C L1M = maximal allowable order in the expansion of the
C scattering matrix in generalized spherical functions,
C Eqs. (6)-(11)
C
C NG = number of quadrature division points and weights
C in the mu and mu_0 discretization of the reflection matrix,
C NG = N in Eq. (24)
C
C NQUADR specifies the type of quadrature formula used:
C = 1: special formula given by Eq. (25)
C = 2: Gaussian formula
C = 3: Markov formula
C
C NCASE specifies the type of output, as follows:
C = 1: computation of the (1,1) element of the reflection
C matrix in the scalar approximation
C = 2: computation of the upper left 2x2 submatrix of the
C zeroth component of the reflection matrix and the
C full 4x4 higher-order Fourier components. This
C option can be used when the incident light is
C unpolarized or if only the first two Stokes
C parameters of the incident light are nonzero.
C = 3: computation of the upper left 2x2 submatrix of the
C zeroth component of the reflection matrix and
C the upper left 3x3 submatrices of the higher-order
C Fourier components. This is the so-called 3x3
C approximation useful in cases when the incident
C light is unpolarized and the 4th Stokes parameter
C of the reflected light is expected to be
C negligibly small.
C = 4: computation of the full 4x4 Fourier components
C of the reflection matrix
C
C EP = absolute accuracy with which the Fourier components
C of the reflection matrix must be computed
PARAMETER (L1M=10000, NG=49, NPH=NG*4, NQUADR=2,
* NCASE=4, EP=1D-7)