1 / 28

Global historic in situ upper air data for climate change research

Global historic in situ upper air data for climate change research. Ramella Pralungo Lorenzo & Haimberger Leopold . Project task:

farren
Télécharger la présentation

Global historic in situ upper air data for climate change research

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Global historic in situ upper air data for climate change research Ramella Pralungo Lorenzo & HaimbergerLeopold Lorenzo Ramella Pralungo - April 2012

  2. Project task: • comprehensive homogenization of the global upper air dataset, which consists of observed time series (T, Wind, RH, SH) coming from radiosondes and tracked balloons (1905 onward). Summary • Method: • development of an unified automatic homogenization system that analyzes and adjusts upper air temperature, wind and humidity datasettogether, using background forecasts from reanalysis as reference. • Scientific goals: • better knowledge of climate from 1920s onward; • improvement in climate monitoring and forecast; • homogenized data as input to ECMWF reanalysis; • use the homogeneous series as test reference for climate models. Lorenzo Ramella Pralungo - April 2012

  3. Temporal data coverage: Data coverage raw time series reanalysis time series • raw data from 1920 (some stations with data in 1905...) up to now; • only one re-analysis back to 1900 and the other more recently but more accurate; Lorenzo Ramella Pralungo - April 2012

  4. The radiosonde/balloon archives are: Input data • ERA-40 (1957-2002), ERA-INTERIM (1979-2011) raw data with departures, provided from ECMWF. • CHUAN contains profiles from almost 4000 station records from 1904-2007. Altitude/pressure level data, often at asynoptic times. • The data have been interpolated, rearranged, merged and stored as NetCDF files to obtain a global and homogeneous archive. The global background forecast archives are: • ERA-40 (1957-2002), ERA-INTERIM (1979-2011); • 20th Century Reanalyses (1900-2011, uses only surface data, provided by NOAA). Lorenzo Ramella Pralungo - April 2012

  5. U wind component: Bronwsville (USA) 72250, January 1963, Vertical profile log(p) vs altitude CHUAN and ERA The conversionaltitude => pressure uses as reference the 20CR for the geopotencial. Lorenzo Ramella Pralungo - April 2012

  6. U wind: Bronwsville (USA) 72250, 1963 Correlation CHUAN vs ERA-40 CHUAN and ERA After the conversion Alt => Press CHUAN fits with the ERA. Since the archives are omogeneus, they could be merge in a global one. Lorenzo Ramella Pralungo - April 2012

  7. The radiosonde/balloons archives ERA-40, ERA-INTERIM and CHUAN are merged, to obtain a homogeneous archive, with rules: The merged archive • Lat, Lon and Alt must be the same in the source archives; • Data priority: (1) => ERA-Int, (2) => ERA-40, (3) => CHUAN • Spikes: if departures(day,press,time) > 4mean_dep(press) => obs dropped out The time series VIEWER INTERESTING: TEMPERATURE Station: 010393 ->1905 Station: 004018, 061641,007145 U WIND Station: 072270 -> @850hPa Station: 093112 -> @700hPa Wind Speed Station: 016716 -> @200hPa Lorenzo Ramella Pralungo - April 2012

  8. Homogenization: remove artificial breaks and trends Starting point is theautomatic homogenization method for upper air data called RAdiosoneOBservationCOrrection using REanalyses (RAOBCORE, Haimberger 2007b) daily data advantages BG weakly dependent from OBS RAOBCORE drawback BG are satellite data dependent (itself biased) These trends: • have non-climatic reasons; • are due to improvements/changes in observation practice; c Radiosonde and tracked balloons observations need homogenization Lorenzo Ramella Pralungo - April 2012

  9. Long Temperature time series are tested using 20CR as reference. RAOBCORE v2.0 The 20thCentury Reanalysis: • dependents only on surface pressure and temperature data: the references are totally independent from the analysed data; • is quite smooth since it is an ensemble mean: it may suffer in the detection of fast phenomena; • shows high variance, especially at the upper pressure levels: the departures may be noisy and rather large; • evidences unexpected warm and cold pools to respect ERA-40/Int: they may introduce artificial shifts in the corrected time series. Lorenzo Ramella Pralungo - April 2012

  10. Bronwsville (USA) RAOBCORE 2.0 first results Big inhomogeneities are highlighted by the statistic test The big trend in the ‘50s is well balanced out Strong correction is applied only in the break region. Lorenzo Ramella Pralungo - April 2012

  11. Bronwsville (USA) RAOBCORE 2.0 first results Inhomogeneities are highlighted by the statistic test Visible in an improvement in the time series. Soft corrections are applied only in the break region. Lorenzo Ramella Pralungo - April 2012

  12. using 20CR using ERA Bronwsville ( USA) RAOBCORE 2.0 first results • In the ERA BG(12)-BG(00) is visible the jump due to the satellite data acquisition. • In the same period the 20CR is smooth. Lorenzo Ramella Pralungo - April 2012

  13. Angra do Heroismo (Portugal) RAOBCORE 2.0 first results • The time series look well homogenized: also the really early data for which no metadata are available; • In the adjusted data a negative offset is present: may it be imputable to a 20CR bias? Lorenzo Ramella Pralungo - April 2012

  14. Keflavik (Island) using 20CR using ERA RAOBCORE 2.0 first results • goodbreaks improvement; • not well corrected the break in the1950 (too few data); • extreme warm shift induced by a 20CR bias. • not well corrected the break in the1975-79 ( bias in ERA ); Lorenzo Ramella Pralungo - April 2012

  15. Keflavik (Island) using 20CR using ERA RAOBCORE 2.0 first results • good breaks improvement; • a cold shift is induced by a 20CR bias. • not well corrected the break in the1975-79 ( bias in ERA ); Lorenzo Ramella Pralungo - April 2012

  16. There are strong offset between 20CR and ERA archives. RAOBCORE v2.0 actual problem Keflavik (Island) T departures (OBS-AN) @ 200hPa, 00 T departures (OBS-AN) @ 500hPa, 00 ERA-40 ERA-Int 20CR ERA-40 ERA-Int 20CR 0.8 K 6.5 K Lorenzo Ramella Pralungo - April 2012

  17. RAOBCORE 2.0 Next improvements: • implement a unified homogenization method for simultaneous analysis of Temperature, Wind and Humidity; • the impact of different background forecasts (20CR, ERA-Int and ERA-40) time series needs to be investigated further; • improve the algorithm for break point detection and correction (from fixed length temporal window to dynamic temporal window, penalized maximums likelihood methods(Caussinus and Mestre 2004 and Hadzimustafic 2010)). Lorenzo Ramella Pralungo - April 2012

  18. Thanks for your attention Questions? Suggestions? Lorenzo Ramella Pralungo - April 2012

  19. It is necessary rearranging the dataset to the ERA archive standard Progress 2/5 The Comprehensive Historical Upper Air Network (CHUAN) has been rearranged in pressure level and at synoptic time (00 and 12) The CHUAN archive contains: uncorrected altitude (wind observations) and pressure (temperature observations) radiosonde-data and the launch times are variable. • To homogenize the series: • the data must be comparable in pressure level and observation time with ERA archives; • the departures from the reanalysis are essential. Lorenzo Ramella Pralungo - April 2012

  20. Progress 3/5 The Comprehensive Historical Upper Air Network (CHUAN) BASIC IDEA Using the 20Century Reanalysis (provided by NOAA) as reference: 1) it is possible to standardize the archive: 1.1) with a altitude to not standard pressure conversion; 1.2) with a not standard pressure to standard pressure interpolation; 1.3) with a time interpolation; 2) it is possible to obtain the analysis departures, prerequisite for the homogenization. Lorenzo Ramella Pralungo - April 2012

  21. Progress…details 3.1/5 The Comprehensive Historical Upper Air Network BASIC IDEA 1) It is necessary a altitude to pressure conversion using the 20 Century Reanalysis as reference: • The 20CR is gridded 2° X 2° • Bilinear interpolation to the CHUAN station • Thermal Wind Relation • But Tm is not well determinable; • 1.1) Since the geopotential is proportional to the pressure logarithm, using a weight function: • It is possible obtain the weight a to determinate the corresponding pressure: • 1.2) The is not standard • Linear interpolation to the standard level CHIAN archive standarized in pressure Lorenzo Ramella Pralungo - April 2012

  22. Progress…detail 3.2/5 The Comprehensive Historical Upper Air Network BASIC IDEA 1.3)Because before 1957 the launch time was not standard, a temporal interpolation is needed: • Since the 20CR is gridded by 6 hours, the cubic interpolation can be used: • If CHUAN observation 2)To compute the departures, one ASSUMPTION is necessary: • the departures are constant in time • the time interpolation does not influence the departures 3)To obtain the CHUAN observation at the synoptic time 00 and 12: * closest CHUAN observation to t00 or t12 Lorenzo Ramella Pralungo - April 2012

  23. CHUAN time interpolation 3.3/5 T or U or V The 4 closest observations to the analysis time are used for the interpolation analysis time * CHUAN 20CR time [h] 12 18 0 12 18 0 0 6 6 tc ta time window 20_CR & CHUAN interpolation Lorenzo Ramella Pralungo - April 2012

  24. Lorenzo Ramella Pralungo - April 2012

  25. Lorenzo Ramella Pralungo - April 2012

  26. Lorenzo Ramella Pralungo - April 2012

  27. Lorenzo Ramella Pralungo - April 2012

  28. Angra do Heroismo 008508 =>[38.73;-27.07], Portugal RAOBCORE 2.0 first results RAOBCORE 2.0 using 20CR The big breaks around 1970 are solved applying strong and sharp corrections at 30 hPa.The other breaks suffers for not sufficient adjustments. RAOBCORE using ERA The corrections are in general more gradually and they fit better to homogenize the serie. Lorenzo Ramella Pralungo - April 2012

More Related