fulfil a release and sharing platform for lightning information oriented to communities and the general public. The DLPS could provide a lightning information sharing platform based on a network, as well as release the real-time lightning information, lightning forecasting and early warning products on the Internet.

3.2. General design of the DLPS

3.2.1. System architecture and key functions

To achieve the aforementioned goals of the proposed system, the entire architecture of the DLPS is classified into five tiers. These tiers are data acquisition, data storage and management, integrated analysis and display, thematic application and release, as well as sharing (see Figure 2).

• Tier 1, data acquisition tier.

Figure 2. Architecture of the DLPS.

Lightning detection instruments are responsible for collecting lightning flash records, namely, raw data, which are sent to the DLPS by a communication network. All detector devices must be strictly tested before they are utilised to ensure the quality of raw data. Various characteristic parameters of different types of lightning, such as time of lightning strike, current intensity of lightning, return stroke number and so on, are detected via an antenna array. This array is used to obtain lightning records of different waves, such as very low frequency, low frequency and very high frequency. These raw data are immediately processed by the optimal matching algorithm and professional model of lightning positioning to obtain real-time ripe data of lightning. Error checking is performed to control the quality of intermediate and ripe data. The third-generation wireless network, which is rapid and inexpensive, is used to transmit data.

• Tier 2, data storage and management tier.

The DLPS can realise the organisation of massive lightning data in a systematic way, as well as achieve a seamless integration of spatial and attribute data. The input and management of all kinds of lightning data, which mainly contain spatial and attribute data, can be fulfilled in a lightning spatial DBMS. Spatial data contain vector graphic features with geometric properties, such as general geographic data, ripe data (CC and CG) and historical lightning data, as well as raster data stored pixel by pixel (e.g. lightning optical data, background images and multiple graphical results data from various statistical analysis operations). Attribute data consist of the descriptions of the corresponding spatial data and lightning thematic tabular data, such as status tables of lightning sub-detectors and raw data (CC and CG) from sub-detectors. In this tier, two data input modes, automatic and manual, are provided to enable real-time updates of these highly dynamic lightning records and relatively stable general data. When a new lightning occurs, the lightning records are automatically detected. The intermediate and ripe data are also automatically updated in real time. Relatively stable general data are regularly renewed by the NGCC as well as national surveying and mapping agencies. The lightning spatial DBMS also supports multi-user concurrent use and offers two export modes: automatic and manual. To provide strong data-level support for other systems in the DLPS, methods such as user management, data archiving, data backup and recovery are applied to ensure database security.

• Tier 3, integrated analysis and display tier.

This tier is also the intermediate tier and is responsible for the comprehensive analysis and application of lightning data. The layer has three functions. The first function is to implement visual analysis and a display platform of lightning. This platform is developed based on 2D and 3D GIS development kits to achieve real-time tracking and display of lightning activities, fixed time period and animation playback of historical lightning data, visualisation of lightning product results, searching and retrieval, statistical analysis, spatial analysis, thematic maps, statistical charts and other forms of graphing output, as well as 3D simulation and roaming of a lightning scene. The comprehensive analysis of lightning data for relevant professional agencies, such as civil administration and meteorological agencies, is also supported. The second function is to achieve a LSIG platform. A LSIG system is developed based on computer grid and GIS technologies. This system mainly offers regular and irregular forms of grid construction under different scales. General data and lightning characteristics can be collected within a grid to form a lightning thematic grid. This system can also realise some GIS functions, such as search and retrieval based on the grid. By a window analysis of the lightning grid, the spatio-temporal distribution pattern of lightning can be studied. The third function is to offer releasable data for the upper release and sharing system, including lightning data and analysis results, such as live lightning map and window analysis results.

• Tier 4, thematic application tier.

This tier provides specialised models for lightning applications. It mainly includes lightning location, simulation, as well as forecasting and warning systems. The positioning algorithm and optimised model retrieves the up-to-date raw data, which are automatically processed and again archived into the database. Lightning simulation and forecasting systems pick up ripe data from database via a web-based interface. These ripe data are analysed using an integrated analysis platform and their own professional models. Professional analysis results, such as lightning forecasting and hazard warning maps, are also provided for the upper system.

• Tier 5, release and sharing tier.

This tier provides a window for lightning information release and sharing. More specifically, this layer is responsible for data sharing and release of lightning information, lightning product results, historical lightning data, queries and professional analysis results.

3.2.2. Data processing flow of the DLPS

As aforementioned in preceding sections, lightning data can be generalised into general, ripe, raw and result data. The first three types have been defined. Result data are the products of the comprehensive analysis and applications of general data and lightning thematic data based on the DLPS platforms, mainly including thematic maps, data tables, thematic grids and 3D spatial data.

The data processing flow of the four major categories in the DLPS is shown in Figure 3. First, a variety of relevant lightning data are collected. All types of raw lightning data (CC, CG and optical) can be detected by the station network. After the pre-processing of positioning computation and standardisation, these data are stored in the database. Second, these data are retrieved to realise lightning live display, query, statistical analysis, grid construction and other operations through platforms. Third, these operation results are stored in the database if necessary. Fourth, several relevant lightning analysis agencies use their professional models to achieve thematic applications according to their own special needs. Some professional and useful results are then formed. Fifth and finally, result data are released, fulfilling the sharing service on the network.

Figure 3. Data processing flow in the DLPS.

4. Implementation of the DLPS

According to the architecture and components of the DLPS aforementioned, a framework of the DLPS has been primarily implemented by the HUST. Visual Basic and C sharp programming languages within the.NET framework are chosen to facilitate the development of DLPS.

4.1. Lightning spatial database (LSDB)

Lightning events dynamically vary with the space and time, producing a large volume of data each day. The three-tier system configuration is regarded as fairly suitable for the database implementation of a large data volume, as well as for supporting the manipulation and analysis of temporary data processing (Al-Sabhan et al. 2003). Therefore, a three-tier architecture, with a front-end application tier, middleware tier and data/resource tier, is implemented in the proposed system (see Figure 4).

Figure 4. Three-tier architecture of lightning spatial database in the DLPS.

The data/resource tier is the most vital tier in the database and provides data-level support for the entire system. The powerful storage performance of the DBMS is utilised. Based on the structured query language, the front-end GIS applications interact with the database system via the middleware spatial data engine (SDE), application programming interface and ActiveX data objects, which contains remote data objects and data access objects.

4.2. Lightning geographic information system (LGIS)

The LGIS in the DLPS is a lightning-oriented geographic information system with basic functions of GIS and some advanced functions on thematic lightning. The LGIS contains both 2D and 3D platforms. In the 2D LGIS, lightning activities and detection status are monitored in real time. Historical lightning data can be queried and replayed in a dynamic way within a particular region by choosing the start and end time. Lightning characteristics such as thunder day, hour and density are important parameters in lightning research and decision-making. These parameters can also be determined on the side of the 2D LGIS. In the 3D LGIS, the 3D scene of basic terrain can be modelled and simulated by overlaying the remotely sensed imagery and the processed DEM data. Based on this, the 3D scene of thematic lightning can be reproduced from the real world in the ESRI® ArcScene® environment (see Figure 5).

Figure 5. 2D lightning real-time monitoring of the current day. The CG of the whole day is divided into four groups, rendered with different colours. The diamond symbol represents the state of the detector, the green and the red represent normal and abnormal, respectively.

4.3. Lightning spatial information grid

The LSIG in the DLPS is put forward for the first time in the fields of lightning application and service, which integrates GIS, spatial information grid and database techniques. The aim of the LSIG is to provide a better grid-based lightning analysis and service to help users find more spatio-temporal patterns of lightning. The LSIG offers multi-level grids. Before building the basic grid within a particular region, the grid grain should be defined. After lightning characteristic values are collected into the basic grid, the lightning thematic grid can be obtained. Subsequently, numerous spatio-temporal window statistics and analyses can be carried out on the LSIG side (see Figure 6).

Figure 6. Lightning thematic grid construction. The Hainan province is partitioned into 10×10 grids, 100 parts altogether. In this figure, the polygon is a regular rectangle, which can also be irregular. According to the mean thunderstorm days of different years stored in each grid, statistical analysis is carried out.

4.4. Lightning information release and sharing system (LIRS)

On the side of the LIRS (Figure 7), the user interface is based on a web-browser and will be developed with ActiveX controls as well as ArcServer secondary developer kits. The control of ActiveX, which is used with hypertext makeup language in this system, is a modular piece that performs tasks as well as communicates with other programs and modules over the Internet/intranet. Based on a hypertext transportation protocol request–response mechanism, people can browse and query up-to-date lightning activities released by the government, and then send their requests to share lightning spatial data via the website. The LIRS can provide a better understanding of lightning and consequently increase our awareness of protective measures against it.

Figure 7. Information flow of the LIRS.

5. Key technologies of the DLPS

5.1. Storage and security technology of massive spatial data

The DLPS contains not only actual lightning data but also other relevant space and ground data. Compared with traditional character-type data, these relevant lightning data have typical features in space, attribute, topology and time. When the data quantity reaches the terabyte level, a great challenge to data storage and security is posed. Hence, the storage and security technology of massive and heterogeneous spatial data is one of the key factors that must be considered in evaluating the practical value of a system.

The aforementioned technologies face many difficulties, such as searching for a better solution to the integrated organisation and storage problems of the multi-source, heterogeneous and massive lightning data. Another difficulty lies in finding a better scheme for overcoming database conflicts and hidden dangers brought about by multi-user concurrent access. Traditional databases cannot easily meet the storage requirements of massive spatial data. The combination of SDE technology and Oracle 10g relational database technology has shown tremendous advantages in spatial data storage. In the DLPS, this combined technology is also considered to achieve an effective organisation of massive spatial data as well as a seamless link between spatial data and attribute data. Regarding data conflict and security issues, these may be addressed by version control mechanism, automatic recovery and backup mechanism for ensuring the database security.

5.2. Customisation technology of thematic lightning GIS

Geographic information science is widely applied in meteorology. However, traditional information management systems are still commonly used in such lightning-related domains as monitoring and location, information management and application, spatio-temporal analysis, as well as nowcasting decision support. These traditional systems cannot meet the growing needs for lightning digitisation and visual management. Therefore, the GIS customisation technology of thematic lightning must be enacted to obtain powerful GIS advantages in data management and spatial analysis. ArcEngine is a flexible, application-neutral programming environment that comprises easy-to-build reusable program blocks and code bases for future extension. The ArcEngine technology may be used to build a customised GIS visualisation application system for lightning. The proposed setup does not require the complete ArcGIS Desktop environment, but only the ArcGIS Engine runtime library (taking about 301 MB of disk space) (Liang et al. 2007).

5.3. Multi-level grid construction oriented to lightning

Lightning spatial data cover a large range of areas. These complex data need to be scientifically manipulated to study more lightning patterns in space and time. Multi-level grid construction technology provides a heuristic solution.

Nevertheless, the construction of lightning multi-level grids is challenging. The lightning mesh generation standard needs to be defined, a probable grid-encoding method must be chosen and there must be a suitable representation for these multi-level grid data. Regarding the mesh generation standard definition, if the grain is too large, some key details may be lost. If the grain is too small, data redundancy results. Therefore, appropriate grain selection is essential in building a lightning thematic grid. Different levels of grids and the properties within a grid adopt a uniform encoding method, such as the classic quad-tree coding approach. The ‘level of detail’ technology can be used to represent a multi-level grid of lightning.

6. Conclusion and future work

The present article tries to integrate GIS, databases, RS, visual reality, 3D visualisation, spatial information grid and computer networking to build a DLPS from the perspective of the digital earth. Based on a careful investigation of existing researches on lightning, some basic knowledge and substantial progresses on digital lightning have been discussed. First, from the viewpoint of the digital earth application, digital lightning is a novel concept and application mode. The DLPS is an important step towards digital lightning from a mere concept to a realistic system. Second, from the perspective of data integration of the digital earth, the DLPS encompasses numerous types of data, such as lightning thematic data, general geospatial data, remotely sensed imagery and lightning product results. Third, from the standpoint of system application and service, the DLPS is deemed as an effective and integrated multi-platform tool for data management, complex analysis, 3D visualisation and information sharing of heterogeneous lightning data. Fourth and finally, for user groups, the highly targeted and practical DLPS offers a universal platform for lightning research by relevant agencies and the public, with the aim of elucidating the characteristics of lightning and the protective measures against it.

At the time of paper submission, the DLPS has been primarily implemented according to the proposed architecture and components. However, DLPS construction is a long-term and strategic task. The current article is a tentative study on the framework and development of the system, and hopes to provide guidance to the future lightning research and hazard mitigation. Although some key technologies of digital lightning have been discussed, there are still some technical problems. Techniques such as lightning signal analysis, feature extraction and data optimisation need further study. Therefore, there is a broad space in which to perfect the system in the near future.

Acknowledgements

The authors would like to appreciate the financial support for this study from the National Key Technology R&D Program of China (Grant#2008BAC36B01), the Program for New Century Excellent Talent of Ministry of Education of China (Grant #NCET-07-0340), the Natural Science Foundation for Outstanding Scholarship of Hubei Province in China (Grant#2008CDB364) and the National Natural Science Foundation of China (Grant #41072199 and 40672179). The authors also would like to thank the anonymous reviewers for their helpful comments and useful suggestions.