The Tropospheric Humidity Trends of NCEP/NCAR Reanalysis before
Satellite Era
Shi-Keng Yang (SKY)
Masao Kanamitsu
Wesley Ebisuzaki
Gerald Potter
Sept-Oct, 2003, CWB, Taipei
NCEP/NCAR REANALYSIS lwcf 4901-7812 12-month running mean
31
31.5
32
32.5
33
33.5
34
34.5
49 54 59 64 69 74
Year
W/M
**2
The Problem
Outline
• LWCF CSOLR & OLR RH• Natural variability from AMIP ensemble• RH & T trends sampled from 30 Stations• Chronology of measurement changes• A Hygrometer Simulator • Conclusion• Ideas for Taiwan Area Regional Reanalysis
LWCF = CSOLR-OLR
NCEP/NCAR REANALYSIS CS OLR Anomaly 4901~9812 12-month running mean
-10
-8
-6
-4
-2
0
2
4
6
8
10
49 52 55 58 61 64 67 70 73 76 79 82 85 88 91 94 97
Year
CS O
LR W
/M**
2Trop CS OLR
SH mid CS OLR
NH Mid CS OLR
NCEP/NCAR REANALYSIS OLR Anomaly 4901~9812 12-month running mean
-10
-8
-6
-4
-2
0
2
4
6
8
10
49 52 55 58 61 64 67 70 73 76 79 82 85 88 91 94 97
Year
OLR
(W/M
**2)
tropics OLR
SH mid lat OLR
NH mid lat OLR
-6
-4
-2
0
2
4
6
Year
RH
%
49 52 55 58 61 64 67 70 73 76 79 82 85 88 91 94 97
Tropics
SH mid lat
NH mid lat
NCEP/NCAR REANALYSIS 500mb RH Anomaly 4901~9712 12-month running mean
What are the causes?
• Natural variability?• Instrument changes?• Algorithm/System changes? Station increases?
The Attributes of Reanalysis GDAS and the Model used for AMIP runs
Reanalysis AMIP-ensemble10
• Convection Scheme SAS RAS• SW Radiation Lacis & Hansen (1974) Chou et al (1992, 96)• Boundary Layer Local Diff Non-Local• Orography Mean Smooth Enhanced• Resolution T62L28 T42L24• Soil Moisture w/ nudging interactive• Snow Obs (fixed on ‘72) Climatology• Radiation Resolution Linear grid/hourly
Guassiagrid/hourly
30 US Stations Suggested by J. Christy
Red – 300 hPa
Yellow – 500 hPa
Green – 700 hPa
Black – 850 hPa
Red – 300 hPa
Yellow – 500 hPa
Green – 700 hPa
Black – 850 hPa
Green – 75~79 RH mean
Black – 49~53 RH mean
Green – 76~78
Black – 49~51
T Profiles from 36-month means
RH differences
Green – 76~78
Black – 49~51
RH Profiles from 36-month means
T-dif Profiles from 36-month means
Taiwan Area R-1 Temperature Anomaly
Taiwan Area R-1 RH Anomaly
R-1 Taipei Relative Humidity Anomaly
R-1 Taipei Temperature Anomaly
• A Chronology of changes in the US Radiosonde Network from Elliott and Gaffen (1991) plotted on LWCF anomaly to demonstrate the probable impacts.
• 1: 1957 observation time changed on IGY;
• 2: 1960 introduced white-coated temperature elements;
• 3: 1965 Introduced carbon humidity element;
• 4: 1969 changed from manual system to a time-share computer system;
• 5: 1972 redesigned relative humidity ducts introduced;
• 6: 1973 “motorboating” lower RH values as 19% when measured lower than 20%;
• 7: 1974 introduced semi-automatic mini-computer-based system;
• 8: 1979 Satellite measurements incorporated into Reanalysis;
• 9: 1980 New carbon hygristors introduced;
• 10:1985 Introduced fully automatic mini-computer-based system;
• 11:1988 Introduced Precalibrated hygristor replacing individual preflight calibration;
• 12:1988 Introduced new VIZ sonde with new humidity duct;
• 13:1989 Introduced fully automatic micro-computer-based system.
Sensor in Use
• Goldbeaters Skin – (Russia, China)
• Carbon Hygristor – (USA[part], India, China?)
• RS80 Humicap A or H – (50% of global net work)
• RS90 Humicap – (Finland +??)
• Meteolabor “Snow White” – (working reference)
Challenge of Radiosonde water vapor measurements I
• Very large range of saturation vapor pressure with temperature– 20 oC 23 hPa– -20 oC 1.2 hPa– -50 oC 0.04 hPa – -80 oC 0.0005hPa
Challenge of Radiosonde water vapor measurementsII
• Sensor temperature induced error– @ 20 oC 6%/deg wrt water – @-20 oC 7%/deg wrt water– @-50 oC 7%/deg wrt water
Radiosonde intercomparison experimentJohn Nash, UK Met Office
Radiosonde intercomparison experimentJohn Nash, UK Met Office
Radiosonde intercomparison experimentJohn Nash, UK Met Office
Radiosonde intercomparison experimentJohn Nash, UK Met Office
Hygristor Time constants
• Carbon hygristor 1~2 sec at sfc
• Thin Film 1~2 sec at sfc
• Goldbeater, Hair >10 sec at sfc, 5min 300 hPa
Hygrometer simulatoran educational toy
(T) = 3+ 5(15-T), T< 15 C
RH(i) = RHe + (RH (i-1) - RHe ) exp (-t/ (T))
Time-lag constant of a hygristor is a function of temperature. Based on Nash and Schmidlin, 1987, we determine that the rate of increase to be 5 sec/ K, so that the time-lag constant can reach 5 minutes at 300 hPa, where the temperature is lower than -40oC. Balloons ascend at 15 f/s.
Hygrometer Response Lag
100
200
300
400
500
600
700
800
900
1000
-60 -40 -20 0 20 40 60 80 100
T in C, and RH in %
Hei
gh
t in
hP
a
1000 hPa
900 hPa
800 hPa
700 hPa
600 hPa
500 hPa
400 hPa
300 hPa
Temp in C
Conclusion
• The trend of upper air humidity within NCEP/NCAR Reanalysis appears to be an artifact, caused by long time-lag constant in the older hygrometers, and other factors.
• Implies that similar problems, in general, in the time- series radiosonde time-series before the satellite era.
• Significant implication on the earth energy balance and cloud fields of the Reanalysis.
• No humidity climatology yet!• Suggestions: future Reanalyses includes a special fixed observation
system sub-analysis using only the limited well-known, high quality, well calibrated, fixed number stations for GDAS, such that a baseline reference analysis for the full analysis can be established .
Ideas for Taiwan Area Regional Reanalysis
• Homogeneous High Resolution Environmental Database suitable for comprehensive applications (3000 papers published using R-1)
• An Integration of modern analysis-assimilation system with historical data. Optimization the use of historical data.
• Periodical Benchmarks of Science-Technology advancement
Some guidelines• Build a system that you can do analysis the 2nd easier than the 1st time• Analysis- ReAnalysis- On-going Reprocess• Analysis Design –Evaluation Cycle ( once every 5~10 years for global)• Modular Analysis System Design• Calibration of measurements and Products• Multiple Stage Process• Multiple Stage Monitoring [ level 3 -> 2 -> 1; 100M/sec max cap by a
person]• Multiple Stage Archival • Other principles
– Process team is the analysis team, stack holder– Periodical Outside Evaluation– Re-packaging at each cycle – Start with EDR, and recycle for CDR– Mutiple algorithms if possible.
Status: NCEP Regional Reanalysis
Geoff DiMego (NCEP/EMC)
Eugenia Kalnay (U Md)
Fedor Mesinger (UCAR)15 March 2000
NCEP
Where America’s climate and weather services begin
Office of Global Programs
O B J E C T I V E
• A long-term set of consistent climate data produced at a regional scale
• 51 Years of the NCEP/NCAR Global Reanalysis have been completed!
Regional Reanalysis Rationale
• NCEP’s Meso Eta Model produces more accurate data assimilation and forecast results than the current global system.
• Improved data sources and techniques are available to improve upon the Global reanalysis
• Higher resolution for North American domain– 32 km / 45 layer Meso Eta regional reanalysis– T62 (~200 km) / 28 layers global reanalysis
Eta 2m temperature Eta 2m specific humidity
AVN 2m temperature AVN 2m specific humidity
Eta skin temperature Eta soil moisture 0-10cm
AVN skin temperature AVN soil moisture 0-10cm
Regional Reanalysis Features
• Unique resources can be added to the global reanalysis baseline:– assimilation of observed precipitation & cloud– direct analysis of satellite radiances with
3DVAR– mature Meso Eta Model due, in part, to GCIP
supported developments
• Additional data sources • 22 year period to be reanalyzed: 1982-2003