| 1 |
C $Header: /u/gcmpack/MITgcm/pkg/exf/exf_interp.F,v 1.24 2010/04/07 20:52:30 jmc Exp $ |
| 2 |
C $Name: $ |
| 3 |
|
| 4 |
#include "EXF_OPTIONS.h" |
| 5 |
|
| 6 |
CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC |
| 7 |
C Flux Coupler using C |
| 8 |
C Bilinear interpolation of forcing fields C |
| 9 |
C C |
| 10 |
C B. Cheng (12/2002) C |
| 11 |
C C |
| 12 |
C added Bicubic (bnc 1/2003) C |
| 13 |
C C |
| 14 |
CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC |
| 15 |
|
| 16 |
_RL FUNCTION LAGRAN(i,x,a,sp) |
| 17 |
|
| 18 |
INTEGER i |
| 19 |
_RS x |
| 20 |
_RL a(4) |
| 21 |
INTEGER sp |
| 22 |
|
| 23 |
C- local variables: |
| 24 |
INTEGER k |
| 25 |
_RL numer,denom |
| 26 |
|
| 27 |
numer = 1. _d 0 |
| 28 |
denom = 1. _d 0 |
| 29 |
|
| 30 |
#ifdef TARGET_NEC_SX |
| 31 |
!CDIR UNROLL=8 |
| 32 |
#endif /* TARGET_NEC_SX */ |
| 33 |
do k=1,sp |
| 34 |
if ( k .ne. i) then |
| 35 |
denom = denom*(a(i) - a(k)) |
| 36 |
numer = numer*(x - a(k)) |
| 37 |
endif |
| 38 |
enddo |
| 39 |
|
| 40 |
lagran = numer/denom |
| 41 |
|
| 42 |
RETURN |
| 43 |
END |
| 44 |
|
| 45 |
C---+----1----+----2----+----3----+----4----+----5----+----6----+----7-|--+----| |
| 46 |
|
| 47 |
SUBROUTINE exf_interp( |
| 48 |
I infile, |
| 49 |
I filePrec, |
| 50 |
O arrayout, |
| 51 |
I irecord, xG_in, yG, |
| 52 |
I lon_0, lon_inc, |
| 53 |
I lat_0, lat_inc, |
| 54 |
I nx_in, ny_in, method, mythid) |
| 55 |
|
| 56 |
implicit none |
| 57 |
|
| 58 |
C infile (string) :: name of the binary input file (direct access) |
| 59 |
C filePrec (integer) :: number of bits per word in file (32 or 64) |
| 60 |
C arrout ( _RL ) :: output array |
| 61 |
C irecord (integer) :: record number to read |
| 62 |
C xG,yG :: coordinates for output grid to interpolate to |
| 63 |
C lon_0, lat_0 :: lon and lat of sw corner of global input grid |
| 64 |
C lon_inc :: scalar x-grid increment |
| 65 |
C lat_inc :: vector y-grid increments |
| 66 |
C nx_in,ny_in (integer) :: size in x & y direction of input file to read |
| 67 |
C method :: 1,11,21 for bilinear; 2,12,22 for bicubic |
| 68 |
C :: 1,2 for tracer; 11,12 for U; 21,22 for V |
| 69 |
C myThid (integer) :: My Thread Id number |
| 70 |
C |
| 71 |
|
| 72 |
#include "SIZE.h" |
| 73 |
#include "EEPARAMS.h" |
| 74 |
#include "PARAMS.h" |
| 75 |
|
| 76 |
C subroutine variables |
| 77 |
character*(*) infile |
| 78 |
integer filePrec, irecord, nx_in, ny_in |
| 79 |
_RL arrayout(1-OLx:sNx+OLx,1-OLy:sNy+OLy,nSx,nSy) |
| 80 |
_RS xG_in (1-OLx:sNx+OLx,1-OLy:sNy+OLy,nSx,nSy) |
| 81 |
_RS yG (1-OLx:sNx+OLx,1-OLy:sNy+OLy,nSx,nSy) |
| 82 |
_RL lon_0, lon_inc |
| 83 |
c _RL lat_0, lat_inc(ny_in-1) |
| 84 |
_RL lat_0, lat_inc(*) |
| 85 |
integer method, mythid |
| 86 |
|
| 87 |
C functions |
| 88 |
external lagran |
| 89 |
_RL lagran |
| 90 |
|
| 91 |
C local variables |
| 92 |
integer e_ind(snx,sny),w_ind(snx,sny) |
| 93 |
integer n_ind(snx,sny),s_ind(snx,sny) |
| 94 |
_RL px_ind(4), py_ind(4), ew_val(4) |
| 95 |
_RL arrayin(-1:nx_in+2 , -1:ny_in+2) |
| 96 |
_RL NorthValue |
| 97 |
_RL x_in (-1:nx_in+2), y_in(-1:ny_in+2) |
| 98 |
integer i, j, k, l, js, bi, bj, sp, interp_unit |
| 99 |
#ifdef TARGET_NEC_SX |
| 100 |
integer ic, ii, icnt |
| 101 |
integer inx(snx*sny,2) |
| 102 |
_RL ew_val1, ew_val2, ew_val3, ew_val4 |
| 103 |
#endif |
| 104 |
_RS xG(1-OLx:sNx+OLx,1-OLy:sNy+OLy,nSx,nSy) |
| 105 |
_RL ninety |
| 106 |
PARAMETER ( ninety = 90. ) |
| 107 |
_RS threeSixtyRS |
| 108 |
PARAMETER ( threeSixtyRS = 360. ) |
| 109 |
|
| 110 |
C put xG in interval [ lon_0 , lon_0+360 [ |
| 111 |
do bj=myByLo(myThid),myByHi(myThid) |
| 112 |
do bi=myBxLo(myThid),myBxHi(myThid) |
| 113 |
do j=1-OLy,sNy+OLy |
| 114 |
do i=1-OLx,sNx+OLx |
| 115 |
xG(i,j,bi,bj) = xG_in(i,j,bi,bj)-lon_0 |
| 116 |
& + threeSixtyRS*2. |
| 117 |
xG(i,j,bi,bj) = lon_0+mod(xG(i,j,bi,bj),threeSixtyRS) |
| 118 |
enddo |
| 119 |
enddo |
| 120 |
enddo |
| 121 |
enddo |
| 122 |
|
| 123 |
call exf_interp_read( |
| 124 |
I infile, filePrec, |
| 125 |
O arrayin, |
| 126 |
I irecord, nx_in, ny_in, mythid) |
| 127 |
|
| 128 |
C setup input longitude grid |
| 129 |
do i=-1,nx_in+2 |
| 130 |
x_in(i) = lon_0 + (i-1)*lon_inc |
| 131 |
enddo |
| 132 |
|
| 133 |
C setup input latitude grid |
| 134 |
y_in(0) = lat_0 - lat_inc(1) |
| 135 |
y_in(-1)= lat_0 - 2.*lat_inc(1) |
| 136 |
y_in(1) = lat_0 |
| 137 |
do j=2,ny_in |
| 138 |
y_in(j) = y_in(j-1) + lat_inc(j-1) |
| 139 |
enddo |
| 140 |
do j=ny_in+1,ny_in+2 |
| 141 |
if (y_in(j-1).eq.ninety) then |
| 142 |
y_in(j) = 2 * ninety - y_in(j-2) |
| 143 |
else |
| 144 |
i = max(1,ny_in-1) |
| 145 |
y_in(j) = min( y_in(j-1)+lat_inc(i), ninety ) |
| 146 |
endif |
| 147 |
enddo |
| 148 |
|
| 149 |
C enlarge boundary |
| 150 |
do j=1,ny_in |
| 151 |
arrayin(0,j) = arrayin(nx_in,j) |
| 152 |
arrayin(-1,j) = arrayin(nx_in-1,j) |
| 153 |
arrayin(nx_in+1,j) = arrayin(1,j) |
| 154 |
arrayin(nx_in+2,j) = arrayin(2,j) |
| 155 |
enddo |
| 156 |
do i=-1,nx_in+2 |
| 157 |
arrayin(i,0) = arrayin(i,1) |
| 158 |
arrayin(i,-1) = arrayin(i,1) |
| 159 |
arrayin(i,ny_in+1) = arrayin(i,ny_in) |
| 160 |
arrayin(i,ny_in+2) = arrayin(i,ny_in) |
| 161 |
enddo |
| 162 |
|
| 163 |
C For tracer (method=1,2) set to northernmost zonal-mean value |
| 164 |
C at 90N to avoid sharp zonal gradients near the Pole. |
| 165 |
C For U (method=11,12) set to zero at 90N to minimize velocity |
| 166 |
C gradient at North Pole |
| 167 |
C For V (method=11,12) set to northernmost zonal value at 90N, |
| 168 |
C as is already done above in order to allow cross-PoleArctic flow |
| 169 |
do j=ny_in,ny_in+2 |
| 170 |
if (y_in(j).eq.ninety) then |
| 171 |
if (method.eq.1 .or. method.eq.2) then |
| 172 |
NorthValue = 0. |
| 173 |
do i=1,nx_in |
| 174 |
NorthValue = NorthValue + arrayin(i,j) |
| 175 |
enddo |
| 176 |
NorthValue = NorthValue / nx_in |
| 177 |
do i=-1,nx_in+2 |
| 178 |
arrayin(i,j) = NorthValue |
| 179 |
enddo |
| 180 |
elseif (method.eq.11 .or. method.eq.12) then |
| 181 |
do i=-1,nx_in+2 |
| 182 |
arrayin(i,j) = 0. |
| 183 |
enddo |
| 184 |
endif |
| 185 |
endif |
| 186 |
enddo |
| 187 |
|
| 188 |
do bj = mybylo(mythid), mybyhi(mythid) |
| 189 |
do bi = mybxlo(mythid), mybxhi(mythid) |
| 190 |
|
| 191 |
C check validity of input/output coordinates |
| 192 |
#ifdef ALLOW_DEBUG |
| 193 |
if ( debugLevel.GE.debLevC ) then |
| 194 |
do j=1,sny |
| 195 |
do i=1,snx |
| 196 |
if ( xG(i,j,bi,bj) .lt. x_in(0) .or. |
| 197 |
& xG(i,j,bi,bj) .ge. x_in(nx_in+1) .or. |
| 198 |
& yG(i,j,bi,bj) .lt. y_in(0) .or. |
| 199 |
& yG(i,j,bi,bj) .ge. y_in(ny_in+1) ) then |
| 200 |
print*,'ERROR in S/R EXF_INTERP:' |
| 201 |
print*,' input grid must encompass output grid.' |
| 202 |
print*,'i,j,bi,bj' ,i,j,bi,bj |
| 203 |
print*,'xG,yG' ,xG(i,j,bi,bj),yG(i,j,bi,bj) |
| 204 |
print*,'nx_in,ny_in' ,nx_in ,ny_in |
| 205 |
print*,'x_in(0,nx_in+1)',x_in(0) ,x_in(nx_in+1) |
| 206 |
print*,'y_in(0,ny_in+1)',y_in(0) ,y_in(ny_in+1) |
| 207 |
STOP ' ABNORMAL END: S/R EXF_INTERP' |
| 208 |
endif |
| 209 |
enddo |
| 210 |
enddo |
| 211 |
endif |
| 212 |
#endif /* ALLOW_DEBUG */ |
| 213 |
|
| 214 |
C compute interpolation indices |
| 215 |
do j=1,sny |
| 216 |
do i=1,snx |
| 217 |
if (xG(i,j,bi,bj)-x_in(1) .ge. 0.) then |
| 218 |
w_ind(i,j) = int((xG(i,j,bi,bj)-x_in(1))/lon_inc) + 1 |
| 219 |
else |
| 220 |
w_ind(i,j) = int((xG(i,j,bi,bj)-x_in(1))/lon_inc) |
| 221 |
endif |
| 222 |
e_ind(i,j) = w_ind(i,j) + 1 |
| 223 |
enddo |
| 224 |
enddo |
| 225 |
#ifndef TARGET_NEC_SX |
| 226 |
C use the original and more readable variant of the algorithm that |
| 227 |
C has unvectorizable while-loops for each (i,j) |
| 228 |
do j=1,sny |
| 229 |
do i=1,snx |
| 230 |
js = ny_in*.5 |
| 231 |
do while (yG(i,j,bi,bj) .lt. y_in(js)) |
| 232 |
js = (js - 1)*.5 |
| 233 |
enddo |
| 234 |
do while (yG(i,j,bi,bj) .ge. y_in(js+1)) |
| 235 |
js = js + 1 |
| 236 |
enddo |
| 237 |
s_ind(i,j) = js |
| 238 |
enddo |
| 239 |
enddo |
| 240 |
#else /* TARGET_NEC_SX defined */ |
| 241 |
C this variant vectorizes more efficiently than the original one because |
| 242 |
C it moves the while loops out of the i,j loops (loop pushing) but |
| 243 |
C it is ugly and incomprehensible |
| 244 |
icnt = 0 |
| 245 |
do j=1,sny |
| 246 |
do i=1,snx |
| 247 |
s_ind(i,j) = ny_in*.5 |
| 248 |
icnt = icnt+1 |
| 249 |
inx(icnt,1) = i |
| 250 |
inx(icnt,2) = j |
| 251 |
enddo |
| 252 |
enddo |
| 253 |
do while (icnt .gt. 0) |
| 254 |
ii = 0 |
| 255 |
!CDIR NODEP |
| 256 |
do ic=1,icnt |
| 257 |
i = inx(ic,1) |
| 258 |
j = inx(ic,2) |
| 259 |
if (yG(i,j,bi,bj) .lt. y_in(s_ind(i,j))) then |
| 260 |
s_ind(i,j) = (s_ind(i,j) - 1)*.5 |
| 261 |
ii = ii+1 |
| 262 |
inx(ii,1) = i |
| 263 |
inx(ii,2) = j |
| 264 |
endif |
| 265 |
enddo |
| 266 |
icnt = ii |
| 267 |
enddo |
| 268 |
icnt = 0 |
| 269 |
do j=1,sny |
| 270 |
do i=1,snx |
| 271 |
icnt = icnt+1 |
| 272 |
inx(icnt,1) = i |
| 273 |
inx(icnt,2) = j |
| 274 |
enddo |
| 275 |
enddo |
| 276 |
do while (icnt .gt. 0) |
| 277 |
ii = 0 |
| 278 |
!CDIR NODEP |
| 279 |
do ic=1,icnt |
| 280 |
i = inx(ic,1) |
| 281 |
j = inx(ic,2) |
| 282 |
if (yG(i,j,bi,bj) .ge. y_in(s_ind(i,j)+1)) then |
| 283 |
s_ind(i,j) = s_ind(i,j) + 1 |
| 284 |
ii = ii+1 |
| 285 |
inx(ii,1) = i |
| 286 |
inx(ii,2) = j |
| 287 |
endif |
| 288 |
enddo |
| 289 |
icnt = ii |
| 290 |
enddo |
| 291 |
#endif /* TARGET_NEC_SX defined */ |
| 292 |
do j=1,sny |
| 293 |
do i=1,snx |
| 294 |
n_ind(i,j) = s_ind(i,j) + 1 |
| 295 |
enddo |
| 296 |
enddo |
| 297 |
|
| 298 |
if (method.eq.1 .or. method.eq.11 .or. method.eq.21) then |
| 299 |
|
| 300 |
C bilinear interpolation |
| 301 |
sp = 2 |
| 302 |
do j=1,sny |
| 303 |
do i=1,snx |
| 304 |
arrayout(i,j,bi,bj) = 0. |
| 305 |
do l=0,1 |
| 306 |
px_ind(l+1) = x_in(w_ind(i,j)+l) |
| 307 |
py_ind(l+1) = y_in(s_ind(i,j)+l) |
| 308 |
enddo |
| 309 |
#ifndef TARGET_NEC_SX |
| 310 |
do k=1,2 |
| 311 |
ew_val(k) = arrayin(w_ind(i,j),s_ind(i,j)+k-1) |
| 312 |
& *lagran(1,xG(i,j,bi,bj),px_ind,sp) |
| 313 |
& +arrayin(e_ind(i,j),s_ind(i,j)+k-1) |
| 314 |
& *lagran(2,xG(i,j,bi,bj),px_ind,sp) |
| 315 |
arrayout(i,j,bi,bj)=arrayout(i,j,bi,bj) |
| 316 |
& +ew_val(k)*lagran(k,yG(i,j,bi,bj),py_ind,sp) |
| 317 |
enddo |
| 318 |
#else |
| 319 |
ew_val1 = arrayin(w_ind(i,j),s_ind(i,j)+1-1) |
| 320 |
& *lagran(1,xG(i,j,bi,bj),px_ind,sp) |
| 321 |
& +arrayin(e_ind(i,j),s_ind(i,j)+1-1) |
| 322 |
& *lagran(2,xG(i,j,bi,bj),px_ind,sp) |
| 323 |
ew_val2 = arrayin(w_ind(i,j),s_ind(i,j)+2-1) |
| 324 |
& *lagran(1,xG(i,j,bi,bj),px_ind,sp) |
| 325 |
& +arrayin(e_ind(i,j),s_ind(i,j)+2-1) |
| 326 |
& *lagran(2,xG(i,j,bi,bj),px_ind,sp) |
| 327 |
arrayout(i,j,bi,bj)= |
| 328 |
& +ew_val1*lagran(1,yG(i,j,bi,bj),py_ind,sp) |
| 329 |
& +ew_val2*lagran(2,yG(i,j,bi,bj),py_ind,sp) |
| 330 |
#endif /* TARGET_NEC_SX defined */ |
| 331 |
enddo |
| 332 |
enddo |
| 333 |
elseif (method .eq. 2 .or. method.eq.12 .or. method.eq.22) then |
| 334 |
|
| 335 |
C bicubic interpolation |
| 336 |
sp = 4 |
| 337 |
do j=1,sny |
| 338 |
do i=1,snx |
| 339 |
arrayout(i,j,bi,bj) = 0. |
| 340 |
do l=-1,2 |
| 341 |
px_ind(l+2) = x_in(w_ind(i,j)+l) |
| 342 |
py_ind(l+2) = y_in(s_ind(i,j)+l) |
| 343 |
enddo |
| 344 |
#ifndef TARGET_NEC_SX |
| 345 |
do k=1,4 |
| 346 |
ew_val(k) = |
| 347 |
& arrayin(w_ind(i,j)-1,s_ind(i,j)+k-2) |
| 348 |
& *lagran(1,xG(i,j,bi,bj),px_ind,sp) |
| 349 |
& +arrayin(w_ind(i,j) ,s_ind(i,j)+k-2) |
| 350 |
& *lagran(2,xG(i,j,bi,bj),px_ind,sp) |
| 351 |
& +arrayin(e_ind(i,j) ,s_ind(i,j)+k-2) |
| 352 |
& *lagran(3,xG(i,j,bi,bj),px_ind,sp) |
| 353 |
& +arrayin(e_ind(i,j)+1,s_ind(i,j)+k-2) |
| 354 |
& *lagran(4,xG(i,j,bi,bj),px_ind,sp) |
| 355 |
arrayout(i,j,bi,bj)=arrayout(i,j,bi,bj) |
| 356 |
& +ew_val(k)*lagran(k,yG(i,j,bi,bj),py_ind,sp) |
| 357 |
enddo |
| 358 |
#else |
| 359 |
ew_val1 = |
| 360 |
& arrayin(w_ind(i,j)-1,s_ind(i,j)+1-2) |
| 361 |
& *lagran(1,xG(i,j,bi,bj),px_ind,sp) |
| 362 |
& +arrayin(w_ind(i,j) ,s_ind(i,j)+1-2) |
| 363 |
& *lagran(2,xG(i,j,bi,bj),px_ind,sp) |
| 364 |
& +arrayin(e_ind(i,j) ,s_ind(i,j)+1-2) |
| 365 |
& *lagran(3,xG(i,j,bi,bj),px_ind,sp) |
| 366 |
& +arrayin(e_ind(i,j)+1,s_ind(i,j)+1-2) |
| 367 |
& *lagran(4,xG(i,j,bi,bj),px_ind,sp) |
| 368 |
ew_val2 = |
| 369 |
& arrayin(w_ind(i,j)-1,s_ind(i,j)+2-2) |
| 370 |
& *lagran(1,xG(i,j,bi,bj),px_ind,sp) |
| 371 |
& +arrayin(w_ind(i,j) ,s_ind(i,j)+2-2) |
| 372 |
& *lagran(2,xG(i,j,bi,bj),px_ind,sp) |
| 373 |
& +arrayin(e_ind(i,j) ,s_ind(i,j)+2-2) |
| 374 |
& *lagran(3,xG(i,j,bi,bj),px_ind,sp) |
| 375 |
& +arrayin(e_ind(i,j)+1,s_ind(i,j)+2-2) |
| 376 |
& *lagran(4,xG(i,j,bi,bj),px_ind,sp) |
| 377 |
ew_val3 = |
| 378 |
& arrayin(w_ind(i,j)-1,s_ind(i,j)+3-2) |
| 379 |
& *lagran(1,xG(i,j,bi,bj),px_ind,sp) |
| 380 |
& +arrayin(w_ind(i,j) ,s_ind(i,j)+3-2) |
| 381 |
& *lagran(2,xG(i,j,bi,bj),px_ind,sp) |
| 382 |
& +arrayin(e_ind(i,j) ,s_ind(i,j)+3-2) |
| 383 |
& *lagran(3,xG(i,j,bi,bj),px_ind,sp) |
| 384 |
& +arrayin(e_ind(i,j)+1,s_ind(i,j)+3-2) |
| 385 |
& *lagran(4,xG(i,j,bi,bj),px_ind,sp) |
| 386 |
ew_val4 = |
| 387 |
& arrayin(w_ind(i,j)-1,s_ind(i,j)+4-2) |
| 388 |
& *lagran(1,xG(i,j,bi,bj),px_ind,sp) |
| 389 |
& +arrayin(w_ind(i,j) ,s_ind(i,j)+4-2) |
| 390 |
& *lagran(2,xG(i,j,bi,bj),px_ind,sp) |
| 391 |
& +arrayin(e_ind(i,j) ,s_ind(i,j)+4-2) |
| 392 |
& *lagran(3,xG(i,j,bi,bj),px_ind,sp) |
| 393 |
& +arrayin(e_ind(i,j)+1,s_ind(i,j)+4-2) |
| 394 |
& *lagran(4,xG(i,j,bi,bj),px_ind,sp) |
| 395 |
arrayout(i,j,bi,bj)= |
| 396 |
& +ew_val1*lagran(1,yG(i,j,bi,bj),py_ind,sp) |
| 397 |
& +ew_val2*lagran(2,yG(i,j,bi,bj),py_ind,sp) |
| 398 |
& +ew_val3*lagran(3,yG(i,j,bi,bj),py_ind,sp) |
| 399 |
& +ew_val4*lagran(4,yG(i,j,bi,bj),py_ind,sp) |
| 400 |
#endif /* TARGET_NEC_SX defined */ |
| 401 |
enddo |
| 402 |
enddo |
| 403 |
else |
| 404 |
stop 'stop in exf_interp.F: interpolation method not supported' |
| 405 |
endif |
| 406 |
enddo |
| 407 |
enddo |
| 408 |
|
| 409 |
RETURN |
| 410 |
END |
Parent Directory
| 
Revision Log
| 
Revision Graph