- Количество слайдов: 16
Range Checks of Coastal Environmental Monitoring Data Lei Hu Dauphin Island Sea Lab lhu@dis. org Brenda Leroux Babin Louisiana Universities Marine Consortium bbabin@lsu. edu
Dauphin Island Sea Lab http: //www. mymobilebay. com/
Louisiana Universities Marine Consortium http: //weather. lumcon. edu/
Introduction • Purpose of real-time monitoring system Provide researchers, educators, and the public with quality data for planning and decision making. • Quality Assurance Actions taken while data is being collected to ensure that the data are "good. " • Quality Control Steps taken after the data is collected to ensure the data is "good" before the data is released to the public.
Quality Assurance of Real-time Oceanography Data (QARTOD) • Quality Control First Step: insure completeness of the data set 1. transmission check—valid message was received 2. time stamp validity check 3. a check for service schedule of hardware Second Step: various range checks on data 1. initial gross range checks based on instrument specifications and range checks based on climatology of the area. 2. checks against trends in data and checks against nearby stations. QARTOD suggests that this Quality Control should be automated and yet contain a manual check by a human component.
Planning An automated range-check system Range Checking Criteria based on ranges established by NERR (by NOAA’s National Estuarine Research Reserve (NERR). Meteorological Data Air Temperature Wind Direction Wind Speed Barometric Pressure Solar Radiation Quantum Radiation Precipitation Relative Humidity Hydrographic Data Water Temperature Water Level Salinity Dissolved Oxygen Percent: Dissolved Oxygen mg/L -10 °C < And < 50 °C 0 degree < And <360 degrees 0. 5 m/s < And <30 m/s 900 mb < And < 1060 mb 0 KW/m 2 < And < 1. 5 KW/m 2 0 µE/m 2/s < And < 2700 µE/m 2/s 0 mm/min < And <3 mm/min 0% < And < 100% -5 °C < And < 45 °C At Dauphin Island Station: -3. 0 ft. < And < 5. 0 ft. At Meaher Park Station: 1. 0 ft. < And < 6. 0 ft. At Middle Bay Light Station: 2. 0 meters< And <6. 0 meters 0 ppt < And < 40 ppt 0% < And < 200% 0 mg/L < And < 20 mg/L
Programming • Created the flag fields in the same data table. • Range check logic was translated into SQL stored procedure. Runs every 30 minutes, 15 minutes after data are harvested. 1 minute meteorological data, 30 minute hydrographical data • Quality Flag The quality of real-time data is described by an aggregate quality flag recommended by QARTOD I: -9 = missing value 0 = quality not evaluated 1 = bad 2 = questionable/suspect 3 = good
SQL stored procedure that checks the range of meteorological parameters (scheduled to run every 30 minutes by SQL Server Agent) update a set precip 1 Flag = case when precip 1*2. 54 >3 then 2 when precip 1 is null then -9 when precip 1*2. 54 <=3 then 3 end, airtemp 1 Flag= case when airtemp 1 > 50 or airtemp 1 <-10 then 2 when airtemp 1 is null then -9 when airtemp 1 >=-10 and airtemp 1<= 50 then 3 end, solarrad 1 Flag= case when solarrad 1 >1. 5 or solarrad 1 < 0 then 2 when solarrad 1 is null then -9 when solarrad 1 >=0 and solarrad 1 <=1. 5 then 3 end, quantumrad 1 Flag= case when quantumrad 1 >2700 or quantumrad 1 < 0 then 2 when quantumrad 1 is null then -9 when quantumrad 1 >=0 and quantumrad 1 <=2700 then 3 end, winddir 1 Flag = case when winddir 1 > 360 or winddir 1 < 0 then 2 when winddir 1 is null then -9 when winddir 1 >=0 and winddir 1 <=360 then 3 end, /*convert windspeed from knots to meter per second*/ windspeed 1 Flag= case when windspeed 1/1. 9438445 >30 then 2 when windspeed 1 is null then -9 when windspeed 1/1. 9438445 <= 30 then 3 end, /*convert inches of Mercury to millibar*/ bar_pressure 1 Flag= case when bar_pressure 1/0. 02953007 >1060 or bar_pressure 1/0. 02953007 < 900 then 2 when bar_pressure 1 is null then -9 when bar_pressure 1/0. 02953007 >=900 and bar_pressure 1/0. 02953007 <=1060 then 3 end from ['+@station+'_'+@Year+'_met_min] as a WHERE airtemp 1 Flag is null')
Implementation Search for data with flag value of 2 –suspicious data. Once data are flagged, another program checks all the flags, four times a day. It scans records that come in within the last six hours on the SQL server. If it finds any data with flag value of 2, it will send an email to the Data Manager and technical support personnel. Subject: Middle Baylight Station Data Out of Range From: lhu@disl. org“ To: lhu@disl. org, mdardeau@disl. org, kweis@disl. org Middle Baylight Station Data Out of Range. Please click http: //www. mobilebaynep. com/mondata/rangestatus. cfm? stationid=188 to see detail.
The final decision on the quality of the data 1. Physical Processes. Example: strong winds may result in unusually low tides 2. Possible malfunction of instruments. Example: accumulation of fouling organisms on sensor will affect the reading of dissolved oxygen. 3. Biological Processes. Algea bloom may result in high dissolved oxygen reading. Update the flag, or add comments to the record Password protected website: http: //www. mobilebaynep. com/mondata/login. cfm.
The Review process is captured. Metadata has been registered with National Coastal Data Development Center The data flags are available together with the data when web visitors download Data from the DISL website (http: //www. mobilebaynep. com/mondata/dislcaimer. cfm)
Acknowledgement We want to thank Mike Dardeau, Holly Hebert, and technicians at the respective sites whose assistance made this report possible.