sptgpsd.f File Reference

Transform spectral to polar stereographic gradients. More...

Go to the source code of this file.

Functions/Subroutines
subroutine	sptgpsd (IROMB, MAXWV, KMAX, NPS, KWSKIP, KGSKIP, NISKIP, NJSKIP, TRUE, XMESH, ORIENT, WAVE, XN, YN, XS, YS)
	This subprogram performs a spherical transform from spectral coefficients of scalar fields to gradient fields on a pair of polar stereographic grids. More...

Detailed Description

Transform spectral to polar stereographic gradients.

Program history log:

Date	Programmer	Comments
96-02-29	IREDELL	Initial.
1998-12-15	IREDELL	OpenMP directives inserted.

Author

IREDELL

Date

96-02-29

Definition in file sptgpsd.f.

Function/Subroutine Documentation

sptgpsd()

subroutine sptgpsd	(		IROMB,
			MAXWV,
			KMAX,
			NPS,
			KWSKIP,
			KGSKIP,
			NISKIP,
			NJSKIP,
			TRUE,
			XMESH,
			ORIENT,
		real, dimension(*)	WAVE,
		real, dimension(*)	XN,
		real, dimension(*)	YN,
		real, dimension(*)	XS,
		real, dimension(*)	YS
	)

This subprogram performs a spherical transform from spectral coefficients of scalar fields to gradient fields on a pair of polar stereographic grids.

The wave-space can be either triangular or rhomboidal. The wave and grid fields may have general indexing, but each wave field is in sequential 'ibm order', i.e., with zonal wavenumber as the slower index. The two square polar stereographic grids are centered on the respective poles, with the orientation longitude of the southern hemisphere grid 180 degrees opposite that of the northern hemisphere grid. The vectors are automatically rotated to be resolved relative to the respective polar stereographic grids.

The transform is made efficient by combining points in eight sectors of each polar stereographic grid, numbered as in the following diagram. The pole and the sector boundaries are treated specially in the code. Unfortunately, this approach induces some hairy indexing and code loquacity, for which the developer apologizes.

   \ 4 | 5 /
    \  |  /
   3 \ | / 6
      \|/
   ----+----
      /|\
   2 / | \ 7
    /  |  \
   / 1 | 8 \

The transforms are all multiprocessed over sector points. transform several fields at a time to improve vectorization. Subprogram can be called from a multiprocessing environment.

Parameters

IROMB	Spectral domain shape (0 for triangular, 1 for rhomboidal)
MAXWV	Spectral truncation
KMAX	Number of fields to transform
NPS	Odd order of the polar stereographic grids
KWSKIP	Skip number between wave fields (defaults to (MAXWV+1)*((IROMB+1)*MAXWV+2) if KWSKIP=0)
KGSKIP	Skip number between grid fields (defaults to NPS*NPS if KGSKIP=0)
NISKIP	Skip number between grid i-points (defaults to 1 if NISKIP=0)
NJSKIP	Skip number between grid j-points (defaults to NPS if NJSKIP=0)
TRUE	Latitude at which PS grid is true (usually 60.)
XMESH	Grid length at true latitude (M)
ORIENT	Longitude at bottom of northern PS grid (southern PS grid will have opposite orientation.)
WAVE	Wave fields
XN	Northern polar stereographic x-gradients
YN	Northern polar stereographic y-gradients
XS	Southern polar stereographic x-gradients
YS	Southern polar stereographic y-gradients

Author

IREDELL

Date

96-02-29

Definition at line 71 of file sptgpsd.f.

References splaplac(), sptgpsv(), and spwget().