Introduction
============
Early theory: only intrinsic Q, Qp ~ 2.5 Qs 
Amplitude decay of a single wave: exp(-a(f) x); Q=(pi f)/ (a(f) c)
Spec. case - Maxwell: Q=Q(f)= const.f   !! abandoned !!
Experiment: Q almost independent on f (obtained if a(f)=const.f)
Real data: strong freq dependence of apparent (total) Q, 
where Q=Qo f**n, n<=1
Apparent Q(f) explained by scattering: 1/Q=1/Qi + 1/Qsc

campillo_plantet_1991.pdf
=========================
France, 200-1000 km, M 2-4, freq 1 to 10 Hz
Qp, Qs are apparent Q from P and S waves
P means Pg, S means Lg (coda of various durations)
Fig. 8,9 => Qp, Qs strong power-law freq increase; ~ 1000
Fig. 10 => Q(f) corrected for Qi=const = 1500 makes Qsc(f) linear
Fig. 7 => 'surprise' Qs > Qp (but 'not much'; Qs ~ 1.5 Qp)

mancilla_2012.pdf
================
Q in mantle from a 600-km deep Mw 6.3 Granada event 2010
f~ 1 - 10 Hz
data "corrected" for crustal Q => mantle Q 
Qp, Qs apparent; no separation into Qi and Qsc
Spectral decay method, Eq. 3, 
   applied to 60s of coda after P and 60s after S
Power-law Q= Qo f**n, Qo ~ 100, n~0.6-1.0
Again: Qp < Qs [and references to others] 


margerin_1999.pdf
================
Mexico, distances < 50 km; magnitude > 4, freq 1-15 Hz
time decay of coda up to 200s
Fig. 1: model; Fig. 2: data fit (1/t diffusion, exp() loss into mantle)
Qsc is denoted as Q*; Q is not separated into Qi and Qsc
Fig. 8 and 9: an ad-hoc Qi ~ 1000 (freq independent) agrees with data   

Davis_et_al-2007.pdf
===================
California, M 2 to 4, around 10 Hz
Fig. 8: total coda Q (=apparent Q), tomography, 
	Q strongly varies spatially  ~100 to 1000 at 10 Hz
Fig. 9a, 9b: explaining Q in terms of multiple reflection 
	(high Q for high freq in Fig. 9a) 
Q is measured, Qsc is modeled by "telegraph equation" Eq 18, 
	Qi is inferred (Qi > 3000 !) 
Conlcusion: Qsc is due to scattering in upper crust 
	(hence local variations)
INTERESTING: 
Point 25: "it is well known .. Qi cannot be separated from Qsc"


gaebler2015a.pdf
================
Table 1 - overview of many papers; Qi ~ 300 to 1000
Theory Monte Carlo solutions to Radiative Transfer Equation; 
Random medium described by its ACF
Fig. 5 and 6  example; Particle counting method; loss of energy due to Qi; 
	no Qsc!! heterogeneous crust !
Random medium produces coda waves with apparent Q. 
	Intrinsic Qi is a model parameter. 
[They assume a fixed theoretical ration of QiP/QiS=2.24, 
	Qi freq independent]
Distance < 200 km, M 3-4 (West Bohemia), coda envelopes
REGIONAL and TELESEISMIC records; TRANSATION and ROTATION. 
RESULT Qi (S waves) = 580-750.
Chap 6 Discussion: Comparisons

gaebler2015b.pdf
================
West Bohemia; M>3; stations CZ and D; Coda envelopes up to 50s
Modeling BOTH elastic and acoustic RTE with similar (!) results
homogeneous background velocity model
Freq 3-24 Hz
Acoustic: scalar waves; dominace of S => applicable to S only
Effect of Qi by multiplying seimogram by exp(-bt), b= (2 pi f)/ Qi 
Simultaneous inversion for the source and site terms (eq 1)
Again assuming Qp/Qs. 
Sec. 5.1: The Qi and Qsc by the parameter fitting. Qi freq dependent.
Results: Table 3 !!!!

mousavi_2017.pdf
================
West Bohemia; 
Qp is Q from P waves, a 2.56s window around P, !NO CODA AT ALL!!
Fig 3: example waveforms
Qp(f) not considered
Path-averaged t* calculated from amplitude spectra of the 2.56s window  
Spectra inverted by eq. 3. Path-dependent t* further inverted for
spatial Q-tomography
Fig 2: stations and ray coverage
Fig 9: RESULT: Q~100-300, low-Q anomaly below Novy Kostel
Conclusion: Much lower Q than in Central Europe;
But their Q for more distant stations show 'normal' Q~500 


