Use of PREMOS & LYRA for the reconstruction of the UV spectrum

Size: px

Start display at page:

Download "Use of PREMOS & LYRA for the reconstruction of the UV spectrum"

Kristopher Henderson
5 years ago
Views:

Lilensten (LPG, Grenoble) + special thanks

1 Use of PREMOS & LYRA for the reconstruction of the UV spectrum T. Dudok de Wit, G. Cessateur, M. Kretzschmar, L. Vieira (LPC2E, Orléans) J. Lilensten (LPG, Grenoble) + special thanks to the instrument teams (TIMED/SEE, SORCE/SIM, PROBA2/LYRA, ground-based instruments)

2 Solar spectral irradiance measurements LASP, Boulder 2

3 Solar spectral irradiance measurements PROBA2/ LYRA LASP, Boulder 2

4 Solar spectral irradiance measurements PICARD/ PREMOS PROBA2/ LYRA LASP, Boulder 2

5 irradiance [mw/m 2 /nm] altitude [km] X EUV UV VIS IR spectral irradiance altitude of absorption ! [nm] based on data from SORCE & TIMED ( ) 3

6 irradiance [mw/m 2 /nm] altitude [km] X EUV UV VIS IR spectral irradiance altitude of absorption ! [nm] based on data from SORCE & TIMED ( ) LYRA 3

7 irradiance [mw/m 2 /nm] altitude [km] X EUV UV VIS IR spectral irradiance altitude of absorption ! [nm] based on data from SORCE & TIMED ( ) LYRA PREMOS 215 nm 266 nm 536 nm 607 nm 782 nm 3

8 irradiance [mw/m 2 /nm] altitude [km] X EUV UV VIS IR spectral irradiance altitude of absorption relative variability [%] absolute variability [mw/m 2 /nm] based on data from SORCE & TIMED ( )! [nm] 64% of the variability from the near UV 34% from the visible relative variability (solar cycle) absolute variability (solar cycle) 4

9 irradiance [mw/m 2 /nm] altitude [km] X EUV UV VIS IR spectral irradiance altitude of absorption relative variability [%] absolute variability [mw/m 2 /nm] LYRA based on data from SORCE & TIMED ( )! [nm] 64% of the variability from the near UV 34% from the visible relative variability (solar cycle) absolute variability (solar cycle) 4

10 irradiance [mw/m 2 /nm] altitude [km] X EUV UV VIS IR spectral irradiance altitude of absorption relative variability [%] absolute variability [mw/m 2 /nm] LYRA based on data from SORCE & TIMED ( )! [nm] PREMOS 64% of the variability from the near UV 34% from the visible relative variability (solar cycle) absolute variability (solar cycle) 4

11 Our problem : interpolate the spectral irradiance 5

12 Our problem : interpolate the spectral irradiance wavelength PREMOS/PICARD, LYRA/PROBA2 EVE/SDO past now future time 5

13 Our problem : interpolate the spectral irradiance wavelength PREMOS/PICARD, LYRA/PROBA2 EVE/SDO past now future time 5

14 Our problem : interpolate the spectral irradiance wavelength PREMOS/PICARD, LYRA/PROBA2 EVE/SDO past now future time 5

15 Our problem : interpolate the spectral irradiance wavelength PREMOS/PICARD, LYRA/PROBA2 EVE/SDO past now future time 5

16 Our problem : interpolate the spectral irradiance wavelength PREMOS/PICARD, LYRA/PROBA2 EVE/SDO past now future time 5

17 Our problem : interpolate the spectral irradiance wavelength PREMOS/PICARD, LYRA/PROBA2 EVE/SDO past now future time 5

18 Interpolation is the problem... Forecas(ng (extrapola+on) difficult except for salient features ( talk by E. Quémerais) Temporal interpola(on difficult, since we s+ll know very li@le about the spectral variability during flares ( talks by J. F. Hochedez and M. Kretzschmar) Interpola(on in wavelength very tricky at high resolu+on (< 0.1 nm) because of atomic lines surprisingly easy otherwise 6

19 Outline Can we use channels from PREMOS & LYRA to reconstruct the Total Solar Irradiance (TSI)? the solar spectral irradiance? What does this tell us about the underlying physics? Such reconstruc+ons are required for space weather products (satellite drag...) and for upper atmospheric models 7

20 What PREMOS & LYRA could look like simulated daily outputs from PREMOS & LYRA (a\er data from SORCE/SIM, SORCE/SOLSTICE, TIMED/EGS & TIMED/XPS) f10.7 relative amplitude [a.u.] LYRA Al LYRA Hb PREMOS 1 PREMOS 2 PREMOS B PREMOS G time [years] PREMOS R TSI 8

21 Long term evolution Different trends in the long term evolu+on (11 year scale) : visible and infrared bands are out of phase with solar cycle [Harder et al., 2009] irradiance [W/m 2 /nm] solar spectrum phase delay vs f10.7 [deg] wavelength! [nm] phase versus f10.7, for 11-year cycle 9

22 Can the TSI variability be reconstructed from a few bands / proxies? 10

23 TSI variability Standard model for TSI variability TSI = facular brightening - sunspot darkening = α Mg II index - β daily sunspot area MgII amplitude [a.u.] DSA TSI Apr04 Jul04 Oct04 Jan05 Apr05 Jul05 time 11

24 TSI varibility But there are occasions where this simple model (chromospheric brightening photospheric darkening) breaks down LYRA Al LYRA Ly amplitude [a.u.] PREMOS 1 PREMOS 2 PREMOS B PREMOS R Chromospheric & Photospheric emissions are sometimes in phase = active network dominates in TSI variability MgII TSI Jul04 Aug04 Sep04 Oct04 Nov04 Dec04 time 12

25 TSI varibility But there are occasions where this simple model (chromospheric brightening photospheric darkening) breaks down LYRA Al LYRA Ly amplitude [a.u.] PREMOS 1 PREMOS 2 PREMOS B PREMOS R Chromospheric & Photospheric emissions are sometimes in phase = active network dominates in TSI variability MgII TSI Jul04 Aug04 Sep04 Oct04 Nov04 Dec04 time 12

26 TSI varibility But there are occasions where this simple model (chromospheric brightening photospheric darkening) breaks down LYRA Al LYRA Ly amplitude [a.u.] PREMOS 1 PREMOS 2 PREMOS B PREMOS R Chromospheric & Photospheric emissions are sometimes in phase = active network dominates in TSI variability MgII TSI Jul04 Aug04 Sep04 Oct04 Nov04 Dec04 time 12

27 Does the TSI vary in phase with solar activity? According to these simple models, the TSI should be in phase with indices for solar ac+vity (for +me scales >> 27 days) Is that really so? We check this by compu+ng the cross phase φ between the TSI and other indices for solar ac+vity φ = x(ω)y (ω) We use con+nuous wavelets. Solar ac+vity is expressed here by the DSA (Daily Sunspot Area) [Preminger & Walton, GRL, 2005]. 13

28 The TSI does NOT vary in phase with solar activity Phase lag (in days) between the TSI and the MgII index, and the DSA, based on 31 years of data diffusion TSI MgII lag time [days] 50 0 behind The TSI lags much more behind the DSA than faculae do (MgII index) 50 ahead characteristic scale [days] 11 years 14

29 The TSI does not vary in phase with solar activity! This result suggests that the excess of irradiance coming from plages and faculae con(nues to enhance the TSI a=er these regions have faded away. This is compa+ble with a diffusive decay of the magne(c field at ac+ve regions [Crouch et al., ApJ 2008]. Unresolved magne+c structures do ma@er... 15

30 Can the spectral variability be reconstructed from a few bands? 16

31 Spectral coherence The variabilty of the solar spectrum is remarkably coherent in spite of the complexity of the underlying processes f10.7 relative amplitude [a.u.] LYRA Al LYRA Hb PREMOS 1 PREMOS 2 PREMOS B PREMOS G time [years] PREMOS R TSI 17

32 Spectral coherence Many authors claim that the solar EUV variability is made of 3 contribu(ons [Lean et al. 1982; Woods et al., 2000; Warren et al., 2001; Feldman et al., 2010;...] quiet Sun + coronal holes + ac+ve regions Amblard et al. [2008] showed, using a sta+s+cal approach, that these can be described by 3 elementary spectra ( regions) quiet Sun + hot corona + cool chromosphere 18

33 Spectral coherence Many authors claim that the solar EUV variability is made of 3 contribu(ons [Lean et al. 1982; Woods et al., 2000; Warren et al., 2001; Feldman et al., 2010;...] quiet Sun + coronal holes + ac+ve regions Amblard et al. [2008] showed, using a sta+s+cal approach, that these can be described by 3 elementary spectra ( regions) quiet Sun + hot corona + cool chromosphere Can the full spectral variability (EUV-UV-visible) also be described that way? 18

34 Spectral coherence Our approach is empirical It is based on the Singular Value Decomposi+on (SVD) + bayesian blind source separa+on We find that > 60% of the variance can be described by a linear combina+on of only 3 contribu+ons (elementary spectra) instrumental noise sets in for > 3 contribu+ons 19

35 Spectral coherence Our approach is empirical It is based on the Singular Value Decomposi+on (SVD) + bayesian blind source separa+on We find that > 60% of the variance can be described by a linear combina+on of only 3 contribu+ons (elementary spectra) instrumental noise sets in for > 3 contribu+ons The full spectral variability EUV-UV-VIS can be adequately described by 3 elementary spectra! 19

36 Proximity map The closer two wavelengths are, the more similar their short term (< 90 days) evolu+on is facular brightening 10 w d f sunspot darkening i c p m LZ LA LL P1 LH P2 PG PR PB t wavelength [nm] Legend PB : PREMOS B PG : PREMOS G PR : PREMOS R P1 : PREMOS 1 P2 : PREMOS 2 LH : LYRA Herzberg LA : LYRA Al t : TSI f : f10.7 m : MgII w : MWSI d : DSA 0 20

37 Interpretation The UV spectrum (< 300 nm) can be properly described by LYRA channels and PREMOS nm channels Idem for visible part, with PREMOS visible channels The Near UV ( nm) is difficult to reconstruct. Unfortunately, this band has the highest absolute variability. Use SODISM? The Near IR (>900 nm) is also problema+c. w d f i c p m LZ LA LL P LH 500 P PG PR PB t

38 Spectrum reconstruction Reconstruc+on by using most PREMOS channels + LYRA Al + LYRA Lyman alpha w d 900 f i c p m LZ LA LL P LH 500 P PG PR PB t In the following we reconstruct the spectrum using channels marked with 22

39 Examples 2 examples of reconstructed irradiance amplitude [a.u.] 4 2 0! = 200 [nm] measured fitted Good reconstruction λ = 200 [nm] relative error < 15 % amplitude [a.u.] ! = 400 [nm] measured fitted Bad reconstruction λ = 400 [nm] relative error ~ 80 % year 23

40 Conclusions The EUV UV spectrum can be properly reconstructed using PREMOS & LYRA data We shall soon provide online nowcasts of the EUV UV spectrum (FP7 SOTERIA project) The reconstruc+on of the near UV is much more challenging. Instrumental errors from SORCE/SIM may be largely responsible for this. The combined use of PREMOS & LYRA will be par+cularly interes+ng for inves+ga+ng the spectral variability during flares. 24

41 Reconstruction error long term (> 90 days) short term (< 90 days) relative error "(!) [%] wavelength! [nm] 25

Recent Solar Spectral Irradiance Observations

Recent Solar Spectral Irradiance Observations M. Meftah 1, G. Cessateur 2, W. Schmutz 3, D. Bolsée 2, A. Shapiro 4, N. Pereira 2, A. Hauchecorne 1, L. Damé 1, A. Irbah 1, S. Bekki 1 (1) CNRS-LATMOS, France