WANG Qisong,FANG Shulin,YIN Danni,ZHAO Yueru

Dalian Sea Environmental Monitoring Central Station,SOA,Dalian 116015,Liaoning province,China

Abstract:The quality control of ocean observation data is conducted by monthly review,yet no on-line quality control system is deployed.An on-line quality control system is provided by this article.The threshold of extreme and rate of change is created by analyzing the character of observation items.The alert will be triggered to remind the operator if the system detects data out of threshold，therefore the on-line data quality control is accomplished.

Keywords:ocean station;data;on-line;quality control

China is a large marine country,with 3 million km2of sea area under its jurisdiction,18 000 km of continental coastline and 14 000 km of island coastline.Each year various marine disasters not only cause huge losses to coastal economy,but also pose a serious threat to people's life and safety[1].The purpose of ocean observation is to obtain the basic marine data and provide basic support for marine economic construction,marine rights and interests maintenance,marine disaster prevention and adaptation to global climate change[2].Ocean station is a fixed marine environment observation facility built on seashore or islands and reefs.The hydro meteorological data observed at ocean station can provide an important basis for marine disaster prediction and decision-making[3].Automatic observation is the main way to obtain ocean observation data.In order to ensure the data quality,for the stations that conduct operational observation through automatic observation instruments,the comparative observation shall not be less than twice a month[4].Thus the delayed monthly data report and the three-level quality control,namely preliminary review,proofreading and preliminary review are conducted[5].Consistency test,rationality test,correlation test and distribution fitting test are the main quality control methods adopted by the monthly report quality control software[6].Real time quality control generally focuses on measures such as manually checking whether the observation data is within the measurement range of its sensor,whether it conforms to the data law curve and whether the abnormal extreme value caused by instrument d amage is taken by the operator on duty[7].Relevant exploration and research have been carried out on the quality control and evaluation standards of marine real-time observation data,including transmission effectiveness,serious error range,threshold test,geophysical threshold interval,change rate continuity,etc.[8]Manual data reading,manual data analysis and the use of quality control software can effectively detect the abnormalities of automatic observations at ocean stations,but there are also deficiencies[9].The local realtime quality control and check of observation data has not been put into practice.This paper first describes the functional design and system development of an online quality control system for automatic observation data of ocean stations,then explains the trial situation of a measuring point,and finally expounds the existing problems of the system.

1 Functional design

1.1 Functional requirements

In order to realize the online quality control of marine observation data,the functional design of the system is carried out with reference to the relevant requirements of theQuality Control Method of Marine Hydro Meteorological Data(technical document TA2014-001 of the National Marine Data and Information Service).Firstly,the system shall be able to read the current marine observation data minute record in real time,obtain the real-time data of each observation element,and then compare it with the set continuity,extreme value and other thresholds,give abnormal prompt or error prompt for the data exceeding the thresholds,and store the prompt information.

1.2 Select online QC observation elements

At present,the automatic observation carried out by the ocean station include tide,sea surface temperature(SST)and sea surface salinity(SSS),wave,wind speed and direction,air temperature,relative humidity,air pressure,precipitation,visibility,etc.The automatic observation level of wave and visibility is low,and the precipitation is discontinuous,which is not included in the online quality control scope for the time being.Online quality control includes seven observation elements:tide,SST and SSS,wind speed and direction,air temperature,relative humidity and air pressure.

1.3 Threshold setting analysis

The threshold setting includes four parts:maximum observed value,minimum observed value,maximum change rate of observed value and minimum change rate of observed value.The threshold setting is based on the ocean station to be deployed by the system,and the observed data over the years are statistically analyzed to obtain the threshold required by the system.

Maximum observation value setting refers to the statistics of the observation values of observation elements over the years,and the maximum value is taken as the maximum observation threshold,which is entered into the system as the basis for system judgment.If the in-situ data is greater than this value,an abnormal warning will be sent.Wind speed includes maximum wind speed and extreme wind speed,both of which are set.At the same time,according to the existing actual observation cases,the judgment condition that the extreme wind speed is greater than the maximum wind speed is added.If the maximum wind speed is greater than or equal to the extreme wind speed,an error warning will be sent.

Minimum observation value setting refers to the statistics of the observation values of observation elements over the years,and the minimum value is taken as the minimum observation threshold and entered into the system as the basis for system judgment.lf the actually obtained observation data is less than this value,an abnormal warning will be sent.

The maximum change rate of observation value refers to analyzing the observation data over the years,setting the judgment time interval and calculating the maximum change rate according to the natural extreme value change characteristics of observation elements.

The maximum change rate of tide is set on the premise of normal semidiurnal tide.The maximum tidal range value is obtained through statistical analysis of the observed data over the years,and then the maximum change rate of tide is obtained by dividing the value by the start and end time interval.The value is entered into the system as the basis for discrimination.If the system detects that the change rate is exceeded,an abnormal warning will be sent.

The SST and SSS are counted according to the daily extreme value(0:00-24:00).Through the statistical analysis of the observed data over the years,the maximum daily change value is obtained,and then the maximum change rate is obtained by dividing the value by the start and end time interval.The value is entered into the system as the discrimination basis.lf the system detects that the change rate is exceeded,an abnormal warning will be sent.

The air temperature,relative humidity and air pressure are counted according to the daily extreme value(8:00 p.m.-8:00 p.m.),the maximum value of daily change is obtained through statistical analysis of the observed data over the years,and then the maximum change rate is obtained by dividing the value by the start and end time value,which is entered into the system as the basis for discrimination.lf the system detects that the change rate is exceeded,an abnormal warning will be sent.

Due to its own characteristics,the change rate of wind speed and direction is large.Here,according to the actual fault cases,the following abnormal logic settings are made:when the wind speed is zero,no abnormal analysis is made on the change rate of wind direction;when the wind speed is not zero,if the average wind direction remains unchanged for 10 minutes,an error prompt will be given.

The automatic wind observation is collected every 3 seconds for 10 minutes,and the average value of wind range and corresponding wind direction is calculated as the average wind speed and corresponding wind direction at the end of the 10 minutes[9].According to the natural attribute of wind direction,when the wind speed is at the time of continuous non static wind,the instantaneous wind direction must change irregularly.lf the average wind speed remains unchanged within 10 minutes,it means that the instantaneous wind direction has not changed within 10 minutes,which is inconsistent with its natural attribute,and must be wrong data.Therefore,the above logic setting is adopted.

2 System development and application

2.1 System development

To effectively integrate with the current marine observation data acquisition software,the software development is based on the existing data acquisition software,further expand the relevant functional modules of the system,simplify the operation to the greatest extent and improve the work efficiency.

The main process of system design includes:

a)Count and calculate the quality control threshold of each observation element according to the existing data of the online quality control station to be carried out;

b)Input the threshold of each observation element into the software;

c)The system reads the real-time observation database and analyzes the minute message according to the observation elements;

d)Store the parsed observation elements into the monitoring database;

e)The system polls the database data regularly,compare the observed element data with the extreme value in the threshold,give an abnormal prompt when the data other than wind is abnormal,and gives an error prompt when the wind data is abnormal;

f)Compare the cached observation element data with the change rate in the threshold,d give an exception prompt when the non-wind data is abnormal,and give an error prompt when the wind data is abnormal.As shown below.

2.2 Application

Fig.1 Main flow chart of the system

The system is deployed in Xingcheng measuring point.While carrying out data collection,it carries out quality control on the real-time data of the measuring point.The deployment time is July 1,2020,and the operation is normal by August 2021.Real time quality control elements are shown in Tab.1.

Tab.1 Quality control elements of measuring points

The system software input interface is shown in Fig.2.

Fig.2 Input interfaces

In order to evaluate and verify the accuracy,effectiveness and stability of the functional design of the system,the data message from October 1,2020 to May 31,2021 is selected for analysis.See Tab.2 functional design evaluation table for details.

Tab.2 Functional design evaluation form

It can be seen from Tab.2 that the system can process all the received effective data,detect extreme value abnormal data and change rate abnormal data,and there is no crash during operation.lt is preliminarily proved that the system operates stably,detects abnormal data accurately,derives reliable data,and meets the functional design requirements.The abnormality detected by the system is shown in Fig.3.

Fig.3 Abnormality result

3 Existing problems

3.1 The threshold used during the system evaluation period(August 1,2020-March 31,2021)is lower than the theoretical value

At the initial stage of system operation,after inputting the theoretical threshold,abnormal data cannot be detected.ln order to verify the system function,the method of lowering the threshold is used for verification.

3.2 Not all ocean observation features have been covered

Observation elements such as wave,visibility and precipitation are not included in the quality control scope of the system.The automation of wave and visibility observation is low.lt is mainly manual observation,1-3 times a day,and empirical observation is mostly used.The necessity of online quality control is not very prominent.Precipitation is discrete,not continuous and unpredictable,therefore the necessity of on-line quality control is also less prominent.

3.3 There is no in-depth analysis on the quality control correlation of different marine observation elements

There is also a certain correlation between marine observation elements,which is not deeply analyzed in this paper:for example,the relationship between SST and air temperature is obviously different due to different seasons and time.ln winter,the temperature of sea surface before freezing is generally higher than the air temperature,and in spring and autumn,the sea water temperature is generally lower than the air temperature,while in summer,the relationship between them is constantly changing due to the difference between morning and evening.

4 Conclusion

The development of real-time online quality control system for ocean observation data enriches the quality control methods of ocean observation data.The system can identify data abnormalities more timely and effectively,and will play a significant role in improving the quality of ocean observation data.At the same time,it will provide reference for operational online quality control of ocean observation data.

Acknowledgement

This article is sponsored by the project of North Bureau of SOA 2017(Development and application of online quality control system for automatic observation data of Ocean Station No.2017B12).Thanks for the helping from Ll Guanyu and QlU Wenbo from National Ocean Technology Center,who delivered great supportments during the period of project employment.