Monitoring the quality of global radiosonde Monitoring the quality - - PowerPoint PPT Presentation

Monitoring the quality of global radiosonde Monitoring the quality of global radiosonde humidity data using ground humidity data using ground-

• based GPS

based GPS measurements measurements

Junhong (June) Wang Junhong (June) Wang Liz Zhang Liz Zhang NCAR/EOL NCAR/EOL

Collaborators: Aiguo Dai (NCAR), Teresa Van Hove and Randolph H. Ware, COSMIC/UCAR Acknowledgement: Joey Comeaus (SCD), Dennis Shea (CGD), Imke Durre (NOAA/NCDC) Thanks NCAR Director Office’s support through NCAR Opportunity Fund

GOALS

1)To develop an analysis technique for deriving a global, 2-hourly data set of atmospheric precipitable water (PW) using existing ground- based GPS measurements of zenith path delay (ZPD), 2)To use GPS PW data to monitor the quality of global radiosonde humidity data and estimate the diurnal sampling errors in twice-daily radiosonde humidity.

How does it work and Why using GPS data? How does it work and Why using GPS data?

Total delay = Ionosphere + dry + wet

• All weather
• Continuous measurements
• High temporal resolution
• High accuracy (~1-2 mm)
• Long term stability

DATA: DATA: Global ZPD data: ~359 stations, 1997 Global ZPD data: ~359 stations, 1997-

• present, 2

present, 2-

• hourly

hourly

ANALYSIS TECHNIQUE AND VALIDATION

Ps from GPS surface-met data Ps from global surface synoptic

• bservations with adjustment

ZWD = ZPD - ZHD

Tm from ERA-40 with horizontal and vertical interpolation

Output: PW = ∏ * ZWD ∏ = f (Tm) Comparisons with radiosonde, MWR and others from field experiments for validations

Input: ZPD = ZHD + ZWD

) , ( 2779 . 2 H f P ZHD

s

λ × =

m m

T T PW PW ∆ ≈ Π ∆Π = ∆

∑ ∑ ∫ ∫

= =

∆ ∆ ≈ ≡

N i i i vi N i i i vi v v m

z T P z T P dz T P dz T P T

1 2 1 2

“Global Estimates of Water-Vapor-Weighted Mean Temperature of the Atmosphere for GPS Applications” (Wang et al. 2005)

• 1. Radiosonde data: The Integrated Global Radiosonde

Archive (IGRA) from NCDC, 1938 to present.

• 2. ERA-40: ~1.125°X1.125° (TL 159), 60 vertical levels, 6-

hourly, 1957-2002.

• 3. NCEP/NCAR reanalysis (NNR): ~1.875°X1.875° (T62),

28 vertical levels, 6 hourly, 1948-present.

• 4. Bevis Tm-Ts relationship: Tm = 70.2 + 0.72*Ts

ERA40-IGRA NCEP/NCAR-IGRA

Annual mean Tm difference between reanalysis and IGRA

• 10% and 16% of stations

with |∆Tm| < 2K for ERA-40 and NNR

• ERA40 better than NNR
• ERA40: a better option for

global estimate of Tm

Annual mean Tm difference between Bevis and IGRA

~1-6K colder with maximum marine stratus regions Warmer with maximum over Mountains

Diurnal biases in Tm_Bevis

Tm = 70.2 + 0.72*Ts

Amplitude of Tm/Ts diurnal cycle

Tm_Bevis Tm_ERA40 Ts

12 UTC 00 UTC 18 UTC 06 UTC

Six-year (1997-2002) seasonal mean of PW (mm)

Monitoring of the “health” of global radiosonde humidity data

2001 (DeltaElevation < 100m; Distance < 50 km)

• 3
• 2.5
• 2
• 1.5
• 1
• 0.5

0.5 1 1.5 2 2.5 3 3.5 4

6 8 9 6 1 7 6 7 4 2 2 9 4 2 9 4 9 6 9 9 6 1 6 1 4 4 9 3 9 8 6 8 2 2 1 8 9 5 7 1 9 4 9 7 5 8 9 5 6 4 7 1 8 1 4 1 8 7 1 9 1 3 9 3 1 1 2 9 4 9 9 8 6 4 4 7 1 4 9 4 6 1 7 8 1 6 3 8 8 2 7 3 5 7 2 4 3 7 2 2 2 7 2 6 1 9 1 5 9 2 6 1 9 9 8 1 5 4 2 6 2 6 1 2 3 7 4 9 1 9 3 8 5 5 5 9 1 5 1 4 6 3 5 7 4 9 4 5 4 5 1 1 5 7 3 6 3 3 3 4 5 2 9 5 7 2 2 1 8 2 4 4 3 2 9 5 4 7 1 2 2 9 1 2 1 2 7 8 3 9 7 9 1 3 6 6 8 9 5 3 2

PW (mm RAOB-GPS) RS80A RS80H RS90 Shang MRZ VIZ-type Meisei

RS80A VIZ-type

Diurnal PW sampling errors in twice Diurnal PW sampling errors in twice-

• daily

daily radiosonde data (U.S.) radiosonde data (U.S.)

Dai et al. (2002)

CONCLUSIONS AND FUTURE WORK

1) An analysis technique is developed to create a global, 2- hourly PW dataset. The technique needs to be validated and improved if necessary. 2) Preliminary analysis of PW differences between GPS and radiosonde data at 45 stations around the globe shows dry biases at most of Vaisala stations, but moist biases at all stations using carbon hygristor. This type of comparisons will be done to multi-year data, and more analyses are needed. 3) Errors in seasonal mean humidity due to under-sampling the diurnal cycle with twice-daily synoptic sounding are small (within ±2%) over the globe. More careful and detailed analyses will be done to quantify the error and its spatial and temporal distributions. The sampling error of

• nce-daily sounding will be also estimated.

Diurnal biases in Bevis Tm

Amplitudes of Tm diurnal cycle

Ts

Tm_Bevis Tm_RAOB

Comparisons of PW from GPS and radiosonde on Oct. 21-25, 2003 in La Jolla (9 km apart, 134/69 m elevations)

From NOAA/FSL (2004 Technical Review)

• 8. Monitoring of the quality of radiosonde humidity data for NWP

Six-year (1997-2002) seasonal mean of diurnal sampling errors

Diurnal sampling errors (%)

• f twice

daily radiosonde data