Trends in the Upper Stratosphere - Lower Mesosphere Philippe Keckhut, Chantal Claud, Bill Randel, NOAA/CPC
Temperature trends Rockets 8°S-34°N Lidar OHP 44°N Significant trends 1-3 K/decade Homogeneous from 8°s to 44°N Keckhut et al., J. Geophys. Res., p447, 1999 Beig et al, Rev. Geoph., 2003 STTA-SPARC Ramaswamy et a., Rev. Geoph., 2001 MTTA/IAGA-ICMA Beig et al, Rev. Geoph., 2003
Data available SiteLat.PeriodCont.instrumentReference La Réunion Low latitude stations Table Mountain, US Wallops Island, US Ryori, Japan OHP, France Hohenpeissenberg, Gr Volgograd, Russia Heiss Island NCEP SSU Nash 21°S 8°S-34°N 34.4°N 37.5°N 39°N 44°N 47.8°N 48.7°N 80.6°N Global Zonal Yes No Yes No Yes No Yes Lidar NDSC Rocket US Lidar NDSC Rocket US Rocket Japan Lidar NDSC Rocket US SSU Fadhuile et al. Keckhut et al., 1999, Baldwin et al., Schmidlin et al., 1998 Keckhut and Kodera, 1990 Ramaswamy et al., Kubicki et al., submitted- Keckhut et al., Keckhut et al., 2001 Austin et al., Ramaswamy et al ….
Instrumental changes on soviet rocket Volgograd sensor changes Estimated from the time serie analyses Estimated from the aerothermic calculations Raw data Corrected data Kubicki et al., submitted toJASTP, 2004.
Tidal interferences They induce large interferences in data comparisons, trends and satellite validations 6K Keckhut et al., J. Geophys. Res., p10299, 1996 Keckhut et al., J. Geophys. Res., p447, 1999 Solar heating or other biais related with Solar time +
Tidal interferences Volgograd Time of launch Averaged temperature km 2:00 10:00 15:00 Kubicki et al., submitted toJASTP, 2004.
Global trend estimates NCEP/NOAA analyses appear to be biased by tides SSU-Nash is probably not biased, but provides only zonal means. NCEP analyses at 1 hPa (≈50 km) Keckhut et al., J. Geophys. Res., p546, 2001
Tidal variability The use of tidal models ? Tidal variability ? Data assimilation? Tidal representation Reproduce realistic tides Tidal variability is around 20% Tidal variability can be reproduced with realistic ozone and water vapor fields and planetary waves forcing 3D Rose/Reprobus model at SA Diurnal Semi-diurnal Morel et al., JASTP, p251, 2004
The multi-parameter regressions (AMOUNTS) ( Hauchecorne et al., 1991; Keckhut et al., 1995) To evaluate temperature trends and variability (for data and model outputs) and Spurious changes, it is necessary to parametrize the variability: T(t) = m + St + ATrend + BSolar + CQBO + DENSO + EAO + F.Step(ti) + Nt The A, B, C, D, E, F terms represent the amplitude of trends / factors of variability and bias. The residuals (AR(1)) include all the variability not considered in the parametrization. The analysis of the residual terms : model inadequacies the degree of confidence of the analysis Data are filtered according to time of the day
Heiss Island 81°N
Heiss Island Trends Trends are similar with Volgograd
Trends as a function of latitude Volgograd OHP, _ _ Wallops, --- Riory, …. US tropical °°°° US tropical Wallops OHP Volgograd Riory Summer Winter Kubicki et al., submitted toJASTP, US tropical: 8°S-34°N Wallops Island: 37,5°N Ryori, Japan: 39°N OHP, France 44°N Volgograd 49°N
Updated stratospheric temps from SSU/MSU * Thanks to John Nash (MetO), Jim Miller, Mel Gelman and Roger Lin (NOAA CPC) note ‘flattening’ of trends near stratopause small long-term cooling in middle stratosphere MSU4 SSU15x SSU25 SSU35x SSU47x
Upper stratosphere: SSU vs. HALOE SSU 47x ~43-57 km HALOE integrated to approximate SSU 47x
Stratospheric Sounding Unit (SSU) * operational measurements since 1979 * ~10-15 km thick layer temperatures synthetic channels
SSU 35X, 36X, 47X (35, 42, 50 km)
Other NDSC lidar data sets 35X 36X 47X
La Réunion lidar station 21°S Warming of around 2 K/decade In agreement with SSU Only 10 years
Bill: How do we interpret stratopause variability? ? ~43-57 km
Bill: How do we interpret stratopause variability? ? ~43-57 km
Methodologic error The methodologic error is strongly related to the noise level After 20 years methodologic errors are negligeable Non linear trend estimates are possible –on domain 2 and depend on data length and residual noise and –on domain 3 only related to residual noise Kerzenmacher et al., QJSRT, 2005
Conclusions Three types of data sets –Rocket –Lidar –SSU Discontinuities –Rocket : pb with time of measurements and sensor changes –Lidar : autocalibrated, darktime measurements, larger biais around 30 and km –SSU/NCEP: Tides R and L in good agreement with SSU around 50 km larger trends in tropics around km Flattering after 1995 in good agreement with NDSC Lidars. Need more investigations. How to take into account slope changes: use ozone data as forcing.