1. 1 / 22 Deutscher Wetterdienst Lindenberg Meteorological Observatory Richard Assmann Observatory What final GRUAN observations may consist of and look like Holger Vömel GRUAN Lead Center Lindenberg Meteorological Observatory German Meteorological Service

2. 2 / 22 Overview MotivationMotivation Measurement uncertaintiesMeasurement uncertainties What reference data may look likeWhat reference data may look like

3. 3 / 22 Measured quantities What GRUAN wants to measure: TemperatureTemperature Water VaporWater Vapor PressurePressure OzoneOzone WindWind ….…. Focus on temperature and water vapor Principle should apply to all reference measurements

4. 4 / 22 Temperature and water vapor Accuracy requirement: Water vapor:2% Temperature: 0.1K (troposphere) 0.2K (stratosphere) Long term stability: Water vapor:1% Temperature: 0.05K

5. 5 / 22 Reference radiosonde: Current status

6. 6 / 22 Quality quantification  Quality control fails, criteria are too stringent for the sensors we have  Try quality quantification Quantify how good sensors areQuantify how good sensors are Quantify how good sensors are as function of altitudeQuantify how good sensors are as function of altitude

7. 7 / 22 Sources of uncertainty Sensor calibration: Accuracy of calibration reference Accuracy of calibration model, …..Sensor calibration: Accuracy of calibration reference Accuracy of calibration model, ….. Sensor integration: Integration into radiosonde Telemetry limitations, …..Sensor integration: Integration into radiosonde Telemetry limitations, ….. Sensor characterization: Time lag variation of polymer sensor Controller stability of frostpoint hygrometer Production variability, …..Sensor characterization: Time lag variation of polymer sensor Controller stability of frostpoint hygrometer Production variability, ….. External influences: Radiation error Balloon contamination Sensor icing, …..External influences: Radiation error Balloon contamination Sensor icing, …..

8. 8 / 22 GRUAN as reference Distinguish between sources of uncertainty that: Can be corrected  correct and quantify remaining uncertaintyCan be corrected  correct and quantify remaining uncertainty Can not be corrected  flag as questionableCan not be corrected  flag as questionable

9. 9 / 22 GRUAN as reference GRUAN data (as Reference data) should: Capture and describe uncertaintyCapture and describe uncertainty Quantify uncertaintyQuantify uncertainty Test uncertainty where feasibleTest uncertainty where feasible

10. 10 / 22 Sources of uncertainty Example : Temperature Sensor calibration: Accuracy of calibration reference Accuracy of calibration modelSensor calibration: Accuracy of calibration reference Accuracy of calibration model Sensor integration: Integration into radiosonde Telemetry limitationsSensor integration: Integration into radiosonde Telemetry limitations Sensor characterization: Time lag variation of polymer sensor Controller stability of frostpoint hygrometer Production variabilitySensor characterization: Time lag variation of polymer sensor Controller stability of frostpoint hygrometer Production variability External influences: Radiation error Balloon contamination Sensor icingExternal influences: Radiation error Balloon contamination Sensor icing Vaisala RS92 daytime 0.01K (est.) 0.01K (est.) 0.01K (est.) 0.01K (est.) ? * - 0.2 K * (strat.) ? >1.0K (temporarily) (* after correction is applied)

11. 11 / 22 Temperature: Ground check

12. 12 / 22 Temperature: Radiation effect Actinic flux example: - Lindenberg (LUAMI) - low level stratus - high level cirrus - high level cirrus

13. 13 / 22 Temperature: Radiation uncertainty Remaining uncertainty after radiation correction (Vaisala RS92 only)

14. 14 / 22 Sources of uncertainty Example: Humidity Sensor calibration: Accuracy of calibration reference Accuracy of calibration modelSensor calibration: Accuracy of calibration reference Accuracy of calibration model Sensor integration: Integration into radiosonde Telemetry limitationsSensor integration: Integration into radiosonde Telemetry limitations Sensor characterization: Time lag variation of polymer sensor Controller stability of frostpoint hygrometer Production variabilitySensor characterization: Time lag variation of polymer sensor Controller stability of frostpoint hygrometer Production variability External influences: Radiation error Balloon contamination Sensor icingExternal influences: Radiation error Balloon contamination Sensor icing Vaisala RS92 daytime 0.1% RH (est.) 0.1% RH (est.) 0.1% RH (est.) 0.1% RH (est.) < 5% RH 1% (default setting) 10% ? 10% ? > 40 % (rel. error) ? (troposphere) ? > 40 % (rel. error) ? (troposphere) ? (No correction are applied)

15. 15 / 22 Example Humidity: Ground check 0% 0% check

16. 16 / 22 Example Humidity: Ground check 100% 100% check

17. 17 / 22 Example Humidity: Radiation error

18. 18 / 22 Example Humidity: Combined uncertainies Uncertainties considered: Ground check: +/- 0.5% absolute Integer resolution: +/- 0.5% absolute Radiation correction: < 10%relative Time lag correction: < 10%relative Production variability: < 4%relative

19. 19 / 22 Example Humidity: Corrected profile including uncertainties

20. 20 / 22 GRUAN data: Generalization Final data contain: Manufacturer output dataManufacturer output data All known correctionsAll known corrections Vertically resolved uncertaintiesVertically resolved uncertainties Ground check independent of manufacturer‘s ground check (where feasible)Ground check independent of manufacturer‘s ground check (where feasible) Reference to corrections and processingReference to corrections and processing Meta data for complete descriptionMeta data for complete description

21. 21 / 22 GRUAN data: Generalization For traceability need: Raw phyiscal data (e.g. temperature or water vapor, no filtering, no smoothing, no corrections)Raw phyiscal data (e.g. temperature or water vapor, no filtering, no smoothing, no corrections) Documentation of all corrections (preferred published papers)Documentation of all corrections (preferred published papers) Some engineering data (e.g. Humidity sensor temperature, raw frostpoint optics signal, ozone cell current and pump temperature,...)Some engineering data (e.g. Humidity sensor temperature, raw frostpoint optics signal, ozone cell current and pump temperature,...) Version number of dataVersion number of data

22. 22 / 22 GRUAN data: Generalization Application of vertical uncertainty profile: Metric for sensor comparisonMetric for sensor comparison Quantification of sensor ‘Agreement‘Quantification of sensor ‘Agreement‘ Quantification of vertical range of sensor outputQuantification of vertical range of sensor output Quantification of ‘Reference’ instrumentQuantification of ‘Reference’ instrument Identification of need for improvementIdentification of need for improvement

23. 23 / 22 Humidity: Radiation correction

24. 24 / 22 Humidity: Ground check 0%