Forecasting Technology of Tidebound Water Levels
Research content
Statistical methods and numerical models are utilized to study the highly accurate operational forecast technology of tidebound water level and realtime dynamic report technology of water level in ports and waterways (the forecast error of tidebound duration is controlled within 20 minutes and, the forecast error of tidebound water level is controlled within 20 cm). Meanwhile, the draft data of vessels is also used to study the tidebound waterlevel forecast system so that the operat ion requirements of the system can be satisfied.
Technology roadmap
The core of the technology is to study the fluctuation regular pattern of seawater. The fluctuation of seawater is a natural phenomenon; we call it tidal phenomenon. In fact, the fluctuation of tide level or water level is composed of two parts, namely the periodic water level and the nonperiodic water level. The former is the astronomical tide level produced by the combined action of astrotidegenerating force while the latter is positive and negative storm surge caused by meteorological, geological and hydrological factors [51, 52, 53]. Among them, the main geological factor that affects the significant changes of water level is the tsunami caused by submarine earthquakes. Due to the limitations of the current earthquake forecasting technology and the wide continental shelf along the coast of China, as well as the peripheral barrier formed by archipelago, tsunamis caused by submarine earthquakes exert relatively small impacts on the coastal waterlevel forecast of China [50, 54]. Therefore, in this waterlevel forecast study, the influence of factors other than geological factors on the water level will be considered and methods to accurately forecast the future water level will be given. See the following technology roadmap for the specific forecast methods:
The tidebound waterlevel forecast technique established by this project organically couples the meteorological and tidelevel monitoring data, the statistical forecasting method of tide level, the dynamics model of atmosphere and a realtime visualization system to achieve the accurate and rapid forecast of tidebound water levels. The technology roadmap of the system is shown in Figure 3.1.
As shown in Figure 3.1, the steps to construct the system are as follows. First, select the key coastal ports and waterways in China as the target sea areas for waterlevel forecasting, and collect hydrometeorological and geographic data for these sea areas. Then, use the harmonic analysis method on the basis of historical longterm measured waterlevel sequence to forecast the astronomical tide of these sea areas. Meanwhile, use the high resolution meteorologicalmarine coupling numerical model and consider the multielement forcing conditions including tide, runoff, meteorology, etc. to calculate the water level of the target sea areas of more than one year and forecast the future water level of these sea areas. Then use the filtering method to filter out the tidal part of the calculated waterlevel value by numerical mode and obtain the forecast results of positive and negative water level. After that,
FIGURE 3.1: Construction technology roadmap of tidebound waterlevel forecast system add the positive and negative waterlevel results calculated by the harmonic analysis method and numerical model, respectively, to preliminarily obtain the waterlevel forecast results of these sea areas. To ensure the accuracy, when performing forecasting each time, the forecast results are corrected according to the forecast error of the past 24 hours. Finally, complete the construction of the tidebound waterlevel forecast system by combining with draft data of different vessels and information display technology [55, 56].
The core part of this technical study is the forecast of tidebound water levels. After years of experimentation and innovation, the final equation for tidebound water level L constructed and adopted in this project is as follows [57, 58, 59]:
where: 1) In the application example of this project, the astronomical tide level C is forecast with harmonic constant of 116 tidal constituents (including longterm sa and ssa atmospheric tidal constituent) adjusted by the multiyear measured waterlevel sequence data at the port and the waterway.
 2) is the nonperiodic level of positive and negative water level forecast by the dynamics model, including the level of positive and negative storm surge caused by meteorological elements and runoff elements, and the waterlevel change caused by the nonlinear action between seawater tidal current due to these elements. In this project, the highresolution meteorologicalmarine coupling model is used first, including the waterlevel C,'_{m} changes caused by tides, runoff and meteorological elements. By calculating the waterlevel sequence of more than one year and using the method of harmonic analysis, the amplitude and tidal epoch at that point are obtained, and the astronomical tide level sequence Q{ (including sa and ssa atmospheric tidal constituent) is then forecast with these amplitude and tidal epoch. Then = Cm ^{—} Ct ^{can} be obtained. At this time, Q' contains the waterlevel changes of positive and negative storm surge caused by the nonlinear action between shortterm meteorological elements, runoff elements and multiple elements.
 3) e is the errorcorrection item. When the waterlevel forecast for the next 48 hours and the next 24 hours is carried out, e is the difference between the measured waterlevel value L_{r} — last2Ahrs and the forecast waterlevel value Lpia.st24h.rs hi the waterlevel forecast for the past 24 hours. When the report of realtime dynamic tidebound water level in the next one hour is carried out, e is the difference between the measured waterlevel value L_{r}i_{as}thr and the forecast waterlevel value L_{p}i_{as}thr in the waterlevel forecast for the past one hour.
Please refer to the formulated procedures for the specific calculation method and technical details in this technology roadmap.
Formulated procedures
The procedures to build the tidebound waterlevel forecast system of a target sea area are as follows:
Step 1: Organize historical meteorological observation data and tidelevel observation data as well as local topographic data and runoff data, and establish an historical meteorological database, historical hydrological database and a local topographic database, respectively.
Step 2: Use the historical tidelevel data in the historical tidelevel database established in Step 1 to obtain the tidal harmonic constant, and use the harmonic constant to analyze and forecast the astronomical tide level of the location.
Step 3: Obtain the meteorological and tidelevel monitoring data in a realtime manner for data assimilation and error correction of numerical model calculation.
Step 4: Based on the collected local hydrometeorological, runoff and topographic database, use the highresolution meteorologicalmarine coupled numerical models, and consider the multielement forcing conditions including tides, runoff, meteorology, etc. to calculate the waterlevel results of the target sea area of more than one year.
Step 5: Use the data assimilation method to assimilate the realtime meteorological monitoring data in Step 3 into the atmospheric forecasting model (such as WRF), and calculate and obtain the forecast wind field according to such an atmospheric forecasting model; use the calculation result of the tidal model of the large area as the open boundary condition of the tide.
Step 6: Use the wind field and open boundary condition of the tide forecast by the atmospheric forecasting model in Step 5 as the forcing field, and use the same meteorological and marine dynamics model in the Step 4 to calculate the forecast tidebound water level containing such astronomical tide level and positive and negative storm surge due to other factors.
Step 7: Base on the water level sequence result of more than one year calculated in Step 4, and filtering method is utilized to filter the tidebound water level part Q from the water level value ('_{n} which is calculated by numerical model in Step 6, then the forecast result of abnormal seallevel event f is obtained. After that, add the astronomical tide level £ calculated in Step 2 to the positive and negative storm surge results C calculated by the numerical model to preliminarily get the waterlevel forecast results of these sea areas.
Step 8: Use the tidebound waterlevel forecast result in Step 7 and add e and the average forecast error of the past 24 hours to the forecast result to correct the forecast result. Finally, combine with the draft data of different types of vessels and use the visualization technology to display the forecast results on the comprehensive information display system in a realtime manner, thus completing the construction of a tidebound waterlevel forecast system.
The calculation method of the tidal harmonic constant described in Step 2 is to use the leastsquares method to analyze the harmonic constant of the location according to the time length of historical observation data in the historical tide level database established in Step 1. Of which, the amplitude and tidal epoch of the harmonic constant are represented by H and g, respective!}'.
The astronomical tide level described in Step 2 refers to the tide level directly or indirectly caused by tidegenerating force. Based on the harmonic constant of the location, the astronomical tide level is calculated and forecast according to the following expression [60, 61]:
where, oo is the height of the average sea level on the tidal elevation datum and the base surface suitable for navigation. The height of average sea level at the zero point of the tidal station can also be taken as needed, j is the number of tidal constituents. fj,Hj,gj,
The data assimilation method and the method for forecasting the wind field result and the open boundary condition of the tide described in Step 5 are:
Carry out format conversion and quality control of meteorological observation data first, and then carry out data assimilation by using 3D VAR method to provide the initial field and timevarying boundary conditions for the marine dynamics model used in the forecast. Adopt and configure the WRFARW atmospheric dynamics model to provide wind field and related meteorological condit ions for the forecast of the concerned sea area in the next 48 hours. The model uses fully compressible, nonstatic Euler equations; the horizontal grid is the Arakawa C grid; the vertical coordinates are the mass based, terrainfollowing // coordinates that are widely used internationally.
The method of using a marine dynamics model to calculate the forecast tidebound water level f'_{n} including the astronomical tide level and negative storm surge level of the location is:
Based on the wind field results calculated in Step 5, use the harmonic constants of the eight tidal constituents М2, S2, N2, К2 ■ K,0, P and Q provided by NA099 Global Tide Model (http : //www.miz.nao.ac.jp/staf fs/nao 99/indexEn.html) as the forcing conditions of the FVCOM marine dynamics model and the openboundary conditions of the tides. The said marine dynamics model is FVCOM (a unstructured grid, FiniteVolume Coastal Ocean Model) marine dynamics model [62, 63, 64, 65, 66]. Such a model is a marine model of free sea surface, hydrostatics, Boussinesq approximation and primitive equations.
After the WRFARW model operation described in Step 5 is completed, SHELL script will automatically run the FVCOM model and calculate the forecast tidebound water level f'_{m} of the location in the next 48 hours.
The tidebound water level finally constructed and adopted in Step 7 and Step 8 is L = £ + Cm + e.
The methods for realtime display of the forecast result through the visualization technology described in Step 8 are:
Use the mapping technology to display the forecast results of tidebound water level L in the next 48 hours in a realtime manner and display the historical observation data and the forecast data in a curve. Use the web technology to release the information online and provide the information to the port dispatcher and the person taking charge of the marine vessels immediately by relevant communication means, thus providing powerful technical support for the safety and highly efficient operation of the port and navigation and the decisionmaking at the time of abnormal tidelevel changes.
The calculation and visualization process of the forecasting technology is automatically scheduled and executed by computer scripts.
Result verification
Tidebound water level is predicted in water areas of 204 Chinese key coastal ports and waterways in future 48 hours at Qinhuangdao, Tianjin, Yantai, Rizhao and Lianyungang in the next hour, with an average error of no more than 5%. According to verification made by the system via comparing the forecast water level at Tianjin Port, Yantai Port, Rizhao Port and Lianyungang Port in 2012 with the measured value, the average error of the system in tidebound waterlevel forecast for the above ports for the 24hour prediction is less than 10 cm (5% lower than the tidal range of those ports and up to the requirement that the tidebound water level forecast error should be less than 20 cm); the error in the forecast of tidebound duration is less than 15 minutes (up to the requirement that tidebound duration forecast error should be less than 20 minutes); the average errors in the timely and dynamic forecasts of the system on the tidebound water levels of all such ports in the next hour are less than 1 cm and on the tidebound duration are less 2 minutes. See Table 3.1 for details of the forecast errors at those measuring points, and DW_sm, DW.sm24, DT_tm, respectively, indicate the difference between the waterlevel forecast made by the system and the measured water level for the next one hour, future 24 hours, and set out in tide table. A comparison between the forecast tidebound water level in the tide table and the
TABLE 3.1: Comparison between tidebound waterlevel forecast error and the measurements set out in the tide table
No. 
Point 
DW_sm24 
DT_tm 
DW_sml 
Position 

Latitude 
Longitude 

1 
Tianjin 
9 
32 
0.85 
38° 59.0' 
117°47.0' 
2 
Yantai 
9 
22 
0.95 
37^{u}33.3' 
12Г23.5' 
3 
Rizhao 
9 
21 
0.9 
35°22.0' 
119°34.0' 
4 
Lianyungang 
12 
23 
1.1 
34°44.8' 
119°28.7' 
Average 
9.75 
24.5 
0.95 
measured value shows that the system has reduced the average error in water level forecast by 14.75 cm.
Software of operational tidebound waterlevel forecast system
The operational tidebound water level forecast system presets harmonic constants and other configuration information via configuration file and inputs data in the historical observation data file and the realtime observation data sheet of the monitoring information base, in order to export forecast values of tidebound water levels of given ports in the future 48 hours for the users. Input information of the tidebound waterlevel forecast and the running interface are shown in Figure 3.2. At present, the system is incorporated into the “marine environment information support and forecast system for ports and shipping” as a separate module and deals with the release of an information application service.
Socialeconomic benefits and application prospects
The system has improved the accuracy of tidebound waterlevel forecasts for key coastal ports and waterways in China by more than 10 cm, enabling the operational application of tidebound waterlevel forecasting for the first time. The most direct economic benefits brought about by the system is, due to more accurate tidebound waterlevel forecast results, the reduction of freight and oil expense as it allows ships to carry more cargo when entering the port
FIGURE 3.2: Information input and running interface of the operational tidebound waterlevel forecast system
by taking the tide, which contributes to energy saving, emission reduction and environmental protection in an indirect way. Let’s take the following case as an example. There is a ship with a T.P.C. (tons per centimeter) of more than 70 tons carrying 70,000 tons of iron ore and sailing from Brazil to Tianjin Port. If it enters the port by taking the tide, it will have an extra draft of more than 100 cm, that’s to say, it can load additional 70 x 100 = 7000 tons cargo at most at a time. Based on the freight of 25 dollars/ton, the revenue directly generated therefrom is 175,000 dollars; suppose that the fuel consumed in each voyage is 2,000 tons, 0.1 trip is saved by each voyage, which is equivalent to the reduction of 200 ton fuels; based on the fuel price of 660 dollars/ton, a total of
132,000 dollars are saved; in other words, each voyage can directly generate a benefit of above 307,000 dollars. In addition, each voyage reduces the carbon dioxide emission by over 640 tons if we assume that 3.2 tons of carbon dioxide is produced with the consumption of every ton of fuel. If the popularization of such technology is made available in a wider range of Chinese coastal ports and key waterways, not only will remarkable economic benefits be generated, but also constructive results in environmental protection will be produced.
In addition, the system can effectively save ships the trouble of stranding caused by significant water level falling arising from meteorological factors. In 1983, the Feoso Ambassador tanker stranded in Kiaochow Bay, resulting in a spill of oil amounting to 3,343 tons, which contaminated the 230 km coastline and scenic spots in the coastal waters of Qingdao. In 1984, the Jiacui tanker stranded in Kiaochow Bay and spilled 757 tons of oil, casting a cloud on the marine environment in Qingdao which is still in restoration. With the change of climate, the frequency of the occurrence of extreme positive and negative waterlevel events on the sea is on the rise. Such events, however, can be reduced by accurately forecasting the water level of ports and waterways.