Key Issues in Marine Environment Information Support Technology

Research on the operational forecasting technology for tide-bound water levels

Since China’s reform and opening up. China’s import and export trade has increased rapidly. Especially since the 21st century, the pressure on ships entering and leaving ports has become more and more salient. In addition, ports and waterway areas are also affected by abnormal weather events caused by global climate change [1, 2]. For instance, the positive and negative water elevation caused by weather, open sea signals and other factors has posed a daunting challenge to the safety of navigation through ports and waterways. Tide-bound water level refers to a high-tide water level enabling the navigation of ships at certain time intervals [3]. Tide-bound water-level value is often required for the design of seaports and the selection of waterways. Although the tide-bound water level value appears around the time when high-tide water happens, in most cases, due to many uncertain factors, such as weather, it still requires experience and complicated calculations to get the accurate water-level value, otherwise navigable water level may be designed to be too low, hindering the navigation, or an area of water may be dredged too deep, causing serious waste. Moreover, inaccurate knowledge of tide-bound water levels will affect the scheduling of ships and cause many risks under abnormal conditions. Therefore, an accurate value of tide-bound water level is important in many areas, such as shipping and port engineering, especially in the ports and waterways that have low water depth and large tidal range. However, as abnormal weather events increase, the accuracy and timeliness of conventional tide-bound water-level estimation methods can no longer meet the current demand for tide-bound water-level forecasting [4]. The main reason is that the physical basis of the empirical method is weak, and it is difficult to accurately understand the impact of changes in the weather and marine environment in a large area on the local tide-bound water level. Furthermore, the previous dynamic numerical model is subject to the data and calculation accuracy regarding water depth and topography. So, the model cannot replace the empirical formula to forecast the accuracy of tide level to some extent [5].

Therefore, the United States conducted research on real-time water level forecasting in the 1990s. China also carried out such research in the late 1990s. The basic method is to conduct numerical simulation by using wind fields (including forecasting wind fields), calculate the value of positive and negative water elevation, and generate forecast water-level data in combination with the corresponding astronomical tide level [6]. As the accuracy of wind field data is rather limited, the calculated value of positive and negative water elevation can be greatly different. By the time project implementation begins, the development of real-time water-level forecasting has not yet started. However, the project has improved the accuracy of positive and negative water- elevation forecasting in the 24 hours by using China’s coastal hydrometeorological measurement data that has the longest time span and highest accuracy and using highly refined coastal topography data in combination with the empirical statistical approaches and dynamic model calculation methods. In combination with tide-level forecasting through astronomical tide, the project has also improved the real-time forecasting capacity of the actual water level, providing an effective water-depth reference for vessels entering and leaving ports. During the storm surge, the value of positive and negative storm surge and changes in tide level can be estimated at any time, which can not only provide technical support for maintaining the operating efficiency of China’s offshore ports, but can also improve the safety and facilitate the development of China’s shipping industry [7]. The social and economic benefits thereof are clear.

Research on emergency response technology for marine oil spills

In recent years, China has accelerated the building of emergency response systems for marine oil spills, but there are still many problems.

There is a lack of marine oil-spill emergency response management systems [8]. At present, China’s maritime emergency management involves multiple departments including marine sectors, fishery sectors, environmental protection sectors, transport and maritime sectors, and customs and border defense sectors. In addition, coastal provinces, cities and departments only manage their adjacent waters by themselves, and it is difficult to form an integrated management pattern among different departments, provinces and cities.

Key Issues in Marine Environment Information Support Technology 3

If things continue this way, the comprehensiveness and uniformity of China’s marine management functions will be weakened, and it will be difficult to form an efficient and scientific management system for oil spill emergency response.

There is a lack of emergency response equipment and materials for marine oil spill accidents. According to reports, after the “7.16” accident in Dalian (two pipelines exploded on July 16, 2010 at Dalian Xingang Port), Liaoning’s Maritime Department promptly laid a 7,000-meter-long oil boom in the contaminated waters and organized nearly 20 ships for cleaning the waters. Then the department arranged to get oil booms (more than 2,000 meters long), oil absorption felts and other decontamination materials from Hebei, Shandong, Tianjin and other areas. However, they were still insufficient to respond to the emergency. It is understood that starting on .1 uly 18, Dalian mobilized 800 fishing boats to participate in oil collection. On the 20th, the number of fishing boats reached more than 1,200. Being one of the three largest oil-spill emergency equipment warehouses in China, Dalian Port is designed to be able to deal with oil spills by 1,000-ton ships. Generally, this port can control and clear oil spills caused by ships in near-shore waters. However, according to the feedback on the “7.16” accident, the port’s emergency response capability obviously fell short because it failed to promptly deal with the oil spill [9, 10]. Some experts pointed out that due to technical constraints, China’s ability to efficiently recover and remove spilled oil in a large area after a marine oil spill accident happens has been weak. For example, up to now, China still has no world-class emergency oil-recover}' vessel, let alone the number of oil-recovery devices that match the vessels. So, the lack of emergency response equipment and materials for marine oil-spill accidents seriously hinders the building of emergency response systems.

There is a lack of emergency response teams made up of professionals. The emergency team is responsible for dealing with the oil-spill accident, including a certain scale and quantity of pollutant clean-up equipment and instruments and well-trained operators. However, there is a lack of training for emergency team members in China. Due to the complexity of emergency treatment of oil spills, the support from a variety of disciplines, such as environment science, information science and medicine, is quite necessary. In addition, experts in relevant fields are needed to provide guidance and technical consulting services at crucial times during emergency response. Furthermore, necessary training for the relevant emergency team members is required.

Technical research on sediment transport in estuary and coastal areas

The interaction between waves and tidal currents in the near-shore area is more salient and complicated than in other areas [12]. Tidal current fields affect wave fields in two areas: 1) wave changes caused by current velocity and direction; and 2) the effect of tide level fluctuation on waves. The hydrodynamics of deep sea areas is dominated by the effect of currents: when tidal currents are in the opposite direction of waves, wave height increases; when tidal currents are in the same direction of waves, wave height decreases. In the shallow sea area, tide-level changes are relatively obvious; accordingly, the wave changes caused by tide-level changes are also obvious. As tide level rises and falls, wave height and tidal cycles change synchronously. The effect of waves on tidal currents is mainly reflected in radiation stress acting on circulation in shallow waters. At the same time, the bottom shear stress caused by waves also increases the bottom friction of tidal currents. Therefore, before the simulation of sediment movement, it is necessary to simulate the complex hydrodynamic processes in the near-shore areas, including the coupling of wind, waves and currents, and their interaction with near-shore buildings [13, 14].

Fundamentals of Fluid Mechanics and Marine Dynamics Modelling

Research content

To establish a high-resolution, long-time marine dynamics model, it is necessary to consider small-scale processes, large-scale phenomena, the fine resolution of the model and computational efficiency. Therefore, model selection is very important. In the modelling process, this book selects the advanced international numerical model, develops the atmosphere, marine wave and marine current modules, and integrates these modules into a marine dynamics model.

Atmosphere module

Introduction to the model

The atmosphere module is a high-resolution surface-wind forecasting system. The atmosphere model in the forecasting system adopts WRF, a mesoscale atmosphere model that can simulate the sea surface wind field under extreme weather conditions. In this model, dynamics and physical processes are very mature [15, 16]. Especially in the case where the typhoon process is taken into account, the WRF model can find the corresponding initial field-generation scenario [17, 18]. The model is completely compressible and non-static; and all control equations are written in the flux form. The terrain-following hydrostatic barometric coordinates are used as the vertical coordinates. Using the Arakawa C horizontal and vertical staggered grid can help improve the accuracy of high-resolution simulation. Time integration adopts the full-time splitting scheme; the outer loop Runge-Kutta technique involves a relatively large time step; the inner loop is the time integral of sound waves [19], which can allow for relatively large time steps and shorten the calculation time while ensuring the stability of the integral. In order to meet the needs of simulating the actual weather, the model has a set of physical processes and parameterization processes, including cloud microphysical processes, cumulus convective parameterization schemes, long-wave radiation, short-wave radiation, boundary layer turbulence, near-surface layer, land surface parameterization and sub-grid turbulence diffusion. The WRF model uses inherited software design, multi-level parallel decomposition algorithms, selective software management tools and intermediate software package structures. The WRF model’s horizontal grid can be accurate to 1 km or even finer, making the model a tool for accurately forecasting major weather characteristics (these characteristics included cloud scale in the past, now the}' also include synoptic scale and other scales). Moreover, the WRF model has advanced 3D variational assimilation technique (3DVAR), which can fully and effectively assimilate various data and information into the initial field of the model, so it can provide the model with more accurate initial value and greatly improve the quality of numerical forecast [15].

The kernel of the forecasting system uses the WRF-ARW Version 3.1.1 released on July 31, 2009. The model uses fully compressible, non-static Euler equations; the horizontal grid is the Arakawa C grid; the vertical coordinates are the mass-based, terrain-following r/ coordinates, and the т/ layer can be changed as needed [20]. The model framework involves the dynamic and physical processes of atmospheric changes. The physical model includes the microphysical model that describes phase transformation of water vapor and cloud physical processes; the cumulus parameterization scheme has taken into account the effects of cumulus clouds on sub-grid scales, a multi-layer land surface model has taken into account the soil and vegetation patterns from ground’s simple thermodynamic process; the planetary boundary layer model involves second-order turbulence closure schemes or non-local K-closure schemes; and the atmospheric radiation pattern has taken into account both the multi-spectral long-wave radiation scheme and the simple short-wave radiation scheme for cloud and ground surface radiation. The V3.1.1 version involves an additional radiation scheme RTMMG, including long-wave and short-wave radiation schemes. The long-wave radiation scheme was developed from the RRTM (Rapid Radiative Transfer Model) scheme in MM5; RRTMG also uses MCICA (Monte Carlo Independent Cloud Approximation) technology to effectively describe the role of sub-scale clouds [21]. In addition, the terrain drag effect of the model and the high-order boundary-layer turbulence closure scheme are added, and variable time steps can be used to shorten the integral time in single zone simulation.

WRF-ARW mainly consists of the following modules: (1) WRF Preprocessing System (WPS) Module: in this module, the simulated area is set, topographic data is interpolated into the simulated area and meteorological data from other models (such as global model) is interpolated into the simulated area. This module is designed to provide a background field for simulation. (2) WRF Data Assimilation (WRFDA) Module: the assimilation scheme (in this paper, 3DVAR is applied) is used to assimilate the objective data such as observations at stations, satellite data and radar data to improve the initial field and boundary conditions required for simulation. This module is optional. (3) Master program module for numerical simulation (ARW): this module is designed to generate the initial background field and time- varying side boundary conditions required for simulation; numerical integration equation. (4) Post-processing module: this module is designed to analyze and process the model’s output results and display them in graphic form [22, 23, 24].

At present, the dynamic framework and calculation scheme for the WRF model are quite effective and ideal for the simulation of mesoscale weather phenomena [25].

Model configuration

In order to explain the model configuration clearly, Bohai Sea in China is taken as an example, especially the area at Tianjin Port. Considering the sea area and resolution comprehensively, the double-nesting technique is adopted in the WRF simulation. In the coarser-resolution area (10'), the weather system of the entire sea area studied can be captured. In the higher-resolution area (2'), the high-resolution simulation and reproduction of the weather system of Bohai Sea (especially the sea area at Tianjin Port) can be realized.

System framework

The sea surface wind real-time forecast system is completely modular and mainly includes the following modules:

  • 1) Data collection and processing module.
  • a) Real-time download the forecast field data of global atmospheric model to provide initial field and time-varying boundary conditions for the model.
  • b) Real-time download the conventional radiosonde data, ground and ship observation data of global GTS.
  • c) Real-time download China’s offshore unconventional satellite radiation and observation data.
  • d) Conduct decoding, data quality control and format conversion of the observation data.
  • 2) Real-time forecasting module.
  • a) Assimilate the conventional and unconventional observation data on the basis of the forecast field data of global atmospheric model to provide initial field and time-varying boundary conditions for the model.
  • b) Running the numerical forecasting system.
  • 3) Post-processing module.
  • a) Extract the sea surface wind data from the WRF results.
  • b) Convert the sea surface wind data from Lambert projection coordinates to regular latitude and longitude coordinates.
  • c) Use GrADS grid point analysis and display system software to draw the sea surface wind at the latitude and longitude grid points into an image.
  • d) Process the sea surface wind at the latitude and longitude grid points into netcdf format.

The interconnection between the modules is controlled by Linux/UNIX- shell script program, which is fully automatic without manual intervention [26].

  • 4) Operation of the forecasting system.
  • a) Data acquisition.

All the data required by the forecast system are from the internet, and they are automatically downloaded with Linux/UNIX-SHELL scripts at fixed times long with download tools such as lftp and wget. After the data is downloaded, the format conversion and quality control of the data are carried out first, and then data assimilation with 3DVAR is carried out to provide initial field and time-varying boundary conditions for the forecast system. The data includes conventional ground observation data, conventional radiosonde data, ship observation data, satellite sea surface wind observation data and satellite radiation observation data. Their distribution is shown in Figure 2.1.

b) Data assimilation.

High-quality forecast results cannot be achieved without a high-quality initial field. Single-time 3DVAR is used for data assimilation to form the initial field of real-time forecast, assimilating observation data within 3 hours or less than 3 hours before and after the moment. Due to the short assimilation window, for the Bohai Sea, single-time 3D VAR can only use a small

Types of observation data used by the forecast system (top left

FIGURE 2.1: Types of observation data used by the forecast system (top left: ground and radiosonde observation; top right: ground automatic observation; bottom left: satellite sea surface wind observation; bottom right: satellite radiation observation)

amount of unconventional observation data, sometimes it uses no such data at all. In order to assimilate more unconventional observation data, the assimilation period should be extended, but the assimilation window for single-time 3DVAR should not exceed 6 hours at most, so a cyclic 3DVAR assimilation scheme can be designed to extend the assimilation period. The basic idea is to use the results of the previous 3DVAR to provide the initial field for the WRF model, and then the WRF model is integrated to the next assimilation moment followed by the next 3DVAR. Such a process should be repeated. Through one-time 3DVAR, the observation data can be assimilated within the time window of the 3DVAR; through multiple-time 3DVAR, all the observation data within a time period can be assimilated. Anot her advantage of the cyclic 3D VAR is that the extended assimilation period is just enough for the model to be dynamically adjusted so as to effectively eliminate the spin-up phenomenon. During 3DVAR assimilation, the background field-error covariance must be given in advance. WRF-3DVAR itself provides a physical space background field-error covariance independent of the specific simulation area and grid points. It is based on the Global Forecast System (GFS) forecast field, which serves as the background field of the model.

Process topology' diagram

FIGURE 2.2: Process topology' diagram

c) Operation process.

GFS grid point data, GTS observation data (conventional radiosonde and ground observation data), satellite observation data and GOOS sea surface temperature (SST) are automatically acquired from the internet and decoded, and WRF-3DVAR data assimilation is carried out to form the initial and boundary values of WRFM5 forecast system. Then the WRF master program (completed by Linux workstation) will be run, and the final result will be graphically processed. The whole process is fully automated. Linux/UNIX cron tools and shell script files control the whole automation process. Process topology' is shown in Figure 2.2.

The forecasting system runs once a day, starting at 08:00 Beijing time and ending at 08:00 Beijing time the day' after tomorrow. The forecast period is 48 hours. The system starts automatically' at 00:00 every night and ends around 02:00 in the morning. The whole operation takes about two hours. See Figure 2.3 for the diagram of operation time control of the forecast system.

WRF initial field-generation scheme considering typhoon process

When extreme weather, such as a typhoon, occurs, WRF initial field generation requires other schemes. The NCAR-AFWA man-made cyclone scheme jointly developed by the US National Center for Atmospheric Research and

Diagram of operation time control of the forecast system

FIGURE 2.3: Diagram of operation time control of the forecast system

Typhoon Bogus in the northwest Pacific by using the NCAR- AFWA man-made cyclone scheme

FIGURE 2.4: Typhoon Bogus in the northwest Pacific by using the NCAR- AFWA man-made cyclone scheme

the Air Force Weather Agency is used to construct an initial Typhoon Bogus based on the typhoon’s position and intensity (Figure 2.4) [27, 28]:

In addition to the maximum wind speed, the typhoon report also contains information such as the radius of wind at 50 knots and 30 knots, while the NCAR-AFWA man-made cyclone scheme only uses the maximum wind speed information. The NCAR-AFWA man-made cyclone scheme will be improved to include information such as the radius of wind at 50 knots and 30 knots. Compared with the original NCAR-AFWA scheme, the improved scheme can accurately describe the structure of the wind field on the periphery of the typhoon [29].

A relatively simple function is adopted for wind profile in the original NCAR-AFWA scheme:

V(r) denotes tangential wind speed; r denotes the distance from the center of the typhoon; Vmax and Rmax denote maximum wind speed and the radius of maximum wind, respectively, in the typhoon observation report, a is -0.5 in general. According to the actual observation, the result of calculation near the center of the typhoon by using the formula is ideal, while the peripheral structure of the typhoon is not very good.

If the structure of the wind field on the periphery of the typhoon is taken into account, the following wind profile forms will be considered:

The parameter b in the above formula is to be determined. If the typhoon observation report contains information about the radius of wind at 50 knots and 30 knots, then the value of parameter b can be determined. Therefore, a new wind profile can be constructed based on the above two formulas:

where V (r) and V2(r) denote tangential wind speed in Eqns. (2.1) and (2.2), respectively: Rc denotes the distance from the center of the typhoon when Ri(r) - V2(r) = 0.

Model application examples and result verification

Examples of atmosphere module applications are as follows:

1) Automatic operation status

The forecast system runs regularly once a day, with the period of forecast validity being 48 hours. If the actual needs and calculation conditions allow, it can be run twice, and the period of forecast validity can be extended to 72 hours. Figure 2.5 shows the forecast results of the wind field on April 14, 2010.

2) Verification of wind field forecast results

Based on the wind speed data gathered from the coastal meteorological ground observation stations, offshore ships and islands in the Circum-Bohai Sea Region and North Yellow Sea (see Figure 2.6), we had the statistical analysis of the forecast wind field from March 10 to May 10 (two months).

Diagrams of forecast results of sea surface wind field on April 14, 2010 (the legend represents different wind speed levels in meters per second)

FIGURE 2.5: Diagrams of forecast results of sea surface wind field on April 14, 2010 (the legend represents different wind speed levels in meters per second)

Distribution of observatories used for statistics (observatories within gray areas in the figure)

FIGURE 2.6: Distribution of observatories used for statistics (observatories within gray areas in the figure)

Statistical results of wind speed and wind direction (RMSE

FIGURE 2.7: Statistical results of wind speed and wind direction (RMSE: root mean square error of wind speed; MEAN: average error of wind direction)

The statistical results are shown in Figure 2.7. The results show that the root mean square error of wind speed is between 2.2 m/s and 3.2 m/s, the average error of wind direction is between 24 degrees and 36 degrees, the root mean square error of wind speed within 24 hours is less than 2.5 m/s, and the average error of wind direction is less than 28 degrees. The forecast results are ideal.

< Prev   CONTENTS   Source   Next >