WVSS-II Moisture Observations – A Tool for Validating and Monitoring Satellite Moisture Data

1. WVSS-II Moisture Observations –A Tool for Validating and Monitoring Satellite Moisture Data Ralph Petersen1, Lee Cronce1, Erik Olson1, Wayne Feltz1, David Helms2 and Randy Baker3 1 Cooperative Institute for Meteorological Satellite Studies (CIMSS), University of Wisconsin – Madison 2 NOAA, National Weather Service, Office of Science and Technology, Silver Spring, Maryland 3 United Parcel Service, Louisville, Kentucky

2. Measuring Moisture from Commercial Aircraft • Efforts underway for more than a decade • Research instruments not appropriate for “day-to-day”, “real world” application • Initial experiments were made using a “stand-alone” Temperature/Relative Humidity sensor called the Water Vapor Sensing System (WVSS-I) • Used humidity sensors “similar” to those used on radiosondes • Test results showed: • Substantial Biases and RMS values that exceeded WMO specification • Systems became contaminated by everyday airport “gunk”, e.g. de-icer, dirt on runways, etc.

3. Measuring Moisture from Commercial Aircraft • Second-generation Water Vapor Sensing System (WVSS-II) measures MixingRatiodirectly • Uses a laser-diode system to measure number of water molecules passing sensor • Testing on UPS 757s • Justified by UPS from savings in improved fog forecasts • Re-engineered electronics • Improved mechanics • Final tests in 2009-2010 • New installation underway at SouthWest Airlines

4. EXPERIMENTAL DESIGN • Independent ground-truth • assessments of the WVSS-II • systems were conducted for • multiple years and multiple seasons • The WVSS-II humidity data were compared with rawinsonde and ground based remote sensing systems. • - Between 25 and 30 UPS B757 aircraft provided WVSS-II data via MDCRS. • Rawinsondes observations were made at the UPS hubs • WVSS-II equipped planes land / take off daily • Verifying Roab observations made at night • Avoids need for radiation correction to Raob data

5. Comparing 2005 & 2006 Validation Results 2005 2006 Engineering changes made after the 2005 test were at least partially successful in improving WVSS-II data taken, but only during ascent. • Modified systems produce consistent small negative Biases at all levels. • Random error component improved - ~0.4 g/kg, a 50-65% improvement BUT: Still unacceptably numbers of ‘bad’ systems and high degradation rate Only ascending reports > 2 g/kg and <10 g/kg – due to known system deficiencies.

6. Remaining WVSS-II problems addressed in 2007-9 • Three re-engineered units to NOAA were thoroughly tested before • widespread aircraft installation in 2009-2010: • - Data processing hardware replaced with digital systems unaffected • by ambient temperature • - Issues regarding water accumulating in intake tubes corrected. • - All moisture was removed from laser chambers. • - Every laser was tested for long-term stability before use. • - Assessed: • - In Chamber at the NOAA’s Upper-Air Facility    - In Chamber at Deutscher Wetterdienst    - Versus chilled mirror on Research aircraft • - Conducting long-term laser stability tests • - Reporting Precision issues resolved on all UPS aircraft • In 2009-2010: • Replaced 25 WVSS-II units on UPS B-757s • Installing 31 units on Southwest B-737s

7. Chamber Experiments by NOAA and DWD were Very Positive Only substantial differences appear for Specific Humidity below ~0.03 g/kg

8. Initial Comparisons of re-engineered WVSS-II data with co-located surface (METAR) reports First new WVSS-II unit on UPS aircraft agrees very closely with time/space co-located night-time surface observations from September 2009: Mixing Ratio Bias ~ 0.2 g/kg Mixing Ratio Standard Deviation ~ 0.4 g/kg Note: Random Temp error > Dewpoint error Data provided by Randy Baker, UPS

9. EXPERIMENTAL DESIGN of 2009-2010 Tests • Most recent Independent • ground-truth assessments • of the WVSS-II systems • have been conducted for • three periods: • - November 2009, • - May-June 2010, and • - August 2010. • The WVSS-II humidity data were compared with rawinsonde and ground based remote sensing systems. • Between 15 and 20 different UPS B757 aircraft provided WVSS-II data • - Data available via GTS • Rawinsondes observations were made at the UPS hub in Rockford, Illinois – where about 20-25% of the WVSS-II equipped planes land / take off daily.

10. Nov 2009-2010 Validation Results Direct Sounding Intercomparisons Direct Data Comparison: Aircraft data generally fell between bounding Rawinsonde reports

11. Nov 2009-2010 Validation Results Direct Sounding Intercomparisons Direct Data Comparison: Aircraft data generally fell between bounding Rawinsonde reports Large variability within Moist regimes led to large Specific Humidity differences

12. Nov 2009-2010 Validation Results Summary of Direct Specific Humidity Intercomparisons All SH data All Levels Spring SH data only All Levels Differences showed: Aircraft data and rawinsonde reports agreed best in middle SH ranges Positive WVSS-II biases at low rawinsonde values (low bias improbable) Few moist outliers from one case in 10-12 g/kg range – good for moister data

13. Nov 2009-2010 Validation Results Direct Specific Humidity Intercomparisons by Relative Humidity All observations – All Levels Differences showed: Small positive Bias across all RH ranges Random Errors average ~0.5-0.7 g/kg (low bias improbable) Higher Random Errors between 20-25% RH and Near Saturation

14. 2009-2010 Validation Results Specific Humidity (Excludes cases with large time and vertical rawinsonde differences) Systematic Differences: WVSS-II Biases at low levels of 0.1 to +0.4 g/kg from surface to 850 hPa. ±0.2 g/kg above Random Differences (Including Dry/Moist Environments): Differences between aircraft data and rawinsonde reports generally showed variability of 0.3 to 0.7 g/kg from the surface to 600 hPa – decreases aloft. StdDev slightly larger than 1-hour variability between bounding rawinsonde reports (gray shading). Note: Fewer intercomparisons near 800 hPa and above 700 hPa. Greater time and space separation above 650 hPa.

15. 2009-2010 Validation Results Specific Humidity (Excludes cases with large time and vertical rawinsonde differences) Systematic Differences: WVSS-II Biases at low levels of 0.1 to +0.4 g/kg from surface to 850 hPa. ±0.2 g/kg above Ascent and Descent Results Similar Random Differences (Including Dry/Moist Environments): Differences between aircraft data and rawinsonde reports generally showed variability of 0.3 to 0.7 g/kg from the surface to 600 hPa – decreases aloft. StdDev slightly larger than 1-hour variability between bounding rawinsonde reports (gray shading). Note: Fewer intercomparisons near 800 hPa and above 700 hPa. Greater time and space separation above 650 hPa.

16. 2009-2010 Validation Results Relative Humidity (From WVSS-II Humidity & Rawinsonde Temperature) Systematic Differences: RH Biases due to WVSS-II were small positive (1 – ±4%) from surface to 650 hPa. Negative maximum at observation minimum. Random Differences (Including Dry/Moist Environments): Differences between aircraft data and rawinsonde reports generally showed variability of 6 to 9% from the surface to 750 hPa. Above 750 hPa, RH StdDev increases as number of matches decreases and space/time distance increases. Random Differences slightly larger than 1-hour variability between bounding rawinsonde reports (gray shading).

17. 2009-2010 Validation Results Approximating WVSS-II Error and Natural Variability Restricted RMS calculated amongst WVSS-II observations for: Time ranges of 0-15, 15-30, 30-45 and 45-60 minutes Distance ranges of 0-15, 15-30, 30-45 and 45-60 km • Restricted RMSs show (ALL reports,Including Dry/Moist Environments): • Atmospheric Variability more than doubles from 0-15 to 30-45 minute intervals • Spatial Variability increase is consistent, but not as regular as temporal • Total Variability made up of 1) Instrument Error and 2) Atmospheric Variability • Projecting for exact co-locations (ΔT& ΔTZ~0 yields Total Variability < 0.2 g/kg), • Expect Operational WVSS-II Instrument Errors should be ~0.1 g/kg

18. Summary • Engineering/mechanical issues with WVSS-II sensors have been resolved !!! • Tests made over wide range of moisture conditions show: •  Sensors agreed extremely closely with each other • - Overall Specific Humidity (SH) RMS < 0.2 g/kg •  Sensors agreed well with co-located Rawinsonde observations • - Overall SH Bias ~ 0.2 g/kg, SH StDev ~ 0.5 g/kg •  Relative Humidity differences were small • - Overall RH Bias ~ 2.5 %, RH StDev ~ 7.5% •  Ascent and Descent Data show similar results • - Some super-saturations remains, especially at RH>95% • - Evaporation of Droplets/Snow in heated intake tube? • WVSS-II data Meet WMO requirements for mesoscale observations • Additional analysis underway to: • Provide atmospheric variability and observation error to support assimilation • Expand data base in moist environments - using NWS Rawinsonde data • Evaluate enroute, low-moisture environments using satellite data

19. The FutureWVSS-II Installations increasing on SouthWest Airlines B-373 The data are good We need to use them !

20. A Nationwide “Network of Networks”: Investigating the Impacts of a Potential Ground-based Profiling System using an Observing System Simulation Experiment Dave Turner1, Dan Hartung2, Jason Otkin2, Wayne Feltz2, Ralph Petersen2, Erik Janzon2 NOAA / National Severe Storms Laboratory (NSSL)- Norman, OK1 Cooperative Institute for Meteorological Satellite Studies (CIMSS / SSEC)2

21. Observing Weather and Climate from the Ground Up: A Nationwide ‘Network of Networks’ (NRC 2009) In a 2009 report issued by the National Research Council, it was determined that the current state of surface observations and lower-tropospheric profiling capabilities in the United States: “… are too limited in what they measure, too sparsely or unevenly distributed, frequently limited to regional areal coverage, and clearly do not qualify as a mesoscale network of national dimensions.” while “limiting our ability to forecast mesoscale atmospheric processes and high-impact weather.” (NRC 2009) • NRC Recommendation Task #8a: • “8. As a high infrastructure priority, federal agencies and their partners should deploy lidars and radio frequency profilers nationwide at approximately 400 sites to continually monitor lower tropospheric conditions. • Humidity, wind, and diurnal boundary layer structure profiles are the highest priority for a network, the sites for which should have a characteristic spacing of ~125 km but could vary between 50 and 200 km based on regional considerations.” (NRC 2009)

22. Qualifying Instrument Constraints Qualifying ground-based profiling conditions: ... must have the ability to minimally measure vertical profiles from surface to at least 3 km AGL, vertical resolution of the order of 100 meters, temporal resolution of at least 1 sounding every 2 hours, and parameter accuracy meeting WMO minimum accuracy requirements with at least 90% data availability.

23. Tested Ground-based Profiling Systems

24. Doppler Wind Lidar (DWL) • Automated, hardened instrument • Laser operates at 1.5 µm (eye-safe) • Measures radial velocity with better than 0.2 m/s accuracy • Temporal resolution < 10 s • Scanner used to measure in different directions to get speed and direction • Aerosols serve as the backscatter agent (primarily boundary layer systems) • Provides cloud base information needed by other instruments Multiple commercial vendors

25. Microwave Radiometer (MWR) • Automated, hardened instrument • Spectrally resolved downelling radiance on: • 22.2 GHz water vapor absorption line • 60 GHz oxygen absorption band • Temporal resolution < 10 s • Calibration for WV channels performed in field; the temperature channels need to be calibrated with liquid nitrogen every 3-6 months • Some sensitivity to clouds, but can profile in all non-precipitating conditions Multiple commercial vendors

26. Atmospheric Emitted Radiance Interferometer (AERI) • Automated, hardened instrument measuring downwelling IR radiation from 3.3 – 19 µm at 0.5 cm-1 resolution • Calibration accuracy better than 1% ambient radiance • Temporal resolution is 30-s • Spectral radiance provides the information for thermodynamic profiling (primarily clear sky) • Also cloud properties, trace gases, more Commercially available

27. Raman Lidar (used as research quality obs standard) • Built for the DOE Atmospheric Radiation Measurement Program • Designed for autonomous profiling of water vapor in the boundary layer • Deployed to SGP site in 1996, fully operational in 1998 • First Raman lidar to achieve 24 / 7 / 365 status • Diurnal difference in sensitivity is significant • Continuous upgrades improved uptime • 60% uptime from 1998 – 2004 • > 90% uptime from 2004 – today No Commercial Vendor

28. Observing System Simulation Experiment (OSSE) Configuration According to the Glossary of the American Meteorological Society: “An OSSE is an experiment in which a model is assumed to represent the exact behavior of the physical system (the atmosphere) of interest. The model results are then sampled in a fashion to mimic real or hypothetical observing systems. OSSEs are an inexpensive way to guide the design of new or enhanced observing systems or to evaluate the usefulness of existing observations.”

29. WRF Model Specifications • Weather Research and Forecasting (WRF) community model Version 3.0.1.1 • non-hydrostatic regional forecast model • Advanced Research WRF (ARW) dynamical core • TRUTH simulation • Initialized 00 UTC 07 Jan 2008 using the 20-km RUC analyses • 6-km horizontal grid with 52 vertical levels • 48 hour simulation on a single 798 x 516 grid point continental United States domain • Vertical grid spacing decreased from 50-100m in first km to ~500m near model top (65 hPa) • Thompson et al. (2008) microphysics scheme • Yonsei University planetary boundary layer scheme (Hong et al. 2006) • Dudhia (1989) shortwave radiation scheme • Rapid Radiation Transfer Model longwave radiation scheme (Mlawler et al. 1997) • Noah land surface model for surface heat and moisture fluxes • No cumulus parameterization scheme

30. Assimilation System Specifications • Employ the Data Assimilation Research Testbed (DART) assimilation system for all experiments • Assimilation algorithm is based on the Ensemble Adjustment Kalman Filter (EAKF) described by Anderson (2001). • Assimilation experiments • Initial 40-member ensemble valid at 12 UTC 05 Jan 2008 using 40-km NAM analyses was integrated 24 hours to increase spread • 18-km horizontal grid with 37 vertical levels • Kain and Fritsch (1990, 1993) sub-grid scale cumulus parameterization scheme used Different initialization datasets, grid resolutions and parameterization schemes were chosen for the assimilation experiments to limit the risk of performing “identical twin” experiments.

31. Observing System Simulation Experiment (OSSE) Configuration – Validation “Truth” At 12 hr assimilation point - 1200 UTC 07 Jan 2008 Observations were simulated at NWS WSR-88D Sites 100m AGL MIXR (shaded); SLP contoured every 2 hPa; 500m AGL total winds

32. Observing System Simulation Experiment (OSSE) Configuration – Validation “Truth” At 18 hr assimilation point - 1800 UTC 07 Jan 2008 Observations were simulated at NWS WSR-88D Sites 100m AGL MIXR (shaded); SLP contoured every 2 hPa; 500m AGL total winds

33. Observing System Simulation Experiment (OSSE) Configuration – Validation “Truth” At 24 hr assimilation point - 0000 UTC 08 Jan 2008 Observations were simulated at NWS WSR-88D Sites 100m AGL MIXR (shaded); SLP contoured every 2 hPa; 500m AGL total winds

34. Observing System Simulation Experiment (OSSE) Configuration – Validation “Truth” At 6 hr forecast point - 0600 UTC 08 Jan 2008 6-hr total precip (shaded in mm); 4km AGL wind; 850 hPaEq Pot Temp contoured every 2 K; 100m AGL MIXR (shaded); SLP contoured every 2 hPa; 500m AGL total winds

35. Observing System Simulation Experiment (OSSE) Configuration – Validation “Truth” At 12 hr forecast point - 1200 UTC 08 Jan 2008 6-hr total precip (shaded in mm); 4km AGL wind; 850 hPaEq Pot Temp contoured every 2 K; 100m AGL MIXR (shaded); SLP contoured every 2 hPa; 500m AGL total winds

36. Assimilation Experiments One sigma measurement error profiles that were applied to the simulated instrument profiles. One sigma wind magnitude error of 0.3 m s-1 applied at all levels.

37. Assimilation Impact- 850 hPa RMSE Comparison (T, Q, VRMS) Temperature [K] WV Mixing Ratio [g/kg] Vector Wind Speed [m/s] Throughout the assimilation period RMSE is lowest in cases when both additional thermodynamic and momentum profiler observations are assimilated simultaneously - Need Moisture Flux Observations

38. Assimilation Impact- Conclusions • Largest improvements to the BL analyses were made when DWL winds were assimilated jointly with temperature and moisture profiles from the other sensors • Smallest temperature and moisture errors generally occurred during the CONV-RAM and CONV-RD cases (especially the upper troposphere), although the CONV-MAD results were fairly similar in the boundary layer • The tendency for errors to be smallest when both moisture and wind observations are assimilated demonstrates that a more accurate moisture analysis is dependent not only on the availability of accurate moisture observations but also on a better wind analysis and associated improvements in moisture flux and moisture advection between analysis times

39. Forecast Overview • Forecast domain is same as the assimilation domain (i.e., entire CONUS) • Assessment sub-domain selected to determine forecast skill for primary heavy precipitation event

40. Forecast Assessment- 6hr Accumulated Precip valid 1200 UTC 8 Jan 2008 + Moisture Flux Conv. Valid 0900 UTC 8 Jan 2008 Conv+MWR+AERI+DWL Truth Conv+Ram+DWL

41. Forecast Assessment- Improvement over CONV in Moisture Flux (%)

42. Forecast Assessment- Traditional forecast skill statistics

43. Forecast Assessment- MODE Total Interest Values Evaluation Matched Pairs for 06-12 UTC Forecast

44. Forecast Assessment- Conclusions • Overall, improvements in the accumulated precipitation intensity forecast due to better representation of moisture transport result when DWL observations are assimilated, while improvements in the location forecast occur when assimilating thermodynamic observations due to better forecasts of localized moisture features and surface baroclinic zone location and intensity • Assimilating additional thermodynamic obs alone (without DWL winds) did not produce strong enough moisture transport to adequately predict the heaviest precipitation, nor were able to sufficiently reduce errors in the large-scale dynamic forcing of the event • Largest accumulated precipitation forecast improvements in both location and intensity were made when thermodynamic, moisture and wind observations were assimilated simultaneously with conventional observations in the CONV-RD, CONV-MD, CONV-AD, and CONV-MAD cases – Need Moisture Flux Observations

45. The National Mesonet Program In 2008, the National Academy of Sciences (NAS) released Observing Weather and Climate From the Ground Up: A Nationwide Network of Networks, which provided a framework for establishing a National Mesonet Following the release of the NAS report, Congress provided earmark funding for a pilot project focused on mesonets located in the south-central United States Pilot project efforts are focused on developing a comprehensive metadata (“data about the data”) capability for existing Mesonet data obtained via MADIS FY 2010 earmark funding will extend the pilot project to mesonets nationwide, consistent with the guidance provided in the earmark language Senate budget language released July 23, 2010 recommends FY 2011 earmark funding to sustain and further expand the effort Out-year funding is predicated on the development of robust financial and program planning, to include alignment with NOAA’s Next Generation Strategic Plan (NGSP) Extending Mesonet through the boundary layer is critical for forecast improvements 47 47

46. FY09 Omnibus • Following the November 2008 release of the National Research Council’s “From the Ground Up” report, Congress directs NOAA/NWS to establish a National Mesonet: • Maintain operations for existing mesonet network observation agreements • Procure competitively a national mesonet, which will provide the critical information needed to improve short- and medium term weather forecasting • Examine the use of real time mobile data collected from commercial trucking operators • Total Appropriation: $11M • Global Science and Technology (GST) awarded contract for “National Mesonet Pilot Project” • Develop prototype capability (‘plumbing’) to provide surface atmospheric and soil moisture/temperature data with focus on enhanced metadata tool • Support metadata exchange between data providers and applications developers • Focus on documenting metadata for wind, temperature, and soil moisture 48

47. FY 2009 Pilot Project: Networks • Metadata effort follows ISO 19115 for definition of attributes and SensorML for instrument description • Metadata attributes/fields include • Station attributes (e.g., geospatial) • Data Communication attributes • Siting/Exposure attributes • Instrument Attributes • Data Logger Attributes • Quality Control Attributes 49

48. FY10 Omnibus • The FY10 Omnibus Appropriation provides $19M for NOAA/NWS to: • Maintain data procurements from existing surface in-situ mesonet observations, including those added as a result of FY09 expansion • Competitively expand the National Mesonet …to include integration of existing networks…75% urban… 25% non-urban areas • Expansion of the Meteorological Analysis and Data Ingest System (MADIS) • MADIS: One IT Pathway for National Mesonet • Establish National Mesonet Program Office for program oversight and data utilization initiatives • Contracts have been awarded/finalized for “continuation” and “mobile” efforts • FY 2011 Senate budget language (earmark language) • Continue and expand program to additional networks • Establish National Mesonet “Testbed” for both in-situ and ground-based remote sensing • MADIS enhancements • Very tentative given FY11 budget environment! 50

49. From: EMC OSE by Ralph Petersen, Geoff Manikin and Dennis Keyser Tropospheric Improvements greater than those in changing RUC resolution from 40 to 20 km • Significant improvement by including Ascent / Descent data • Positive effects at all levels on Winds, Temp and RH • Above 25,000’, impact comparable to analysis differences • Below 25,000’, impact slightly smaller, but still noteworthy