shellig.f90 Source File


This file depends on

sourcefile~~shellig.f90~~EfferentGraph sourcefile~shellig.f90 shellig.f90 sourcefile~radbelt_kinds_module.f90 radbelt_kinds_module.F90 sourcefile~shellig.f90->sourcefile~radbelt_kinds_module.f90

Files dependent on this one

sourcefile~~shellig.f90~~AfferentGraph sourcefile~shellig.f90 shellig.f90 sourcefile~radbelt_module.f90 radbelt_module.f90 sourcefile~radbelt_module.f90->sourcefile~shellig.f90 sourcefile~radbelt_c_module.f90 radbelt_c_module.f90 sourcefile~radbelt_c_module.f90->sourcefile~radbelt_module.f90

Source Code

!*****************************************************************************************
!>
!  IGRF model
!
!### History
!  * SHELLIG.FOR, Version 2.0, January 1992
!  * 11/01/91-DKB- SHELLG: lowest starting point for B0 search is 2
!  * 1/27/92-DKB- Adopted to IGRF-91 coefficients model
!  * 2/05/92-DKB- Reduce variable-names: INTER(P)SHC,EXTRA(P)SHC,INITI(ALI)ZE
!  * 8/08/95-DKB- Updated to IGRF-45-95; new coeff. DGRF90, IGRF95, IGRF95S
!  * 5/31/00-DKB- Updated to IGRF-45-00; new coeff.: IGRF00, IGRF00s
!  * 3/24/05-DKB- Updated to IGRF-45-10; new coeff.: IGRF05, IGRF05s

module shellig_module

    use radbelt_kinds_module

    implicit none

    private

    integer, parameter :: filename_len = 14 !! length of the model data file names

    ! parameters formerly in `gener` common block
    real(wp), parameter :: Era = 6371.2_wp !! earth radius for normalization of cartesian coordinates (6371.2 km)
    real(wp), parameter :: erequ = 6378.16_wp
    real(wp), parameter :: erpol = 6356.775_wp
    real(wp), parameter :: Aquad = erequ * erequ !! square of major half axis for
                                                !! earth ellipsoid as recommended by international
                                                !! astronomical union
    real(wp), parameter :: Bquad = erpol * erpol !! square of minor half axis for
                                                !! earth ellipsoid as recommended by international
                                                !! astronomical union
    real(wp), parameter :: Umr = atan(1.0_wp) * 4.0_wp / 180.0_wp !! atan(1.0)*4./180.   <degree>*umr=<radiant>

    real(wp), dimension(3, 3), parameter :: u = reshape([+0.3511737_wp, -0.9148385_wp, -0.1993679_wp, &
                                                         +0.9335804_wp, +0.3583680_wp, +0.0000000_wp, &
                                                         +0.0714471_wp, -0.1861260_wp, +0.9799247_wp], [3, 3])
    integer, parameter :: max_loop_index = 3333 !! used in [[shellg]] for the field line integration loop

    type, public :: shellig_type
        private

        character(len=:), allocatable :: igrf_dir !! directory containing the data files

        ! formerly in the `fidb0` common block
        real(wp), dimension(3) :: sp = 0.0_wp

        ! formerly in blank common
        real(wp), dimension(3) :: xi = 0.0_wp
        real(wp), dimension(144) :: h = 0.0_wp !! Field model coefficients adjusted for [[shellg]]

        ! formerly in `model` common block
        integer :: iyea = 0 !! the int year corresponding to the file `name` that has been read
        character(len=:), allocatable :: name !! file name
        integer :: nmax = 0 !! maximum order of spherical harmonics
        real(wp) :: Time = 0.0_wp !! year (decimal: 1973.5) for which magnetic field is to be calculated
        real(wp), dimension(144) :: g = 0.0_wp !! `g(m)` -- normalized field coefficients (see [[feldcof]]) m=nmax*(nmax+2)
        integer :: nmax1 = 0 !! saved variables from the file
        integer :: nmax2 = 0 !! saved variables from the file
        real(wp), dimension(144) :: g_cache = 0.0_wp !! saved `g` from the file

        ! formerly saved vars in shellg:
        real(wp) :: step = 0.20_wp !! step size for field line tracing
        real(wp) :: steq = 0.03_wp !! step size for integration

        ! from feldcof, so we can cache the coefficients
        real(wp), dimension(120) :: gh2 = 0.0_wp ! JW : why is this 120 and g is 144 ???

        real(wp), dimension(:, :), allocatable :: p !! this was `p(8,100)` in the original code.
                                                  !! used for the field line integration loop.
                                                  !! changed it to be allocatable since it was
                                                  !! changed to be p(8,3334).

    contains
        private

        procedure, public :: igrf, igrfc

        procedure, public :: feldcof
        procedure, public :: feldg, feldc
        procedure, public :: shellg, shellc
        procedure, public :: findb0
        procedure :: stoer, feldi
        procedure, public :: set_data_file_dir, get_data_file_dir
        procedure, public :: destroy => destroy_shellig_type

    end type shellig_type

contains
!*****************************************************************************************

!*****************************************************************************************
!>
!  Destroy a [[shellig_type]].

    subroutine destroy_shellig_type(me)
        class(shellig_type), intent(out) :: me
    end subroutine destroy_shellig_type

!*****************************************************************************************
!>
!  Set the directory containing the data files.

    subroutine set_data_file_dir(me, dir)
        class(shellig_type), intent(inout) :: me
        character(len=*), intent(in) :: dir
        me%igrf_dir = trim(dir)
    end subroutine set_data_file_dir

!*****************************************************************************************
!>
!  Get the directory containing the data files.

    function get_data_file_dir(me) result(dir)
        class(shellig_type), intent(in) :: me
        character(len=:), allocatable :: dir
        if (allocated(me%igrf_dir)) then
            dir = trim(me%igrf_dir)//'/'
        else
            dir = 'data/igrf/' ! default
        end if
    end function get_data_file_dir

!*****************************************************************************************
!>
!  Wrapper for IGRF functions.

    subroutine igrf(me, lon, lat, height, year, xl, bbx)

        class(shellig_type), intent(inout) :: me
        real(wp), intent(in) :: lon !! geodetic longitude in degrees (east)
        real(wp), intent(in) :: lat !! geodetic latitude in degrees (north)
        real(wp), intent(in) :: height !! altitude in km above sea level
        real(wp), intent(in) :: year !! decimal year for which geomagnetic field is to
                                  !! be calculated (e.g.:1995.5 for day 185 of 1995)
        real(wp), intent(out) :: xl !! l-value
        real(wp), intent(out) :: bbx !! b_total / b_equatorial ratio

        real(wp) :: bab1, babs, bdel, bdown, beast, &
                    beq, bequ, bnorth, dimo, rr0
        integer :: icode
        logical :: val

        real(wp), parameter :: stps = 0.05_wp

        ! JW : do we need to reset some or all of these ?
        me%sp = 0.0_wp
        me%xi = 0.0_wp
        me%h = 0.0_wp
        me%step = 0.20_wp
        me%steq = 0.03_wp

        call me%feldcof(year, dimo)
        call me%feldg(lat, lon, height, bnorth, beast, bdown, babs)
        call me%shellg(lat, lon, height, dimo, xl, icode, bab1)

        bequ = dimo / (xl * xl * xl)
        if (icode == 1) then
            bdel = 1.0e-3_wp
            call me%findb0(stps, bdel, val, beq, rr0)
            if (val) bequ = beq
        end if
        bbx = babs / bequ

    end subroutine igrf
!*****************************************************************************************

!*****************************************************************************************
!>
!  Alternate version of [[igrf]] for cartesian coordinates.

    subroutine igrfc(me, v, year, xl, bbx)

        class(shellig_type), intent(inout) :: me
        real(wp), dimension(3), intent(in) :: v !! cartesian coordinates in earth radii (6371.2 km)
                                             !! x-axis pointing to equator at 0 longitude
                                             !! y-axis pointing to equator at 90 long.
                                             !! z-axis pointing to north pole
        real(wp), intent(in) :: year !! decimal year for which geomagnetic field is to
                                  !! be calculated (e.g.:1995.5 for day 185 of 1995)
        real(wp), intent(out) :: xl !! l-value
        real(wp), intent(out) :: bbx !! b_total / b_equatorial ratio

        real(wp) :: bab1, bdel, beq, bequ, dimo, rr0
        integer :: icode
        logical :: val
        real(wp), dimension(3) :: b

        real(wp), parameter :: stps = 0.05_wp

        ! JW : do we need to reset some or all of these ?
        me%sp = 0.0_wp
        me%xi = 0.0_wp
        me%h = 0.0_wp
        me%step = 0.20_wp
        me%steq = 0.03_wp

        call me%feldcof(year, dimo)
        call me%feldc(v, b)
        call me%shellc(v, dimo, xl, icode, bab1)

        bequ = dimo / (xl * xl * xl)
        if (icode == 1) then
            bdel = 1.0e-3_wp
            call me%findb0(stps, bdel, val, beq, rr0)
            if (val) bequ = beq
        end if
        bbx = norm2(b) / bequ

    end subroutine igrfc
!*****************************************************************************************

!*****************************************************************************************
!>
    subroutine findb0(me, stps, bdel, value, bequ, rr0)

        class(shellig_type), intent(inout) :: me
        real(wp), intent(in) :: stps
        real(wp), intent(inout) :: bdel
        real(wp), intent(out) :: bequ
        logical, intent(out) :: value
        real(wp), intent(out) :: rr0

        real(wp) :: b, bdelta, bmin, bold, bq1, &
                    bq2, bq3, p(8, 4), r1, r2, r3, &
                    rold, step, step12, zz
        integer :: i, irun, j, n

        step = stps
        irun = 0
        rold = 0.0_wp ! to avoid -Wmaybe-uninitialized warnings

        main: do
            irun = irun + 1
            if (irun > 5) then
                value = .false.
                exit main
            end if
            ! first three points
            p(1, 2) = me%sp(1)
            p(2, 2) = me%sp(2)
            p(3, 2) = me%sp(3)
            step = -sign(step, p(3, 2))
            call me%stoer(p(1, 2), bq2, r2)
            p(1, 3) = p(1, 2) + 0.5_wp * step * p(4, 2)
            p(2, 3) = p(2, 2) + 0.5_wp * step * p(5, 2)
            p(3, 3) = p(3, 2) + 0.5_wp * step
            call me%stoer(p(1, 3), bq3, r3)
            p(1, 1) = p(1, 2) - step * (2.0_wp * p(4, 2) - p(4, 3))
            p(2, 1) = p(2, 2) - step * (2.0_wp * p(5, 2) - p(5, 3))
            p(3, 1) = p(3, 2) - step
            call me%stoer(p(1, 1), bq1, r1)
            p(1, 3) = p(1, 2) + step * (20.0_wp * p(4, 3) - 3.*p(4, 2) + p(4, 1)) / 18.0_wp
            p(2, 3) = p(2, 2) + step * (20.0_wp * p(5, 3) - 3.*p(5, 2) + p(5, 1)) / 18.0_wp
            p(3, 3) = p(3, 2) + step
            call me%stoer(p(1, 3), bq3, r3)
            ! invert sense if required
            if (bq3 > bq1) then
                step = -step
                r3 = r1
                bq3 = bq1
                do i = 1, 5
                    zz = p(i, 1)
                    p(i, 1) = p(i, 3)
                    p(i, 3) = zz
                end do
            end if
            ! initialization
            step12 = step / 12.0_wp
            value = .true.
            bmin = 1.0e4_wp
            bold = 1.0e4_wp
            ! corrector (field line tracing)
            n = 0
            corrector: do
                p(1, 3) = p(1, 2) + step12 * (5.0_wp * p(4, 3) + 8.0_wp * p(4, 2) - p(4, 1))
                n = n + 1
                p(2, 3) = p(2, 2) + step12 * (5.0_wp * p(5, 3) + 8.0_wp * p(5, 2) - p(5, 1))
                ! predictor (field line tracing)
                p(1, 4) = p(1, 3) + step12 * (23.0_wp * p(4, 3) - 16.0_wp * p(4, 2) + 5.0_wp * p(4, 1))
                p(2, 4) = p(2, 3) + step12 * (23.0_wp * p(5, 3) - 16.0_wp * p(5, 2) + 5.0_wp * p(5, 1))
                p(3, 4) = p(3, 3) + step
                call me%stoer(p(1, 4), bq3, r3)
                do j = 1, 3
                    do i = 1, 8
                        p(i, j) = p(i, j + 1)
                    end do
                end do
                b = sqrt(bq3)
                if (b < bmin) bmin = b
                if (b > bold) exit corrector
                bold = b
                rold = 1.0_wp / r3
                me%sp(1) = p(1, 4)
                me%sp(2) = p(2, 4)
                me%sp(3) = p(3, 4)
            end do corrector
            if (bold /= bmin) value = .false.
            bdelta = (b - bold) / bold
            if (bdelta <= bdel) exit main
            step = step / 10.0_wp
        end do main

        rr0 = rold
        bequ = bold
        bdel = bdelta

    end subroutine findb0

!*****************************************************************************************
!>
!  Wrapper to [[shellg]] to be used with cartesian coordinates.
!
!@note In the original code, this was an ENTRY point in [[shellg]] and didn't
!      include all the outputs.

    subroutine shellc(me, v, dimo, fl, icode, b0)

        class(shellig_type), intent(inout) :: me
        real(wp), dimension(3), intent(in) :: v !! cartesian coordinates in earth radii (6371.2 km)
                                         !! * x-axis pointing to equator at 0 longitude
                                         !! * y-axis pointing to equator at 90 long.
                                         !! * z-axis pointing to north pole
        real(wp), intent(in) :: dimo !! dipol moment in gauss (normalized to earth radius)
        real(wp), intent(out) :: fl !! l-value
        integer, intent(out) :: icode !! * =1 normal completion
                                 !! * =2 unphysical conjugate point (fl meaningless)
                                 !! * =3 shell parameter greater than limit up to
                                 !!   which accurate calculation is required;
                                 !!   approximation is used.
        real(wp), intent(out) :: b0 !! magnetic field strength in gauss
        real(wp) :: glat, glon, alt !! not used

        call me%shellg(glat, glon, alt, dimo, fl, icode, b0, v)

    end subroutine shellc

!*****************************************************************************************
!>
!  calculates l-value for specified geodaetic coordinates, altitude
!  and gemagnetic field model.
!
!### Reference
!  * G. KLUGE, EUROPEAN SPACE OPERATIONS CENTER, INTERNAL NOTE
!    NO. 67, 1970.
!  * G. KLUGE, COMPUTER PHYSICS COMMUNICATIONS 3, 31-35, 1972
!
!### History
! * CHANGES (D. BILITZA, NOV 87):
!   - USING CORRECT DIPOL MOMENT I.E.,DIFFERENT COMMON/MODEL/
!   - USING IGRF EARTH MAGNETIC FIELD MODELS FROM 1945 TO 1990

    subroutine shellg(me, glat, glon, alt, dimo, fl, icode, b0, v)

        class(shellig_type), intent(inout) :: me
        real(wp), intent(in) :: glat !! geodetic latitude in degrees (north)
        real(wp), intent(in) :: glon !! geodetic longitude in degrees (east)
        real(wp), intent(in) :: alt !! altitude in km above sea level
        real(wp), intent(in) :: dimo !! dipol moment in gauss (normalized to earth radius)
        real(wp), intent(out) :: fl !! l-value
        integer, intent(out) :: icode !! * =1 normal completion
                                 !! * =2 unphysical conjugate point (fl meaningless)
                                 !! * =3 shell parameter greater than limit up to
                                 !!   which accurate calculation is required;
                                 !!   approximation is used.
        real(wp), intent(out) :: b0 !! magnetic field strength in gauss
        real(wp), dimension(3), intent(in), optional :: v !! cartesian coordinates in earth radii (6371.2 km)
                                                  !!
                                                  !! * x-axis pointing to equator at 0 longitude
                                                  !! * y-axis pointing to equator at 90 long.
                                                  !! * z-axis pointing to north pole
                                                  !!
                                                  !! If this argument is present, it is used
                                                  !! instead of glat,glon,alt. See [[shellc]].

        real(wp) :: arg1, arg2, bequ, bq1, bq2, bq3, c0, c1, c2, c3, &
                    d0, d1, d2, dimob0, e0, e1, e2, ff, fi, gg, &
                    hli, oradik, oterm, r, r1, r2, r3, r3h, radik, &
                    rq, step12, step2, stp, t, term, xx, z, zq, zz
        integer :: i, iequ, n

        real(wp), parameter :: rmin = 0.05_wp !! boundaries for identification of `icode=2 and 3`
        real(wp), parameter :: rmax = 1.01_wp !! boundaries for identification of `icode=2 and 3`

        if (.not. allocated(me%p)) allocate (me%p(8, max_loop_index + 1)) ! because `p(:,n+1)` in the loop

        bequ = 1.0e10_wp

        if (present(v)) then
            me%xi(1) = v(1)
            me%xi(2) = v(2)
            me%xi(3) = v(3)
        else
            me%xi = geo_to_cart(glat, glon, alt)
        end if

        associate (p => me%p)

            radik = 0.0_wp ! to avoid -Wmaybe-uninitialized warnings

            ! convert to dipol-oriented co-ordinates
            rq = 1.0_wp / (me%xi(1) * me%xi(1) + me%xi(2) * me%xi(2) + me%xi(3) * me%xi(3))
            r3h = sqrt(rq * sqrt(rq))
            p(1, 2) = (me%xi(1) * u(1, 1) + me%xi(2) * u(2, 1) + me%xi(3) * u(3, 1)) * r3h
            p(2, 2) = (me%xi(1) * u(1, 2) + me%xi(2) * u(2, 2)) * r3h
            p(3, 2) = (me%xi(1) * u(1, 3) + me%xi(2) * u(2, 3) + me%xi(3) * u(3, 3)) * rq
            ! first three points of field line
            me%step = -sign(me%step, p(3, 2))
            call me%stoer(p(1, 2), bq2, r2)
            b0 = sqrt(bq2)
            p(1, 3) = p(1, 2) + 0.5_wp * me%step * p(4, 2)
            p(2, 3) = p(2, 2) + 0.5_wp * me%step * p(5, 2)
            p(3, 3) = p(3, 2) + 0.5_wp * me%step
            call me%stoer(p(1, 3), bq3, r3)
            p(1, 1) = p(1, 2) - me%step * (2.0_wp * p(4, 2) - p(4, 3))
            p(2, 1) = p(2, 2) - me%step * (2.0_wp * p(5, 2) - p(5, 3))
            p(3, 1) = p(3, 2) - me%step
            call me%stoer(p(1, 1), bq1, r1)
            p(1, 3) = p(1, 2) + me%step * (20.0_wp * p(4, 3) - 3.*p(4, 2) + p(4, 1)) / 18.0_wp
            p(2, 3) = p(2, 2) + me%step * (20.0_wp * p(5, 3) - 3.*p(5, 2) + p(5, 1)) / 18.0_wp
            p(3, 3) = p(3, 2) + me%step
            call me%stoer(p(1, 3), bq3, r3)
            ! invert sense if required
            if (bq3 > bq1) then
                me%step = -me%step
                r3 = r1
                bq3 = bq1
                do i = 1, 7
                    zz = p(i, 1)
                    p(i, 1) = p(i, 3)
                    p(i, 3) = zz
                end do
            end if
            ! search for lowest magnetic field strength
            if (bq1 < bequ) then
                bequ = bq1
                iequ = 1
            end if
            if (bq2 < bequ) then
                bequ = bq2
                iequ = 2
            end if
            if (bq3 < bequ) then
                bequ = bq3
                iequ = 3
            end if
            ! initialization of integration loops
            step12 = me%step / 12.0_wp
            step2 = me%step + me%step
            me%steq = sign(me%steq, me%step)
            fi = 0.0_wp
            icode = 1
            oradik = 0.0_wp
            oterm = 0.0_wp
            stp = r2 * me%steq
            z = p(3, 2) + stp
            stp = stp / 0.75_wp
            p(8, 1) = step2 * (p(1, 1) * p(4, 1) + p(2, 1) * p(5, 1))
            p(8, 2) = step2 * (p(1, 2) * p(4, 2) + p(2, 2) * p(5, 2))
            ! main loop (field line tracing)
            main: do n = 3, max_loop_index
                ! corrector (field line tracing)
                p(1, n) = p(1, n - 1) + step12 * (5.0_wp * p(4, n) + 8.0_wp * p(4, n - 1) - p(4, n - 2))
                p(2, n) = p(2, n - 1) + step12 * (5.0_wp * p(5, n) + 8.0_wp * p(5, n - 1) - p(5, n - 2))
                ! prepare expansion coefficients for interpolation
                ! of slowly varying quantities
                p(8, n) = step2 * (p(1, n) * p(4, n) + p(2, n) * p(5, n))
                c0 = p(1, n - 1)**2 + p(2, n - 1)**2
                c1 = p(8, n - 1)
                c2 = (p(8, n) - p(8, n - 2)) * 0.25_wp
                c3 = (p(8, n) + p(8, n - 2) - c1 - c1) / 6.0_wp
                d0 = p(6, n - 1)
                d1 = (p(6, n) - p(6, n - 2)) * 0.5_wp
                d2 = (p(6, n) + p(6, n - 2) - d0 - d0) * 0.5_wp
                e0 = p(7, n - 1)
                e1 = (p(7, n) - p(7, n - 2)) * 0.5_wp
                e2 = (p(7, n) + p(7, n - 2) - e0 - e0) * 0.5_wp
                inner: do
                    ! inner loop (for quadrature)
                    t = (z - p(3, n - 1)) / me%step
                    if (t > 1.0_wp) then
                        ! predictor (field line tracing)
                        p(1, n + 1) = p(1, n) + step12 * (23.0_wp * p(4, n) - 16.0_wp * p(4, n - 1) + 5.0_wp * p(4, n - 2))
                        p(2, n + 1) = p(2, n) + step12 * (23.0_wp * p(5, n) - 16.0_wp * p(5, n - 1) + 5.0_wp * p(5, n - 2))
                        p(3, n + 1) = p(3, n) + me%step
                        call me%stoer(p(1, n + 1), bq3, r3)
                        ! search for lowest magnetic field strength
                        if (bq3 < bequ) then
                            iequ = n + 1
                            bequ = bq3
                        end if
                        exit inner
                    else
                        hli = 0.5_wp * (((c3 * t + c2) * t + c1) * t + c0)
                        zq = z * z
                        r = hli + sqrt(hli * hli + zq)
                        if (r <= rmin) then
                            ! approximation for high values of l.
                            icode = 3
                            t = -p(3, n - 1) / me%step
                            fl = 1.0_wp / (abs(((c3 * t + c2) * t + c1) * t + c0) + 1.0e-15_wp)
                            return
                        end if
                        rq = r * r
                        ff = sqrt(1.0_wp + 3.0_wp * zq / rq)
                        radik = b0 - ((d2 * t + d1) * t + d0) * r * rq * ff
                        if (r > rmax) then
                            icode = 2
                            radik = radik - 12.0_wp * (r - rmax)**2
                        end if
                        if (radik + radik <= oradik) exit main
                        term = sqrt(radik) * ff * ((e2 * t + e1) * t + e0) / (rq + zq)
                        fi = fi + stp * (oterm + term)
                        oradik = radik
                        oterm = term
                        stp = r * me%steq
                        z = z + stp
                    end if
                end do inner
            end do main
            if (iequ < 2) iequ = 2
            me%sp(1) = p(1, iequ - 1)
            me%sp(2) = p(2, iequ - 1)
            me%sp(3) = p(3, iequ - 1)
            if (oradik >= 1.0e-15_wp) fi = fi + stp / &
                                           0.75_wp * oterm * oradik / &
                                           (oradik - radik)

            ! the minimal allowable value of fi was changed from 1e-15 to 1e-12,
            ! because 1e-38 is the minimal allowable arg. for alog in our envir.
            ! d. bilitza, nov 87.
            fi = 0.5_wp * abs(fi) / sqrt(b0) + 1.0e-12_wp

            ! compute l from b and i.  same as carmel in invar.
            ! correct dipole moment is used here. d. bilitza, nov 87.
            dimob0 = dimo / b0
            arg1 = log(fi)
            arg2 = log(dimob0)
            ! arg = fi*fi*fi/dimob0
            ! if(abs(arg)>88.0_wp) arg=88.0_wp
            xx = 3 * arg1 - arg2
            if (xx > 23.0_wp) then
                gg = xx - 3.0460681_wp
            elseif (xx > 11.7_wp) then
                gg = (((((2.8212095e-8_wp * xx - 3.8049276e-6_wp) * xx + &
                         2.170224e-4_wp) * xx - 6.7310339e-3_wp) * xx + &
                       1.2038224e-1_wp) * xx - 1.8461796e-1_wp) * xx + 2.0007187_wp
            elseif (xx > +3.0_wp) then
                gg = ((((((((6.3271665e-10_wp * xx - 3.958306e-8_wp) * xx + &
                            9.9766148e-07_wp) * xx - 1.2531932e-5_wp) * xx + &
                          7.9451313e-5_wp) * xx - 3.2077032e-4_wp) * xx + &
                        2.1680398e-3_wp) * xx + 1.2817956e-2_wp) * xx + &
                      4.3510529e-1_wp) * xx + 6.222355e-1_wp
            elseif (xx > -3.0_wp) then
                gg = ((((((((2.6047023e-10_wp * xx + 2.3028767e-9_wp) * xx - &
                            2.1997983e-8_wp) * xx - 5.3977642e-7_wp) * xx - &
                          3.3408822e-6_wp) * xx + 3.8379917e-5_wp) * xx + &
                        1.1784234e-3_wp) * xx + 1.4492441e-2_wp) * xx + &
                      4.3352788e-1_wp) * xx + 6.228644e-1_wp
            elseif (xx > -22.0_wp) then
                gg = ((((((((-8.1537735e-14_wp * xx + 8.3232531e-13_wp) * xx + &
                            1.0066362e-9_wp) * xx + 8.1048663e-8_wp) * xx + &
                          3.2916354e-6_wp) * xx + 8.2711096e-5_wp) * xx + &
                        1.3714667e-3_wp) * xx + 1.5017245e-2_wp) * xx + &
                      4.3432642e-1_wp) * xx + 6.2337691e-1_wp
            else
                gg = 3.33338e-1_wp * xx + 3.0062102e-1_wp
            end if
            fl = exp(log((1.0_wp + exp(gg)) * dimob0) / 3.0_wp)

        end associate

    end subroutine shellg

!*****************************************************************************************
!>
!  subroutine used for field line tracing in [[shellg]].
!  calls entry point [[feldi]] in geomagnetic field subroutine [[feldg]]

    subroutine stoer(me, p, bq, r)

        class(shellig_type), intent(inout) :: me
        real(wp), dimension(7), intent(inout) :: p
        real(wp), intent(out) :: bq
        real(wp), intent(out) :: r

        real(wp) :: dr, dsq, dx, dxm, dy, dym, dz, &
                    dzm, fli, q, rq, wr, xm, ym, zm

        ! xm,ym,zm are geomagnetic cartesian inverse co-ordinates
        zm = P(3)
        fli = P(1) * P(1) + P(2) * P(2) + 1.0e-15_wp
        R = 0.5_wp * (fli + sqrt(fli * fli + (zm + zm)**2))
        rq = R * R
        wr = sqrt(R)
        xm = P(1) * wr
        ym = P(2) * wr
        ! transform to geographic co-ordinate system
        me%Xi(1) = xm * u(1, 1) + ym * u(1, 2) + zm * u(1, 3)
        me%Xi(2) = xm * u(2, 1) + ym * u(2, 2) + zm * u(2, 3)
        me%Xi(3) = xm * u(3, 1) + zm * u(3, 3)
        ! compute derivatives
        ! Changed from CALL FELDI(XI,H); XI, H are in COMMON block; results
        ! are the same; dkb Feb 1998.
        ! JW : feb 2024 : xi, h now class variables.
        call me%feldi()
        q = me%H(1) / rq
        dx = me%H(3) + me%H(3) + q * me%Xi(1)
        dy = me%H(4) + me%H(4) + q * me%Xi(2)
        dz = me%H(2) + me%H(2) + q * me%Xi(3)
        ! transform back to geomagnetic co-ordinate system
        dxm = u(1, 1) * dx + u(2, 1) * dy + u(3, 1) * dz
        dym = u(1, 2) * dx + u(2, 2) * dy
        dzm = u(1, 3) * dx + u(2, 3) * dy + u(3, 3) * dz
        dr = (xm * dxm + ym * dym + zm * dzm) / R
        ! form slowly varying expressions
        P(4) = (wr * dxm - 0.5_wp * P(1) * dr) / (R * dzm)
        P(5) = (wr * dym - 0.5_wp * P(2) * dr) / (R * dzm)
        dsq = rq * (dxm * dxm + dym * dym + dzm * dzm)
        Bq = dsq * rq * rq
        P(6) = sqrt(dsq / (rq + 3.0_wp * zm * zm))
        P(7) = P(6) * (rq + zm * zm) / (rq * dzm)

    end subroutine stoer

!*****************************************************************************************
!>
!  Calculates earth magnetic field from spherical harmonics model
!
!### Reference
! ref: g. kluge, european space operations centre, internal note 61,
!      1970.
!
!### History
!  * changes (d. bilitza, nov 87):
!   - field coefficients in binary data files instead of block data
!   - calculates dipol moment
!
!@note In the original code, [[feldc] and [[feldi]] were
!      ENTRY points to this routine

    subroutine feldg(me, glat, glon, alt, bnorth, beast, bdown, babs)

        class(shellig_type), intent(inout) :: me
        real(wp), intent(in) :: glat !! geodetic latitude in degrees (north)
        real(wp), intent(in) :: glon !! geodetic longitude in degrees (east)
        real(wp), intent(in) :: alt !! altitude in km above sea level
        real(wp), intent(out) :: bnorth, beast, bdown !! components of the field with respect
                                                !! to the local geodetic coordinate system, with axis
                                                !! pointing in the tangential plane to the north, east
                                                !! and downward.
        real(wp), intent(out) :: Babs !! magnetic field strength in gauss

        real(wp) :: brho, bxxx, byyy, bzzz, cp, ct, d, f, rho, &
                    rlat, rlon, rq, s, sp, st, t, &
                    x, xxx, y, yyy, z, zzz
        integer :: i, ih, ihmax, il, imax, k, last, m

        ! same calculation as geo_to_cart, but not used here
        ! because the intermediate variables are also used below.
        rlat = glat * umr
        ct = sin(rlat)
        st = cos(rlat)
        d = sqrt(aquad - (aquad - bquad) * ct * ct)
        rlon = glon * umr
        cp = cos(rlon)
        sp = sin(rlon)
        zzz = (alt + bquad / d) * ct / era
        rho = (alt + aquad / d) * st / era
        xxx = rho * cp
        yyy = rho * sp

        rq = 1.0_wp / (xxx * xxx + yyy * yyy + zzz * zzz)
        me%xi = [xxx, yyy, zzz] * rq

        ihmax = me%nmax * me%nmax + 1
        last = ihmax + me%nmax + me%nmax
        imax = me%nmax + me%nmax - 1
        do i = ihmax, last
            me%h(i) = me%g(i)
        end do
        do k = 1, 3, 2
            i = imax
            ih = ihmax
            do
                il = ih - i
                f = 2.0_wp / real(i - k + 2, wp)
                x = me%xi(1) * f
                y = me%xi(2) * f
                z = me%xi(3) * (f + f)
                i = i - 2
                if ((i - 1) >= 0) then
                    if ((i - 1) > 0) then
                        do m = 3, i, 2
                            me%h(il + m + 1) = me%g(il + m + 1) + z * me%h(ih + m + 1) + x * (me%h(ih + m + 3) - &
                                                                       me%h(ih + m - 1)) - y * (me%h(ih + m + 2) + me%h(ih + m - 2))
                            me%h(il + m) = me%g(il + m) + z * me%h(ih + m) + x * (me%h(ih + m + 2) - &
                                                                       me%h(ih + m - 2)) + y * (me%h(ih + m + 3) + me%h(ih + m - 1))
                        end do
                    end if
                    me%h(il + 2) = me%g(il + 2) + z * me%h(ih + 2) + x * me%h(ih + 4) - y * (me%h(ih + 3) + me%h(ih))
                    me%h(il + 1) = me%g(il + 1) + z * me%h(ih + 1) + y * me%h(ih + 4) + x * (me%h(ih + 3) - me%h(ih))
                end if
                me%h(il) = me%g(il) + z * me%h(ih) + 2.0_wp * (x * me%h(ih + 1) + y * me%h(ih + 2))
                ih = il
                if (i < k) exit
            end do
        end do

        s = 0.5_wp * me%h(1) + 2.0_wp * (me%h(2) * me%xi(3) + me%h(3) * me%xi(1) + me%h(4) * me%xi(2))
        t = (rq + rq) * sqrt(rq)
        bxxx = t * (me%h(3) - s * xxx)
        byyy = t * (me%h(4) - s * yyy)
        bzzz = t * (me%h(2) - s * zzz)

        babs = sqrt(bxxx * bxxx + byyy * byyy + bzzz * bzzz)
        beast = byyy * cp - bxxx * sp
        brho = byyy * sp + bxxx * cp
        bnorth = bzzz * st - brho * ct
        bdown = -bzzz * ct - brho * st

    end subroutine feldg

!*****************************************************************************************
!>
!  Alternate version of [[feldg]] to be used with cartesian coordinates

    subroutine feldc(me, v, b)

        class(shellig_type), intent(inout) :: me
        real(wp), dimension(3), intent(in) :: v !! cartesian coordinates in earth radii (6371.2 km)
                                          !! x-axis pointing to equator at 0 longitude
                                          !! y-axis pointing to equator at 90 long.
                                          !! z-axis pointing to north pole
        real(wp), intent(out) :: b(3) !! field components

        real(wp) :: f, rq, s, t, x, xxx, y, yyy, z, zzz
        integer :: i, ih, ihmax, il, imax, k, last, m

        xxx = v(1)
        yyy = v(2)
        zzz = v(3)

        rq = 1.0_wp / (xxx * xxx + yyy * yyy + zzz * zzz)
        me%xi = [xxx, yyy, zzz] * rq

        ihmax = me%nmax * me%nmax + 1
        last = ihmax + me%nmax + me%nmax
        imax = me%nmax + me%nmax - 1
        do i = ihmax, last
            me%h(i) = me%g(i)
        end do
        do k = 1, 3, 2
            i = imax
            ih = ihmax
            do
                il = ih - i
                f = 2.0_wp / real(i - k + 2, wp)
                x = me%xi(1) * f
                y = me%xi(2) * f
                z = me%xi(3) * (f + f)
                i = i - 2
                if ((i - 1) >= 0) then
                    if ((i - 1) > 0) then
                        do m = 3, i, 2
                            me%h(il + m + 1) = me%g(il + m + 1) + z * me%h(ih + m + 1) + x * (me%h(ih + m + 3) - &
                                                                       me%h(ih + m - 1)) - y * (me%h(ih + m + 2) + me%h(ih + m - 2))
                            me%h(il + m) = me%g(il + m) + z * me%h(ih + m) + x * (me%h(ih + m + 2) - &
                                                                       me%h(ih + m - 2)) + y * (me%h(ih + m + 3) + me%h(ih + m - 1))
                        end do
                    end if
                    me%h(il + 2) = me%g(il + 2) + z * me%h(ih + 2) + x * me%h(ih + 4) - y * (me%h(ih + 3) + me%h(ih))
                    me%h(il + 1) = me%g(il + 1) + z * me%h(ih + 1) + y * me%h(ih + 4) + x * (me%h(ih + 3) - me%h(ih))
                end if
                me%h(il) = me%g(il) + z * me%h(ih) + 2.0_wp * (x * me%h(ih + 1) + y * me%h(ih + 2))
                ih = il
                if (i < k) exit
            end do
        end do

        s = 0.5_wp * me%h(1) + 2.0_wp * (me%h(2) * me%xi(3) + me%h(3) * me%xi(1) + me%h(4) * me%xi(2))
        t = (rq + rq) * sqrt(rq)

        b(1) = t * (me%h(3) - s * xxx)
        b(2) = t * (me%h(4) - s * yyy)
        b(3) = t * (me%h(2) - s * zzz)

    end subroutine feldc

!*****************************************************************************************
!>
!  Used for `l` computation.

    subroutine feldi(me)

        class(shellig_type), intent(inout) :: me

        real(wp) :: f, x, y, z
        integer :: i, ih, ihmax, il, imax, k, last, m

        ihmax = me%nmax * me%nmax + 1
        last = ihmax + me%nmax + me%nmax
        imax = me%nmax + me%nmax - 1
        do i = ihmax, last
            me%h(i) = me%g(i)
        end do
        do k = 1, 3, 2
            i = imax
            ih = ihmax
            do
                il = ih - i
                f = 2.0_wp / real(i - k + 2, wp)
                x = me%xi(1) * f
                y = me%xi(2) * f
                z = me%xi(3) * (f + f)
                i = i - 2
                if ((i - 1) >= 0) then
                    if ((i - 1) > 0) then
                        do m = 3, i, 2
                            me%h(il + m + 1) = me%g(il + m + 1) + z * me%h(ih + m + 1) + x * (me%h(ih + m + 3) - &
                                                                       me%h(ih + m - 1)) - y * (me%h(ih + m + 2) + me%h(ih + m - 2))
                            me%h(il + m) = me%g(il + m) + z * me%h(ih + m) + x * (me%h(ih + m + 2) - &
                                                                       me%h(ih + m - 2)) + y * (me%h(ih + m + 3) + me%h(ih + m - 1))
                        end do
                    end if
                    me%h(il + 2) = me%g(il + 2) + z * me%h(ih + 2) + x * me%h(ih + 4) - y * (me%h(ih + 3) + me%h(ih))
                    me%h(il + 1) = me%g(il + 1) + z * me%h(ih + 1) + y * me%h(ih + 4) + x * (me%h(ih + 3) - me%h(ih))
                end if
                me%h(il) = me%g(il) + z * me%h(ih) + 2.0_wp * (x * me%h(ih + 1) + y * me%h(ih + 2))
                ih = il
                if (i < k) exit
            end do
        end do

    end subroutine feldi

!*****************************************************************************************
!>
!  Determines coefficients and dipol moment from IGRF models
!
!### Author
!  * D. BILITZA, NSSDC, GSFC, CODE 633, GREENBELT, MD 20771,
!    (301) 286-9536 NOV 1987.
!
!### History
!  * corrected for 2000 update - dkb- 5/31/2000
!  * updated to IGRF-2000 version -dkb- 5/31/2000
!  * updated to IGRF-2005 version -dkb- 3/24/2000

    subroutine feldcof(me, year, dimo)

        class(shellig_type), intent(inout) :: me
        real(wp), intent(in) :: year !! decimal year for which geomagnetic field is to
                               !! be calculated (e.g.:1995.5 for day 185 of 1995)
        real(wp), intent(out) :: dimo !! geomagnetic dipol moment in gauss (normalized
                                !! to earth's radius) at the time (year)

        real(wp) :: dte1, dte2, erad, gha(144), sqrt2
        integer :: i, ier, j, l, m, n, iyea
        character(len=:), allocatable :: fil2
        real(wp) :: x, f0, f !! these were double precision in original
                          !! code while everything else was single precision

        ! changed to conform with IGRF 45-95, also FILMOD, DTEMOD arrays +1
        character(len=filename_len), dimension(17), parameter :: filmod = [ &
                                                               'dgrf1945.dat ', 'dgrf1950.dat ', 'dgrf1955.dat ', 'dgrf1960.dat ', &
                                                               'dgrf1965.dat ', 'dgrf1970.dat ', 'dgrf1975.dat ', 'dgrf1980.dat ', &
                                                               'dgrf1985.dat ', 'dgrf1990.dat ', 'dgrf1995.dat ', 'dgrf2000.dat ', &
                                                               'dgrf2005.dat ', 'dgrf2010.dat ', 'dgrf2015.dat ', 'igrf2020.dat ', &
                                                                 'igrf2020s.dat']
        real(wp), dimension(17), parameter :: dtemod = [1945.0_wp, 1950.0_wp, 1955.0_wp, &
                                                        1960.0_wp, 1965.0_wp, 1970.0_wp, &
                                                        1975.0_wp, 1980.0_wp, 1985.0_wp, &
                                                        1990.0_wp, 1995.0_wp, 2000.0_wp, &
                                                        2005.0_wp, 2010.0_wp, 2015.0_wp, &
                                                        2020.0_wp, 2025.0_wp]
        integer, parameter :: numye = size(dtemod) - 1 ! number of 5-year priods represented by IGRF
        integer, parameter :: is = 0 !! * is=0 for schmidt normalization
                               !! * is=1 gauss normalization

        logical :: read_file

        !-- determine igrf-years for input-year
        me%time = year
        iyea = int(year / 5.0_wp) * 5
        read_file = iyea /= me%iyea ! if we have to read the file
        me%iyea = iyea
        l = (me%iyea - 1945) / 5 + 1
        if (l < 1) l = 1
        if (l > numye) l = numye
        dte1 = dtemod(l)
        me%name = me%get_data_file_dir()//trim(filmod(l))
        dte2 = dtemod(l + 1)
        fil2 = me%get_data_file_dir()//trim(filmod(l + 1))
        if (read_file) then
            ! get igrf coefficients for the boundary years
            ! [if they have not ready been loaded]
            call getshc(me%name, me%nmax1, erad, me%g, ier)
            if (ier /= 0) error stop 'error reading file: '//trim(me%name)
            me%g_cache = me%g ! because it is modified below, we have to cache the original values from the file
            call getshc(fil2, me%nmax2, erad, me%gh2, ier)
            if (ier /= 0) error stop 'error reading file: '//trim(fil2)
        else
            me%g = me%g_cache
        end if
        !-- determine igrf coefficients for year
        if (l <= numye - 1) then
            call intershc(year, dte1, me%nmax1, me%g, dte2, me%nmax2, me%gh2, me%nmax, gha)
        else
            call extrashc(year, dte1, me%nmax1, me%g, me%nmax2, me%gh2, me%nmax, gha)
        end if
        !-- determine magnetic dipol moment and coeffiecients g
        f0 = 0.0_wp
        do j = 1, 3
            f = gha(j) * 1.0e-5_wp
            f0 = f0 + f * f
        end do
        dimo = sqrt(f0)

        me%g(1) = 0.0_wp
        i = 2
        f0 = 1.0e-5_wp
        if (is == 0) f0 = -f0
        sqrt2 = sqrt(2.0_wp)

        do n = 1, me%nmax
            x = n
            f0 = f0 * x * x / (4.0_wp * x - 2.0_wp)
            if (is == 0) f0 = f0 * (2.0_wp * x - 1.0_wp) / x
            f = f0 * 0.5_wp
            if (is == 0) f = f * sqrt2
            me%g(i) = gha(i - 1) * f0
            i = i + 1
            do m = 1, n
                f = f * (x + m) / (x - m + 1.0_wp)
                if (is == 0) f = f * sqrt((x - m + 1.0_wp) / (x + m))
                me%g(i) = gha(i - 1) * f
                me%g(i + 1) = gha(i) * f
                i = i + 2
            end do
        end do

    end subroutine feldcof

!*****************************************************************************************
!>
!  Reads spherical harmonic coefficients from the specified
!  file into an array.
!
!### Author
!  * Version 1.01, A. Zunde, USGS, MS 964,
!    Box 25046 Federal Center, Denver, CO  80225

    subroutine getshc(Fspec, Nmax, Erad, Gh, Ier)

        character(len=*), intent(in) :: Fspec !! File specification
        integer, intent(out) :: Nmax !! Maximum degree and order of model
        real(wp), intent(out) :: Erad !! Earth's radius associated with the spherical
                                !! harmonic coefficients, in the same units as
                                !! elevation
        real(wp), dimension(*), intent(out) :: Gh !! Schmidt quasi-normal internal spherical
                                           !! harmonic coefficients
        integer, intent(out) :: Ier !! Error number:
                              !!
                              !!  * 0, no error
                              !!  * -2, records out of order
                              !!  * FORTRAN run-time error number

        integer :: iu !! logical unit number
        real(wp) :: g, h
        integer :: i, m, mm, n, nn

        read_file: block
            ! ---------------------------------------------------------------
            !  Open coefficient file. Read past first header record.
            !  Read degree and order of model and Earth's radius.
            ! ---------------------------------------------------------------
            open (newunit=Iu, FILE=Fspec, STATUS='OLD', IOSTAT=Ier)
            if (Ier /= 0) then
                write (*, *) 'Error opening file: '//trim(fspec)
                exit read_file
            end if
            read (Iu, *, IOSTAT=Ier)
            if (Ier /= 0) exit read_file
            read (Iu, *, IOSTAT=Ier) Nmax, Erad
            if (Ier /= 0) exit read_file

            ! ---------------------------------------------------------------
            !  Read the coefficient file, arranged as follows:
            !
            !          N     M     G     H
            !          ----------------------
            !            /   1     0    GH(1)  -
            !           /  1     1    GH(2) GH(3)
            !          /  2     0    GH(4)  -
            !         /  2     1    GH(5) GH(6)
            !      NMAX*(NMAX+3)/2   /  2     2    GH(7) GH(8)
            !         records    \  3     0    GH(9)  -
            !         \      .     .     .     .
            !          \  .     .     .     .
            !      NMAX*(NMAX+2)     \  .     .     .     .
            !      elements in GH      \  NMAX  NMAX   .     .
            !
            !  N and M are, respectively, the degree and order of the
            !  coefficient.
            ! ---------------------------------------------------------------
            i = 0
            main: do nn = 1, Nmax
                do mm = 0, nn
                    read (Iu, *, IOSTAT=Ier) n, m, g, h
                    if (Ier /= 0) exit main
                    if (nn /= n .or. mm /= m) then
                        Ier = -2
                        exit main
                    end if
                    i = i + 1
                    Gh(i) = g
                    if (m /= 0) then
                        i = i + 1
                        Gh(i) = h
                    end if
                end do
            end do main

        end block read_file

        close (Iu)

    end subroutine getshc

!*****************************************************************************************
!>
!  Interpolates linearly, in time, between two spherical
!  harmonic models.
!
!  The coefficients (GH) of the resulting model, at date
!  DATE, are computed by linearly interpolating between the
!  coefficients of the earlier model (GH1), at date DTE1,
!  and those of the later model (GH2), at date DTE2. If one
!  model is smaller than the other, the interpolation is
!  performed with the missing coefficients assumed to be 0.
!
!### Author
!  * Version 1.01, A. Zunde
!    USGS, MS 964, Box 25046 Federal Center, Denver, CO  80225

    subroutine intershc(date, dte1, nmax1, gh1, dte2, nmax2, gh2, nmax, gh)

        real(wp), intent(in) :: date !! Date of resulting model (in decimal year)
        real(wp), intent(in) :: dte1 !! Date of earlier model
        integer, intent(in) :: nmax1 !! Maximum degree and order of earlier model
        real(wp), intent(in) :: gh1(*) !! Schmidt quasi-normal internal spherical harmonic coefficients of earlier model
        real(wp), intent(in) :: dte2 !! Date of later model
        integer, intent(in) :: nmax2 !! Maximum degree and order of later model
        real(wp), intent(in) :: gh2(*) !! Schmidt quasi-normal internal spherical harmonic coefficients of later model
        real(wp), intent(out) :: gh(*) !! Coefficients of resulting model
        integer, intent(out) :: nmax !! Maximum degree and order of resulting model

        real(wp) :: factor
        integer :: i, k, l

        factor = (date - dte1) / (dte2 - dte1)

        if (nmax1 == nmax2) then
            k = nmax1 * (nmax1 + 2)
            nmax = nmax1
        elseif (nmax1 > nmax2) then
            k = nmax2 * (nmax2 + 2)
            l = nmax1 * (nmax1 + 2)
            do i = k + 1, l
                gh(i) = gh1(i) + factor * (-gh1(i))
            end do
            nmax = nmax1
        else
            k = nmax1 * (nmax1 + 2)
            l = nmax2 * (nmax2 + 2)
            do i = k + 1, l
                gh(i) = factor * gh2(i)
            end do
            nmax = nmax2
        end if

        do i = 1, k
            gh(i) = gh1(i) + factor * (gh2(i) - gh1(i))
        end do

    end subroutine intershc

!*****************************************************************************************
!>
!  Extrapolates linearly a spherical harmonic model with a
!  rate-of-change model.
!
!  The coefficients (GH) of the resulting model, at date
!  DATE, are computed by linearly extrapolating the coef-
!  ficients of the base model (GH1), at date DTE1, using
!  those of the rate-of-change model (GH2), at date DTE2. If
!  one model is smaller than the other, the extrapolation is
!  performed with the missing coefficients assumed to be 0.
!
!### Author
!  * Version 1.01, A. Zunde
!    USGS, MS 964, Box 25046 Federal Center, Denver, CO  80225

    subroutine extrashc(date, dte1, nmax1, gh1, nmax2, gh2, nmax, gh)

        real(wp), intent(in) :: date !! Date of resulting model (in decimal year)
        real(wp), intent(in) :: dte1 !! Date of base model
        integer, intent(in) :: nmax1 !! Maximum degree and order of base model
        real(wp), intent(in) :: gh1(*) !! Schmidt quasi-normal internal spherical harmonic coefficients of base model
        integer, intent(in) :: nmax2 !! Maximum degree and order of rate-of-change model
        real(wp), intent(in) :: gh2(*) !! Schmidt quasi-normal internal spherical harmonic coefficients of rate-of-change model
        real(wp), intent(out) :: gh(*) !! Coefficients of resulting model
        integer, intent(out) :: nmax !! Maximum degree and order of resulting model

        real(wp) :: factor
        integer :: i, k, l

        factor = (date - dte1)

        if (nmax1 == nmax2) then
            k = nmax1 * (nmax1 + 2)
            nmax = nmax1
        elseif (nmax1 > nmax2) then
            k = nmax2 * (nmax2 + 2)
            l = nmax1 * (nmax1 + 2)
            do i = k + 1, l
                gh(i) = gh1(i)
            end do
            nmax = nmax1
        else
            k = nmax1 * (nmax1 + 2)
            l = nmax2 * (nmax2 + 2)
            do i = k + 1, l
                gh(i) = factor * gh2(i)
            end do
            nmax = nmax2
        end if

        do i = 1, k
            gh(i) = gh1(i) + factor * gh2(i)
        end do

    end subroutine extrashc

!*****************************************************************************************
!>
!  geodetic to scaled cartesian coordinates

    pure function geo_to_cart(glat, glon, alt) result(x)

        real(wp), intent(in) :: glat !! geodetic latitude in degrees (north)
        real(wp), intent(in) :: glon !! geodetic longitude in degrees (east)
        real(wp), intent(in) :: alt !! altitude in km above sea level
        real(wp), dimension(3) :: x !! cartesian coordinates in earth radii (6371.2 km)
                                !!
                                !! * x-axis pointing to equator at 0 longitude
                                !! * y-axis pointing to equator at 90 long.
                                !! * z-axis pointing to north pole

        real(wp) :: rlat !! latitude in radians
        real(wp) :: rlon !! longitude in radians
        real(wp) :: d, rho

        ! deg to radians:
        rlat = glat * umr
        rlon = glon * umr

        ! JW : it's weird that ct is sin, and st is cos...it was like that in the original code
        associate (ct => sin(rlat), st => cos(rlat), cp => cos(rlon), sp => sin(rlon))
            d = sqrt(aquad - (aquad - bquad) * ct * ct)
            rho = (alt + aquad / d) * st / era
            x = [rho * cp, rho * sp, (alt + bquad / d) * ct / era]
        end associate

    end function geo_to_cart

end module SHELLIG_module