1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
|
// $Id: SubordinateStation.cc,v 1.2 2003/02/04 15:53:46 flaterco Exp $
/* SubordinateStation Implementation for offset station.
Copyright (C) 1998 David Flater.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
#include "common.hh"
SubordinateStation::SubordinateStation (ReferenceStation *rs,
SubordinateStationRef *ssr): Station (rs->context) {
stationRef = ssr;
name = ssr->name;
timeZone = ssr->timezone;
note = ssr->note;
if (timeZone.isNull())
timeZone = rs->timeZone;
coordinates = ssr->coordinates;
constants = rs->constants;
isCurrent = rs->isCurrent;
isHydraulicCurrent = rs->isHydraulicCurrent;
myUnits = rs->myUnits;
origOffsets = ssr->offsets;
residualOffsets = NULL;
reduce_offsets();
rs->constants = NULL;
delete rs;
}
PredictionValue SubordinateStation::minLevel() const {
PredictionValue pv = Station::minLevel();
if (residualOffsets) {
// Yet again...
if (residualOffsets->min.levelAdd.Units() != pv.Units())
residualOffsets->min.levelAdd.Units (pv.Units());
pv = pv * residualOffsets->min.levelMultiply() +
residualOffsets->min.levelAdd;
}
return pv;
}
PredictionValue SubordinateStation::maxLevel() const {
PredictionValue pv = Station::maxLevel();
if (residualOffsets) {
// And again.
if (residualOffsets->max.levelAdd.Units() != pv.Units())
residualOffsets->max.levelAdd.Units (pv.Units());
pv = pv * residualOffsets->max.levelMultiply() +
residualOffsets->max.levelAdd;
}
return pv;
}
void
SubordinateStation::reduce_offsets () {
if (residualOffsets)
delete residualOffsets;
residualOffsets = NULL;
assert (origOffsets);
// Step 1: reduce hairy offsets to simple wherever possible (catch
// hydraulics in step 2). While we're at it, extract current angles.
Offsets *mid = NULL;
if (origOffsets->is_simple())
mid = new SimpleOffsets (*((SimpleOffsets *)origOffsets));
else {
HairyOffsets *h = (HairyOffsets *)origOffsets;
if (isCurrent) { // Extract current angles.
minCurrentAngle = h->min.currentAngle;
maxCurrentAngle = h->max.currentAngle;
}
if (h->max == h->min &&
((!isCurrent) || (h->max.timeAdd == h->floodbegins.timeAdd &&
h->max.timeAdd == h->ebbbegins.timeAdd))) {
mid = new SimpleOffsets (h->max);
#ifdef SUPER_ULTRA_VERBOSE_DEBUGGING
cerr << "Reduced HairyOffsets to simple" << endl;
#endif
} else
mid = new HairyOffsets (*h);
}
// Step 2: extract whatever SimpleOffsets are relevant, and determine
// if any residual offsets remain.
SimpleOffsets s;
if (mid->is_simple()) {
SimpleOffsets *mp = (SimpleOffsets*)mid;
if ((!isHydraulicCurrent)||(mp->levelMultiply() == 1.0))
s = *mp;
else {
// We actually have to "un-simplify" this special case. In the
// case of hydraulic currents, level multipliers cannot be
// treated as simple and applied to the constants because of the
// square root operation.
residualOffsets = new HairyOffsets(*mp);
residualOffsets->max = residualOffsets->min = *mp;
residualOffsets->floodbegins.timeAdd =
residualOffsets->ebbbegins.timeAdd = mp->timeAdd;
#ifdef SUPER_ULTRA_VERBOSE_DEBUGGING
cerr << "Un-simplified offsets for hydraulic station" << endl;
#endif
}
} else {
HairyOffsets *mp = (HairyOffsets*)mid;
residualOffsets = new HairyOffsets(*mp);
// We already know that this can't be reduced to SimpleOffsets.
// There isn't any point in factoring out individual fields since
// hairy offsets need to be processed in the prediction routine in
// both cases. However, it's important to center the time to
// minimize the inaccuracy in predictApproximate for slack
// offsets. (If you were to factor out levelAdd and
// levelMultiply, you would still need to avoid the pitfall of
// messing up the order of operations. For example, factoring out
// a levelAdd but not a levelMultiply would yield incorrect
// results.)
}
// Step 3: if we have residual offsets, center the timeAdd.
if (residualOffsets) {
// If current, then center time for slack. Otherwise, center
// time over max and min.
if (isCurrent) {
s.timeAdd = (residualOffsets->floodbegins.timeAdd + residualOffsets->ebbbegins.timeAdd) / 2;
residualOffsets->floodbegins.timeAdd -= s.timeAdd;
residualOffsets->ebbbegins.timeAdd -= s.timeAdd;
} else
s.timeAdd = (residualOffsets->max.timeAdd + residualOffsets->min.timeAdd) / 2;
residualOffsets->max.timeAdd -= s.timeAdd;
residualOffsets->min.timeAdd -= s.timeAdd;
}
// Step 4: apply SimpleOffsets.
constants->setSimpleOffsets (s);
#ifdef SUPER_ULTRA_VERBOSE_DEBUGGING
cerr << "Orig offsets: " << origOffsets << endl;
cerr << "Mid-offsets: " << mid << endl;
cerr << "Applied offsets: " << &s << endl;
cerr << "Residual offsets: " << residualOffsets << endl;
#endif
// Just "delete mid" yields a segmentation violation.
if (mid->is_simple()) {
SimpleOffsets *mp = (SimpleOffsets*)mid;
delete mp;
} else {
HairyOffsets *mp = (HairyOffsets*)mid;
delete mp;
}
}
SubordinateStation::~SubordinateStation() {
if (constants)
delete constants;
if (markLevel)
delete markLevel;
}
// tm is "uncorrected" or "internal" timestamp
// t_out is "corrected" timestamp (with offsets applied)
// For reference station or simple offsets, they are identical.
// pv_out is undefined if the event is a sun or moon event.
void SubordinateStation::predictExactTideEvent (Timestamp &tm, Direction d,
Timestamp &t_out, EventType &etype_out, Dstr &etype_desc_long,
Dstr &etype_desc_short, PredictionValue &pv_out) {
_predictExactTideEvent (tm, d, etype_out);
t_out = tm;
if (!isSunMoonEvent (etype_out)) {
// Explicit Station:: gets the original, un-corrected value
pv_out = Station::predictApproximate (tm);
// It may be inefficient to do this here, but it's a lot more robust.
if (residualOffsets) {
if (residualOffsets->max.levelAdd.Units() != pv_out.Units())
residualOffsets->max.levelAdd.Units (pv_out.Units());
if (residualOffsets->min.levelAdd.Units() != pv_out.Units())
residualOffsets->min.levelAdd.Units (pv_out.Units());
}
}
// Apply residual offsets.
if (residualOffsets) {
switch (etype_out) {
case max:
t_out += residualOffsets->max.timeAdd;
pv_out = pv_out * residualOffsets->max.levelMultiply()
+ residualOffsets->max.levelAdd;
break;
case min:
t_out += residualOffsets->min.timeAdd;
pv_out = pv_out * residualOffsets->min.levelMultiply()
+ residualOffsets->min.levelAdd;
break;
case slackrise:
t_out += residualOffsets->floodbegins.timeAdd;
break;
case slackfall:
t_out += residualOffsets->ebbbegins.timeAdd;
break;
// Mark should never be allowed if there are residual offsets.
case markrise:
case markfall:
{
Dstr details ("The offending station is ");
details += name;
details += ".";
barf (ILLEGAL_MARK, details);
}
default:
;
}
}
// The etype_desc figuring needs to be delayed until after everything
// so that "Max Flood" versus "Min Ebb" will be correct when offsets
// bump something over the line.
etypedesc (etype_out, pv_out, etype_desc_long, etype_desc_short);
}
// Interpolate approximate tide levels for subordinate stations. This is
// the most difficult algorithm in the whole program. When NOAA needed to
// do this, they just used a simple sine curve to interpolate. Nobody
// liked that answer.
//
// The original algorithm submitted by Jean-Pierre Lapointe
// (scipur@collegenotre-dame.qc.ca), which is in block comments far
// below, got me past the question of how to do the interpolation, but
// assumed the availability of usable high and low tides for the day
// of prediction. Getting these tides turned out to be highly
// non-trivial. It is not as simple as just using the two nearest
// bracketing tide events for the following reasons:
//
// 1. Some locations like to produce lots of maxima that are
// actually below the mean, and minima that are above the mean.
// These data points cause the algorithm to blow up, and must be
// thrown out.
//
// 2. The lower high and higher low tide events produced by mostly
// diurnal tides, as well as the various other "meaningless" maxima
// and minima that happen with such tides, may be closer together
// than the offset intervals. In these cases it is better to just
// apply offsets to the higher high and lower low, and leave the
// others to be interpolated as they will. Ed Wallner makes the
// following relevant quote in his diurnal.txt file:
//
// The NOS tide table notes on using tidal differences include
// the statement that " ... at subordinate stations where the
// tide is chiefly diurnal, the tide corrections are intended to
// be used to approximate the times and heights of the higher
// high and lower low waters. When the lower high water and
// higher low water at the reference station are nearly the same
// height, great reliance should not be placed on the calculated
// corresponding tides at the subordinate station."
//
// In these rare cases, it would actually be more accurate to
// derive tide events from predictApproximate than to faithfully
// apply the offsets with predictExact, but it is more important to
// produce consistent results than to try to make sense of a
// degenerate case.
//
// 3. The choices of highs and lows to use must be deterministic
// and consistent based on a given Timestamp input. This
// complicates the caching of values, but failure to cache here
// would lead to totally unacceptable performance.
//
// The implementation of this algorithm in XTide 1 was not too smooth
// in its approach to solving these problems. It used a fixed search
// interval (originally 26 hours, later expanded to 30) and used the
// highest high and lowest low tides within that interval, with
// kludges to stretch the interval as needed to solve problem #1. To
// make a long story short, it didn't really work; it unfairly
// penalized the accuracy of regular old semi-diurnal stations, and it
// was not consistent. The interval had to be expanded because Dublon
// managed to produce a nearly 29-hour period with no high tide events
// and crash the algorithm (from 1997-09-10 19:18 to 1997-09-12 00:13
// NCST).
//
// A large part of the reason for using this crummy approach in XTide
// 1 was the lack of a facility for searching backwards for tide
// events. Finding previous tide events without a fixed interval to
// search in would have been prohibitive. This problem has been
// corrected in XTide 2.
//
// XTide 2 doesn't use a fixed interval. It just keeps looking for
// tide events until it finds some that are a "reasonable distance"
// away from the mean. "Reasonable" is defined as a fraction of the
// total theoretical amplitude in config.hh.
//
// Since I use a localized mean instead of a global one, we shouldn't
// have to look too far to find a reasonable answer (provided that
// reasonable_fraction is reasonably defined). The caching problem is
// solved by using the actual two nearest tide events as a bracket,
// and forcing a refresh when the bracket is exited.
//
// NOTE 1: It is still theoretically possible to obtain inconsistent
// results if a tide is on the very border of "reasonable" and gets
// thrown over by inaccuracy in the third decimal of a tide
// prediction. If you can demonstrate this bug in practice, you
// deserve a medal or something.
//
// NOTE 2: This algorithm does nothing about slack water offsets.
// Following is Jean-Pierre's explanation of his original algorithm.
// Things that I changed are:
//
// (1) Use a moving mean instead of the datum. Using the datum
// gives unusable results in summer and winter for locations with
// large seasonal variations.
//
// (2) Interpolate multiplicative offsets as well as additive ones.
// Jean-Pierre figured that the changeover from low-tide
// multipliers to high-tide multipliers always happened at zero,
// which is not the case here. We need to interpolate to avoid
// causing a discontinuity.
/************** COMMENT BLOCK ************************
1- Store the following data
Z (datum at reference station)
t (time of prediction at the secondary station)
TH (correction for high tide time at secondary station)
TL (correction for low tide time at secondary station)
RH (correction ratio for high tide height at secondary station)
RL (correction ratio for low tide height at secondary station)
HH (height correction for high tide height at secondary station)
HL (height correction for low tide height at secondary station)
2- Run XTIDE for the reference station for the day of prediction
to find the height of the higher tide and store it as MAX and
to find the height of the lower tide and store it as MIN.
3- Run XTIDE for the reference station at a time T defined as
T = t - (TH + TL) / 2
Store the height calculated by XTIDE as HI (intermediate height)
4- Store S defined as S =HI - Z
5- Run XTIDE for the reference station at a time T defined as:
if S > 0 then
T = t - (TH + TL) / 2 - (TH-TL) / 2 x S / absolute value(MAX - Z)
else
T = t - (TH + TL) / 2 - (TH-TL) / 2 x S / absolute value(MIN - Z)
Store the height calculated by XTIDE as HI (intermediate height)
and calculate HS (height at secondary station at time t) defined as:
if S > 0 then
HS = HI x RH + (HH + HL) / 2 + (HH-HL) / 2 x S/absolute value(MAX - Z)
else
HS = HI x RL + (HH + HL) / 2 + (HH-HL) / 2 x S/absolute value(MIN - Z)
You now have HS the height of the tide at the secondary station at a
time t for this station.
******** END COMMENT BLOCK **********************************************/
PredictionValue SubordinateStation::predictApproximate (Timestamp tm) {
// Get rid of the trivial cases.
if (!residualOffsets)
return Station::predictApproximate (tm);
// Invalidate cache?
if (!bracketleft.isNull())
if (tm < bracketleft || tm > bracketright || usemax.Units() != myUnits)
bracketleft.make_null();
// Refresh cache?
// This is tricky. A criterion of "nearest reasonable high/low
// tide" would violate the requirement for consistency -- you could
// get different results based on where in the interval you are when
// you refresh the cache. One might instead constrain the direction
// of the search so that we always get one tide on each side, of the
// same type as the bracketing tides. But this causes
// discontinuities since the value being used in lieu of an
// unreasonable max or min will change as you cross that max or min.
// So, we have to be extra-cynical, and just always search in the
// same direction.
if (bracketleft.isNull()) {
#ifdef SUPER_ULTRA_VERBOSE_DEBUGGING
cerr << "Refreshing predictApproximate cache" << endl;
#endif
int bhiflag=0, fhiflag=0;
Timestamp b, f;
PredictionValue pv;
// It's plausible that we might miss a tide if we are sitting right
// on top of it -- detect that and correct for it.
{
Timestamp ttm = tm;
int itcnt = 0;
do {
b = bracketleft = next_high_or_low_tide (ttm, bhiflag, backward);
f = bracketright = next_high_or_low_tide (ttm, fhiflag, forward);
ttm += Interval(60);
assert (itcnt++ < 2);
} while (bhiflag == fhiflag);
#ifdef SUPER_ULTRA_VERBOSE_DEBUGGING
if (itcnt > 1)
cerr << " ** Corrected for bhiflag == bloflag!" << endl;
#endif
}
// Make the left side reasonable.
pv = Station::predictApproximate(b);
while (!(isReasonable (b, pv, bhiflag))) {
#ifdef SUPER_ULTRA_VERBOSE_DEBUGGING
cerr << " * Left side unreasonable" << endl;
#endif
int tempflag;
do
b = next_high_or_low_tide (b, tempflag, forward); // See comments.
while (tempflag != bhiflag);
pv = Station::predictApproximate(b);
}
if (bhiflag)
usemax = pv;
else
usemin = pv;
// Make the right side reasonable.
pv = Station::predictApproximate(f);
while (!(isReasonable (f, pv, fhiflag))) {
#ifdef SUPER_ULTRA_VERBOSE_DEBUGGING
cerr << " * Right side unreasonable" << endl;
#endif
int tempflag;
do
f = next_high_or_low_tide (f, tempflag, forward);
while (tempflag != fhiflag);
pv = Station::predictApproximate(f);
}
if (fhiflag)
usemax = pv;
else
usemin = pv;
assert (bhiflag != fhiflag);
#ifdef SUPER_ULTRA_VERBOSE_DEBUGGING
cerr << "predictApproximate cache refreshed" << endl;
cerr << " timestamp: " << tm << endl;
cerr << " l bracket: " << bracketleft << endl;
cerr << " r bracket: " << bracketright << endl;
cerr << " l using: " << b << endl;
cerr << " r using: " << f << endl;
cerr << " usemax: " << usemax << endl;
cerr << " usemin: " << usemin << endl;
#endif
}
// Do the interpolation. The capital letter variable names here
// match up with those used in the algorithm in the comment block
// above.
/* This is the initial guess (average of time offsets) */
Timestamp T = tm - (residualOffsets->max.timeAdd +
residualOffsets->min.timeAdd) / 2;
PredictionValue Z = movingMean(T);
PredictionValue S = Station::predictApproximate(T) - Z;
/* Improve the initial guess. */
double magicnum;
if (S.val() > 0.0)
magicnum = 0.5 * S.val() / fabs(usemax.val() - Z.val());
else
magicnum = 0.5 * S.val() / fabs(usemin.val() - Z.val());
T -= (residualOffsets->max.timeAdd - residualOffsets->min.timeAdd) * magicnum;
PredictionValue HI = Station::predictApproximate(T);
/* Apply the height offsets. */
// It may be inefficient to do this here, but it's a lot more robust.
if (residualOffsets->max.levelAdd.Units() != HI.Units())
residualOffsets->max.levelAdd.Units (HI.Units());
if (residualOffsets->min.levelAdd.Units() != HI.Units())
residualOffsets->min.levelAdd.Units (HI.Units());
return HI * ((residualOffsets->max.levelMultiply() +
residualOffsets->min.levelMultiply()) / 2.0 +
(residualOffsets->max.levelMultiply() -
residualOffsets->min.levelMultiply()) * magicnum)
+ (residualOffsets->max.levelAdd +
residualOffsets->min.levelAdd) / 2.0 +
(residualOffsets->max.levelAdd -
residualOffsets->min.levelAdd) * magicnum;
}
int SubordinateStation::isReasonable (Timestamp tm, PredictionValue
pv, int ishigh) {
// Calculate a fresh mean when testing for reasonableness so that
// the decision will be most consistent.
PredictionValue mean = movingMean (tm);
if (ishigh)
return ((pv - mean) / (constants->maxAmplitude) > reasonable_fraction);
else
return ((mean - pv) / (constants->maxAmplitude) > reasonable_fraction);
}
int SubordinateStation::is_reference_station() {
return 0;
}
|