Newer
Older
#'@rdname CST_Analogs
#'@title Downscaling using Analogs based on large scale fields.
#'
#'@author Carmen Alvarez-Castro, \email{carmen.alvarez-castro@cmcc.it}
#'@author Nuria Perez-Zanon \email{nuria.perez@bsc.es}
#'@description This function perform a downscaling using Analogs. To compute
#'the analogs, the function search for days with similar large scale conditions
#'to downscaled fields in the local scale.
#'The large scale and the local scale regions are defined by the user.
#'The large scale is usually given by atmospheric circulation as sea level
#'pressure or geopotential height (Yiou et al, 2013) but the function gives the
#' possibility to use another field. The local scale will be usually given by
#' precipitation or temperature fields, but might be another variable.
#' The analogs function will find the best analogs based in three criterias:
#' (1) Minimal distance in the large scale pattern (i.e. SLP)
#' (2) Minimal distance in the large scale pattern (i.e. SLP) and minimal
#' distance in the local scale pattern (i.e. SLP).
#' (3) Minimal distance in the large scale pattern (i.e. SLP), minimal
#' distance in the local scale pattern (i.e. SLP) and maxima correlation in the
#' local variable to downscale (i.e Precipitation).
#' The search of analogs must be done in the longest dataset posible. This is
#' important since it is necessary to have a good representation of the
#' possible states of the field in the past, and therefore, to get better
#' analogs. Once the search of the analogs is complete, and in order to used the
#' three criterias the user can select a number of analogsi, using parameter 'nAnalogs' to restrict
#' the selection of the best analogs in a short number of posibilities, the best
#' This function has not constrains of specific regions, variables to downscale,
#' or data to be used (seasonal forecast data, climate projections data,
#' reanalyses data).
#' The regrid into a finner scale is done interpolating with CST_Load.
#' Then, this interpolation is corrected selecting the analogs in the large
#' and local scale in based of the observations.
#' The function is an adapted version of the method of Yiou et al 2013.
#'@references Yiou, P., T. Salameh, P. Drobinski, L. Menut, R. Vautard,
#' and M. Vrac, 2013 : Ensemble reconstruction of the atmospheric column
#' from surface pressure using analogues. Clim. Dyn., 41, 1419-1437.
#' \email{pascal.yiou@lsce.ipsl.fr}
#'
#'@param expL an 's2dv_cube' object containing the experimental field on the large scale for which the analog is aimed. This field is used to in all the criterias. If parameter 'expVar' is not provided, the function will return the expL analog. The element 'data' in the 's2dv_cube' object must have, at least, latitudinal and longitudinal dimensions. The object is expect to be already subset for the desired large scale region.
#'@param obsL an 's2dv_cube' object containing the observational field on the large scale. The element 'data' in the 's2dv_cube' object must have the same latitudinal and longitudinal dimensions as parameter 'expL' and a temporal dimension with the maximum number of available observations.
#'@param expVar an 's2dv_cube' object containing the experimental field on the local scale, usually a different variable to the parameter 'expL'. If it is not NULL (by default, NULL), the returned field by this function will be the analog of parameter 'expVar'.
#'@param obsVar an 's2dv_cube' containing the field of the same variable as the passed in parameter 'expVar' for the same region.
#'@param region a vector of length four indicating the minimum longitude, the maximum longitude, the minimum latitude and the maximum latitude.
#'@param criteria a character string indicating the criteria to be used for the selection of analogs:
#'\itemize{
#'\item{Large_dist} minimal distance in the large scale pattern;
#'\item{Local_dist} minimal distance in the large scale pattern and minimal
#' distance in the local scale pattern; and
#'\item{Local_cor} minimal distance in the large scale pattern, minimal
#' distance in the local scale pattern and maxima correlation in the
#' local variable to downscale.}
#'
#'@import ClimProjDiags
#'
#'@seealso code{\link{CST_Load}}, \code{\link[s2dverification]{Load}} and \code{\link[s2dverification]{CDORemap}}
#'
#'@return An 's2dv_cube' object containing the dowscaled values of the best analogs in the criteria selected.
#'@examples
#'res <- CST_Analogs(expL = lonlat_data$exp, obsL = lonlat_data$obs)
CST_Analogs <- function(expL, obsL, time_obsL, expVar = NULL, obsVar = NULL,
region = NULL, criteria = "Large_dist") {
if (!inherits(expL, 's2dv_cube') || !inherits(obsL, 's2dv_cube')) {
stop("Parameter 'expL' and 'obsL' must be of the class 's2dv_cube', ",
"as output by CSTools::CST_Load.")
}
if (!is.null(expVar) || !is.null(obsVar)) {
if (!inherits(expVar, 's2dv_cube') || !inherits(obsVar, 's2dv_cube')) {
stop("Parameter 'expVar' and 'obsVar' must be of the class 's2dv_cube', ",
"as output by CSTools::CST_Load.")
}
}
if (!is.null(expVar)) {
region <- c(min(expVar$lon), max(expVar$lon), min(expVar$lat), max(expVar$lon))
lonVar <- expVar$lon
latVar <- expVar$lat
} else {
region <- c(min(expL$lon), max(expL$lon), min(expL$lat), max(expL$lon))
lonVar <- expL$lon
latVar <- expL$lat
}
result <- Analogs(expL$data, obsL$data, time_obsL = timevector,
expVar = expVar$data, obsVar = obsVar$data,
criteria = criteria,
lonVar = expVar$lon, latVar = expVar$lat,
region = region, nAnalogs = 1, return_list = FALSE)
if (!is.null(obsVar)) {
obsVar$data <- result$AnalogsFields
return(obsVar)
} else {
obsL$data <- result$AnalogsFields
return(obsL)
}
#'@rdname Analogs
#'@title Search for analogs based on large scale fields.
#'
#'@author Carmen Alvarez-Castro, \email{carmen.alvarez-castro@cmcc.it}
#'@author Nuria Perez-Zanon \email{nuria.perez@bsc.es}
#'
#'@description This function perform a downscaling using Analogs. To compute
#'the analogs, the function search for days with similar large scale conditions
#'to downscaled fields in the local scale.
#'The large scale and the local scale regions are defined by the user.
#'The large scale is usually given by atmospheric circulation as sea level
#'pressure or geopotential height (Yiou et al, 2013) but the function gives the
#' possibility to use another field. The local scale will be usually given by
#' precipitation or temperature fields, but might be another variable.
#' The analogs function will find the best analogs based in three criterias:
#' (1) Minimal distance in the large scale pattern (i.e. SLP)
#' (2) Minimal distance in the large scale pattern (i.e. SLP) and minimal
#' distance in the local scale pattern (i.e. SLP).
#' (3) Minimal distance in the large scale pattern (i.e. SLP), minimal
#' distance in the local scale pattern (i.e. SLP) and maxima correlation in the
#' local variable to downscale (i.e Precipitation).
#' The search of analogs must be done in the longest dataset posible. This is
#' important since it is necessary to have a good representation of the
#' possible states of the field in the past, and therefore, to get better
#' analogs. Once the search of the analogs is complete, and in order to used the
#' three criterias the user can select a number of analogsi, using parameter
#' 'nAnalogs' to restrict
#' the selection of the best analogs in a short number of posibilities, the best
#' ones.
#' This function has not constrains of specific regions, variables to downscale,
#' or data to be used (seasonal forecast data, climate projections data,
#' reanalyses data).
#' The regrid into a finner scale is done interpolating with CST_Load.
#' Then, this interpolation is corrected selecting the analogs in the large
#' and local scale in based of the observations.
#' The function is an adapted version of the method of Yiou et al 2013.
#'@references Yiou, P., T. Salameh, P. Drobinski, L. Menut, R. Vautard,
#' and M. Vrac, 2013 : Ensemble reconstruction of the atmospheric column
#' from surface pressure using analogues. Clim. Dyn., 41, 1419-1437.
#' \email{pascal.yiou@lsce.ipsl.fr}
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
#'
#'@param expL an array of N named dimensions containing the experimental field
#' on the large scale for which the analog is aimed. This field is used to in
#' all the criterias. If parameter 'expVar' is not provided, the function will
#' return the expL analog. The element 'data' in the 's2dv_cube' object must
#' have, at least, latitudinal and longitudinal dimensions. The object is
#' expect to be already subset for the desired large scale region.
#'@param obsL an array of N named dimensions containing the observational field
#'on the large scale. The element 'data' in the 's2dv_cube' object must have
#'the same latitudinal and longitudinal dimensions as parameter 'expL' and a
#' temporal dimension with the maximum number of available observations.
#'@param expVar an array of N named dimensions containing the experimental
#'field on the local scale, usually a different variable to the parameter
#''expL'. If it is not NULL (by default, NULL), the returned field by this
#'function will be the analog of parameter 'expVar'.
#'@param obsVar an array of N named dimensions containing the field of the same variable as the passed in parameter 'expVar' for the same region.
#'@param criteria a character string indicating the criteria to be used for the selection of analogs:
#'\itemize{
#'\item{Large_dist} minimal distance in the large scale pattern;
#'\item{Local_dist} minimal distance in the large scale pattern and minimal
#' distance in the local scale pattern; and
#'\item{Local_cor} minimal distance in the large scale pattern, minimal
#' distance in the local scale pattern and maxima correlation in the
#' local variable to downscale.}
#'@param lonVar a vector containing the longitude of parameter 'expVar'.
#'@param latVar a vector containing the latitude of parameter 'expVar'.
#'@param region a vector of length four indicating the minimum longitude,
#'the maximum longitude, the minimum latitude and the maximum latitude.
#'@param return_list TRUE if you want to get a list with the best analogs FALSE
#'#'if not.
#'@param nAnalogs number of Analogs to be selected to apply the criterias (this
#'is not the necessary the number of analogs that the user can get, but the number
#'of events with minimal distance in which perform the search of the best Analog.
#' The default value for the Large_dist criteria is 1, the default value for
#' the Local_dist criteria is 10 and same for Local_cor. If return_list is
#' False you will get just the first one for downscaling purposes. If return_list
#' is True you will get the list of the best analogs that were searched in nAnalogs
#' under the selected criterias.
#'@import ClimProjDiags
#'@return list with the best analogs (time, distance)
#'@return dowscaled values of the best analogs for the criteria selected.
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
#'require(zeallot)
#'
#' # Example 1:Large_dist
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:180),expL*2)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' downscale_field <- Analogs(expL, obsL, time_obsL)
#' layout(matrix(1:3,1,3,byrow=T))
#' image(expL,main="expL")
#' image(downscale_field$AnalogsFields,
#' main=paste0("Best_Analog ",downscale_field$DatesAnalogs))
#'
#' # Example 2:Large_dist imposing nAnalogs and return_list
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:1980),expL*1.5)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 100)
#' time_obsL <- paste(rep("01", 100), rep("01", 100), 1920 : 2019, sep = "-")
#' nAnalogs=30
#' downscale_field <- Analogs(expL, obsL, time_obsL,nAnalogs=nAnalogs,return_list = TRUE)
#' str(downscale_field)
#' plot.new()
#' layout(matrix(1:3,1,3,byrow=T))
#' image(expL,main="expL")
#' image(downscale_field$AnalogsFields[,,1],
#' main=paste0("Best_Analog ",downscale_field$DatesAnalogs[1]))
#' image(downscale_field$AnalogsFields[,,2],
#' main=paste0("2nd Best_Analog ",downscale_field$DatesAnalogs[2]))
#'
#' # Example 3:Local_dist with obsVar and expVar return_list = FALSE
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:180),expL*2)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' expVar <- expL[1:3,1:3]
#' dim(expVar) <- c(lat = 3, lon = 3)
#' obsVar <- obsL[1:3,1:3,1:10]
#' dim(obsVar) <- c(lat = 3, lon = 3, time = 10)
#' lonmin=-1
#' lonmax=2
#' latmin=30
#' latmax=33
#' region=c(lonmin,lonmax,latmin,latmax)
#' Local_scale <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,obsVar=obsVar,expVar=expVar,
#' criteria="Local_dist",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 5, return_list = FALSE)
#' plot.new()
#' layout(matrix(1:3,1,3,byrow=T))
#' image(expL,main="expL")
#' image(expVar,main="expVar")
#' image(Local_scale$AnalogsFields,
#' main=paste0("Best_Analog ",Local_scale$DatesAnalogs))
#'
#' # Example 4:Large_dist and Local_dist analogs. Local_dist with obsVar and expVar. return_list = FALSE in both
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:180),expL*5)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' expVar <- expL[1:3,1:3]
#' dim(expVar) <- c(lat = 3, lon = 3)
#' obsVar <- obsL[1:3,1:3,1:10]
#' dim(obsVar) <- c(lat = 3, lon = 3, time = 10)
#'
#' # analogs of large scale using criteria 1
#' Large_scale <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,
#' criteria="Large_dist",
#' nAnalogs = 10, return_list = FALSE)
#' # analogs of local scale using criteria 2
#' lonmin=-1
#' lonmax=2
#' latmin=30
#' latmax=33
#' region=c(lonmin,lonmax,latmin,latmax)
#' Local_scale <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,obsVar=obsVar,expVar=expVar,
#' criteria="Local_dist",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 10, return_list = FALSE)
#' plot.new()
#' layout(matrix(1:4,2,2,byrow=T))
#' image(expL,main="expL")
#' image(Large_scale$AnalogsFields,
#' main=paste0("Best_Analog ",Large_scale$DatesAnalogs))
#' image(expVar,main="expVar")
#' image(Local_scale$AnalogsFields,
#' main=paste0("Best_Analog ",Local_scale$DatesAnalogs))
#'
#' # Example 5: Local_dist without obsVar and expVar
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:180),expL*2)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' lonmin=-1
#' lonmax=2
#' latmin=30
#' latmax=33
#' region=c(lonmin,lonmax,latmin,latmax)
#' Local_scale2 <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,
#' criteria="Local_dist",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 10, return_list = FALSE)
#' plot.new()
#' layout(matrix(1:4,2,2,byrow=T))
#' image(expL,main="expL")
#' image(Local_scale2$AnalogsFields,
#' main=paste0("Best_Analog ",Local_scale2$DatesAnalogs))
#' image(expVar,main="expVar")
#' image(Local_scale$AnalogsFields,
#' main=paste0("Best_Analog ",Local_scale$DatesAnalogs))
#'
#' # Example 6:Local_dist with obsVar and expVar return_list = TRUE
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:180),expL*2)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' expVar <- expL[1:3,1:3]
#' dim(expVar) <- c(lat = 3, lon = 3)
#' obsVar <- obsL[1:3,1:3,1:10]
#' dim(obsVar) <- c(lat = 3, lon = 3, time = 10)
#' lonmin=-1
#' lonmax=2
#' latmin=30
#' latmax=33
#' region=c(lonmin,lonmax,latmin,latmax)
#' Local_scale <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,obsVar=obsVar,expVar=expVar,
#' criteria="Local_dist",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 5, return_list = TRUE)
#' str(Local_scale)
#'
#' # Example 7: Local_cor with obsVar and expVar return_list = FALSE
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:180),expL*5)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' expVar <- expL[1:3,1:3]
#' dim(expVar) <- c(lat = 3, lon = 3)
#' obsVar <- obsL[1:3,1:3,1:10]
#' dim(obsVar) <- c(lat = 3, lon = 3, time = 10)
#' lonmin=-1
#' lonmax=2
#' latmin=30
#' latmax=33
#' region=c(lonmin,lonmax,latmin,latmax)
#' Local_corr <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,obsVar=obsVar,expVar=expVar,
#' criteria="Local_cor",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 5, return_list = FALSE)
#' plot.new()
#' layout(matrix(1:3,1,3,byrow=T))
#' image(expL,main="expL")
#' image(expVar,main="expVar")
#' image(Local_corr$AnalogsFields[,,1],
#' main=paste0("Best_Analog ",Local_corr$DatesAnalogs[1]))
#'
#' # Example 8: Local_cor return list TRUE
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:180),expL*5)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' expVar <- expL[1:3,1:3]
#' dim(expVar) <- c(lat = 3, lon = 3)
#' obsVar <- obsL[1:3,1:3,1:10]
#' dim(obsVar) <- c(lat = 3, lon = 3, time = 10)
#' lonmin=-1
#' lonmax=2
#' latmin=30
#' latmax=33
#' region=c(lonmin,lonmax,latmin,latmax)
#' Local_corr <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,obsVar=obsVar,expVar=expVar,
#' criteria="Local_cor",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 5, return_list = TRUE)
#' plot.new()
#' layout(matrix(1:4,2,2,byrow=T))
#' image(expL,main="expL")
#' image(expVar,main="expVar")
#' image(Local_corr$AnalogsFields[,,1],
#' main=paste0("Best_Analog ",Local_corr$DatesAnalogs[1]))
#' image(Local_corr$AnalogsFields[,,2],
#' main=paste0("2nd Best_Analog ",Local_corr$DatesAnalogs[2]))
#'
#' # Example 9: Large_dist, Local_dist, and Local_cor return list FALSE same variable
#' expL <- rnorm(1:20)
#' dim(expL) <- c(lat = 4, lon = 5)
#' obsL <- c(rnorm(1:180),expL*7)
#' dim(obsL) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' # analogs of large scale using criteria 1
#' Large_scale <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,
#' criteria="Large_dist",
#' nAnalogs = 10, return_list = TRUE)
#' # analogs of local scale using criteria 2
#' lonmin=-1
#' lonmax=2
#' latmin=30
#' latmax=33
#' region=c(lonmin,lonmax,latmin,latmax)
#' Local_scale <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,
#' criteria="Local_dist",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 10, return_list = TRUE)
#' # analogs of local scale using criteria 2
#' Local_corr <- Analogs(expL=expL,
#' obsL=obsL, time_obsL=time_obsL,
#' criteria="Local_cor",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 10, return_list = TRUE)
#' plot.new()
#' layout(matrix(1:9,3,3,byrow=T))
#' image(expL,main="expL")
#' image(Large_scale$AnalogsFields[,,1],
#' main=paste0("Best_Analog C1 ",Large_scale$DatesAnalogs[1]))
#' image(Large_scale$AnalogsFields[,,2],
#' main=paste0("Best_Analog C1 ",Large_scale$DatesAnalogs[2]))
#' image(expVar,main="expVar")
#' image(Local_scale$AnalogsFields[,,1],
#' main=paste0("Best_Analog C2 ",Local_scale$DatesAnalogs[1]))
#' image(Local_scale$AnalogsFields[,,2],
#' main=paste0("Best_Analog C2 ",Local_scale$DatesAnalogs[2]))
#' image(expVar,main="expVar")
#' image(Local_corr$AnalogsFields[,,1],
#' main=paste0("Best_Analog C3 ",Local_corr$DatesAnalogs[1]))
#' image(Local_corr$AnalogsFields[,,2],
#' main=paste0("2nd Best_Analog C3 ",Local_corr$DatesAnalogs[2]))
#'
#' # Example 10: Large_dist, Local_dist, and Local_cor return list FALSE different variable
#' expL1 <- rnorm(1:20)
#' dim(expL1) <- c(lat = 4, lon = 5)
#' obsL1 <- c(rnorm(1:180),expL1*5)
#' dim(obsL1) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL1 <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' expVar1 <- expL1[1:3,1:3]
#' dim(expVar1) <- c(lat = 3, lon = 3)
#' obsVar1 <- obsL1[1:3,1:3,1:10]
#' dim(obsVar1) <- c(lat = 3, lon = 3, time = 10)
#' # analogs of large scale using criteria 1
#' Large_scale <- Analogs(expL=expL1,
#' obsL=obsL1, time_obsL=time_obsL1,expVar=expVar1,obsVar=obsVar1,
#' criteria="Large_dist",
#' nAnalogs = 10, return_list = TRUE)
#' # analogs of local scale using criteria 2
#' lonmin=-1
#' lonmax=2
#' latmin=30
#' latmax=33
#' region=c(lonmin,lonmax,latmin,latmax)
#' Local_scale <- Analogs(expL=expL1,
#' obsL=obsL1, time_obsL=time_obsL1,obsVar=obsVar1,expVar=expVar1,
#' criteria="Local_dist",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 10, return_list = TRUE)
#' # analogs of local scale using criteria 3 and another variable so different obsL, expL, obsVar and expVar
#' expL2 <- rnorm(1:20)
#' dim(expL2) <- c(lat = 4, lon = 5)
#' obsL2 <- c(rnorm(1:180),expL2*5)
#' dim(obsL2) <- c(lat = 4, lon = 5, time = 10)
#' time_obsL2 <- paste(rep("01", 10), rep("01", 10), 1994 : 2003, sep = "-")
#' expVar2 <- expL2[1:3,1:3]
#' dim(expVar2) <- c(lat = 3, lon = 3)
#' obsVar2 <- obsL2[1:3,1:3,1:10]
#' dim(obsVar2) <- c(lat = 3, lon = 3, time = 10)
#' Local_corr <- Analogs(expL=expL2,
#' obsL=obsL2, time_obsL=time_obsL2,obsVar=obsVar2,expVar=expVar2,
#' criteria="Local_cor",lonVar=seq(-1,5,1.5),
#' latVar=seq(30,35,1.5),region=region,
#' nAnalogs = 10, return_list = TRUE)
#' plot.new()
#' layout(matrix(1:9,3,3,byrow=T))
#' image(expL1,main="expL Var1")
#' image(Large_scale$AnalogsFields[,,1],
#' main=paste0("BestAn. Var1 C1 ",Large_scale$DatesAnalogs[1]))
#' image(Large_scale$AnalogsFields[,,2],
#' main=paste0("BestAn. Var1 C1 ",Large_scale$DatesAnalogs[2]))
#' image(expVar1,main="expVar1")
#' image(Local_scale$AnalogsFields[,,1],
#' main=paste0("BestAn. Var1 C2 ",Local_scale$DatesAnalogs[1]))
#' image(Local_scale$AnalogsFields[,,2],
#' main=paste0("BestAn. Var1 C2 ",Local_scale$DatesAnalogs[2]))
#' image(expVar2,main="expVar2")
#' image(Local_corr$AnalogsFields[,,1],
#' main=paste0("BestAn. Var2 C3 ",Local_corr$DatesAnalogs[1]))
#' image(Local_corr$AnalogsFields[,,2],
#' main=paste0("2nd BestAn. Var2 C3 ",Local_corr$DatesAnalogs[2]))
#'
Analogs <- function(expL, obsL, time_obsL, expVar = NULL, obsVar = NULL,
criteria = "Large_dist",
lonVar = NULL, latVar = NULL, region = NULL,
nAnalogs = 1, return_list = FALSE) {
# checks
if (!all(c('lon', 'lat') %in% names(dim(expL)))) {
stop("Parameter 'expL' must have the dimensions 'lat' and 'lon'.")
}
if (!all(c('lat', 'lon') %in% names(dim(obsL)))) {
stop("Parameter 'obsL' must have the dimension 'lat' and 'lon'.")
}
if (any(is.na(expL))) {
warning("Parameter 'exp' contains NA values.")
}
if (any(is.na(obsL))) {
warning("Parameter 'obs' contains NA values.")
}
if (is.null(expVar) & !is.null(obsVar)) {
obsVar <- NULL
warning("Parameter 'obsVar' is set to NULL as parameter 'expVar'.")
}
if (!is.null(expVar) & is.null(obsVar)) {
expVar <- NULL
warning("Parameter 'expVar' is set to NULL as parameter 'obsVar'.")
}
if (any(names(dim(obsL)) %in% 'ftime')) {
if (any(names(dim(obsL)) %in% 'time')) {
stop("Multiple temporal dimensions ('ftime' and 'time') found",
"in parameter 'obsL'.")
} else {
time_pos_obsL <- which(names(dim(obsL)) == 'ftime')
names(dim(obsL))[time_pos_obsL] <- 'time'
if (any(names(dim(expL)) %in% 'ftime')) {
time_pos_expL <- which(names(dim(expL)) == 'ftime')
names(dim(expL))[time_pos_expL] <- 'time'
}
}
}
if (any(names(dim(obsVar)) %in% 'ftime')) {
if (any(names(dim(obsVar)) %in% 'time')) {
stop("Multiple temporal dimensions ('ftime' and 'time') found",
"in parameter 'obsVar'.")
} else {
time_pos_obsVar <- which(names(dim(obsVar)) == 'ftime')
names(dim(obsVar))[time_pos_obsVar] <- 'time'
if (any(names(dim(expVar)) %in% 'ftime')) {
time_pos_expVar <- which(names(dim(expVar)) == 'ftime')
names(dim(expVar))[time_pos_expVar] <- 'time'
}
}
}
if (any(names(dim(obsL)) %in% 'sdate')) {
if (any(names(dim(obsL)) %in% 'time')) {
dims_obsL <- dim(obsL)
pos_sdate <- which(names(dim(obsL)) == 'sdate')
pos_time <- which(names(dim(obsL)) == 'time')
pos <- 1 : length(dim(obsL))
pos <- c(pos_time, pos_sdate, pos[-c(pos_sdate,pos_time)])
obsL <- aperm(obsL, pos)
dim(obsL) <- c(time = prod(dims_obsL[c(pos_time, pos_sdate)]),
dims_obsL[-c(pos_time, pos_sdate)])
} else {
stop("Parameter 'obsL' must have a temporal dimension.")
}
}
if (any(names(dim(obsVar)) %in% 'sdate')) {
if (any(names(dim(obsVar)) %in% 'time')) {
dims_obsVar <- dim(obsVar)
pos_sdate <- which(names(dim(obsVar)) == 'sdate')
pos_time <- which(names(dim(obsVar)) == 'time')
pos <- 1 : length(dim(obsVar))
pos <- c(pos_time, pos_sdate, pos[-c(pos_sdate,pos_time)])
obsVar <- aperm(obsVar, pos)
dim(obsVar) <- c(time = prod(dims_obsVar[c(pos_time, pos_sdate)]),
dims_obsVar[-c(pos_time, pos_sdate)])
} else {
stop("Parameter 'obsVar' must have a temporal dimension.")
}
}
if (is.null(region)) {
if (!is.null(lonVar) & !is.null(latVar)) {
region <- c(min(lonVar), max(lonVar), min(latVar), max(latVar))
}
}
position <- Select(expL = expL, obsL = obsL, expVar = expVar, obsVar = obsVar,
criteria = criteria, lonVar = lonVar, latVar = latVar,
region = region)$position
best <- Apply(list(position), target_dims = c('time', 'pos'), fun = BestAnalog,
criteria = criteria,
return_list = return_list, nAnalogs = nAnalogs)$output1
Analogs_dates <- time_obsL[best]
dim(Analogs_dates) <- dim(best)
if (all(!is.null(region), !is.null(lonVar), !is.null(latVar))) {
if (is.null(obsVar)) {
obsVar <- SelBox(obsL, lon = lonVar, lat = latVar, region = region)
Analogs_fields <- Subset(obsVar$data, along = which(names(dim(obsVar)) == 'time'),
#obsVar <- SelBox(obsL, lon = lonVar, lat = latVar, region = region)
Analogs_fields <- Subset(obsVar, along = which(names(dim(obsVar)) == 'time'),
indices = best)
}
warning("One or more of the parameter 'region', 'lonVar' and 'latVar'",
" are NULL and the large scale field will be returned.")
if (is.null(obsVar)) {
Analogs_fields <- Subset(obsL, along = which(names(dim(obsL)) == 'time'),
indices = best)
} else {
Analogs_fields <- Subset(obsVar,
along = which(names(dim(obsVar)) == 'time'),
indices = best)
}
lon_dim <- which(names(dim(Analogs_fields)) == 'lon')
lat_dim <- which(names(dim(Analogs_fields)) == 'lat')
if (lon_dim < lat_dim) {
dim(Analogs_fields) <- c(dim(Analogs_fields)[c(lon_dim, lat_dim)], dim(best))
} else if (lon_dim > lat_dim) {
dim(Analogs_fields) <- c(dim(Analogs_fields)[c(lat_dim, lon_dim)], dim(best))
stop("Dimensions 'lat' and 'lon' not found.")
}
return(list(DatesAnalogs = Analogs_dates, AnalogsFields = Analogs_fields))
BestAnalog <- function(position, criteria = 'Large_dist', return_list = FALSE,
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
nAnalogs = 10) {
pos_dim <- which(names(dim(position)) == 'pos')
if (dim(position)[pos_dim] == 1) {
pos1 <- position
if (criteria != 'Large_dist') {
warning("Dimension 'pos' in parameter 'position' has length 1,",
" criteria 'Large_dist' will be used.")
criteria <- 'Large_dist'
}
} else if (dim(position)[pos_dim] == 2) {
pos1 <- Subset(position, along = pos_dim, indices = 1)
pos2 <- Subset(position, along = pos_dim, indices = 2)
if (criteria == 'Local_cor') {
warning("Dimension 'pos' in parameter 'position' has length 2,",
" criteria 'Local_dist' will be used.")
criteria <- 'Local_dist'
}
} else if (dim(position)[pos_dim] == 3) {
pos1 <- Subset(position, along = pos_dim, indices = 1)
pos2 <- Subset(position, along = pos_dim, indices = 2)
pos3 <- Subset(position, along = pos_dim, indices = 3)
if (criteria != 'Local_cor') {
warning("Parameter 'criteria' is set to", criteria, ".")
}
} else {
stop("Parameter 'position' has dimension 'pos' of different ",
"length than expected (from 1 to 3).")
}
if (criteria == 'Large_dist') {
if (return_list == FALSE) {
pos <- pos1[1]
pos1 <- pos1[1 : nAnalogs]
pos2 <- pos2[1 : nAnalogs]
if(length(best)==1){
warning("Just 1 best analog matching Large_dist and ",
"Local_dist criteria")
}
if(length(best)==1 & is.na(best[1])==TRUE){
stop("no best analogs matching Large_dist and Local_dist criterias")
}
pos <- pos1[as.logical(best)]
pos <- pos[which(!is.na(pos))]
if (return_list == FALSE) {
pos <- pos[1]
pos1 <- pos1[1 : nAnalogs]
pos2 <- pos2[1 : nAnalogs]
best <- match(pos1, pos2)
pos <- pos1[as.logical(best)]
pos <- pos[which(!is.na(pos))]
best <- match(pos, pos3)
pos <- pos[order(best, decreasing = F)]
pos <- pos[which(!is.na(pos))]
if (return_list == FALSE) {
pos[1]
}
return(pos)
Select <- function(expL, obsL, expVar = NULL, obsVar = NULL, criteria = "Large_dist",
lonVar = NULL, latVar = NULL, region = NULL) {
names(dim(expL)) <- replace_repeat_dimnames(names(dim(expL)), names(dim(obsL)))
metric1 <- Apply(list(obsL), target_dims = list(c('lat', 'lon')),
fun = .select, expL, metric = "dist")$output1
if (length(dim(metric1)) > 1) {
dim_time_obs <- which(names(dim(metric1)) == 'time' |
names(dim(metric1)) == 'ftime')
dim(metric1) <- c(dim(metric1), metric=1)
margins <- c(1 : (length(dim(metric1))))[-dim_time_obs]
pos1 <- apply(metric1, margins, order)
names(dim(pos1))[1] <- 'time'
metric1 <- apply(metric1, margins, sort)
names(dim(metric1))[1] <- 'time'
} else {
pos1 <- order(metric1)
dim(pos1) <- c(time = length(pos1))
metric1 <- sort(metric1)
dim(metric1) <- c(time = length(metric1))
dim(metric1) <- c(dim(metric1), metric = 1)
dim(pos1) <- c(dim(pos1), pos = 1)
return(list(metric = metric1, position = pos1))
}
if (criteria == "Local_dist" | criteria == "Local_cor") {
obs <- SelBox(obsL, lon = lonVar, lat = latVar, region = region)$data
exp <- SelBox(expL, lon = lonVar, lat = latVar, region = region)$data
metric2 <- Apply(list(obs), target_dims = list(c('lat', 'lon')),
fun = .select, exp, metric = "dist")$output1
dim(metric2) <- c(dim(metric2), metric=1)
margins <- c(1 : (length(dim(metric2))))[-dim_time_obs]
names(dim(pos2))[1] <- 'time'
metric2 <- apply(metric2, margins, sort)
names(dim(metric2))[1] <- 'time'
if (criteria == "Local_dist") {
metric <- abind(metric1, metric2, along = length(dim(metric1))+1)
position <- abind(pos1, pos2, along = length(dim(pos1))+1)
names(dim(metric)) <- c(names(dim(pos1)), 'metric')
names(dim(position)) <- c(names(dim(pos1)), 'pos')
return(list(metric = metric, position = position))
}
}
obs <- SelBox(obsL, lon = lonVar, lat = latVar, region = region)$data
exp <- SelBox(expL, lon = lonVar, lat = latVar, region = region)$data
metric3 <- Apply(list(obs), target_dims = list(c('lat', 'lon')),
fun = .select, exp, metric = "cor")$output1
dim(metric3) <- c(dim(metric3), metric=1)
margins <- c(1 : (length(dim(metric3))))[-dim_time_obs]
pos3 <- apply(metric3, margins, order, decreasing = TRUE)
names(dim(pos3))[1] <- 'time'
metric3 <- apply(metric3, margins, sort)
names(dim(metric3))[1] <- 'time'
metric <- abind(metric1, metric2, metric3, along = length(dim(metric1)) + 1)
position <- abind(pos1, pos2, pos3, along = length(dim(pos1)) + 1)
names(dim(metric)) <- c(names(dim(metric1)), 'metric')
names(dim(position)) <- c(names(dim(pos1)), 'pos')
return(list(metric = metric, position = position))
}
else {
stop("Parameter 'criteria' must to be one of the: 'Large_dist', ",
"'Local_dist','Local_cor'.")
}
}
.select <- function(exp, obs, metric = "dist") {
if (metric == "dist") {
result <- Apply(list(obs), target_dims = list(c('lat', 'lon')),
fun = function(x) {sum((x - exp) ^ 2)})$output1
} else if (metric == "cor") {
result <- Apply(list(obs), target_dims = list(c('lat', 'lon')),
fun = function(x) {cor(as.vector(x), as.vector(exp))})$output1
replace_repeat_dimnames <- function(names_exp, names_obs, lat_name = 'lat',
if (!is.character(names_exp)) {
stop("Parameter 'names_exp' must be a vector of characters.")
}
if (!is.character(names_obs)) {
stop("Parameter 'names_obs' must be a vector of characters.")
}
latlon_dim_exp <- which(names_exp == lat_name | names_exp == lon_name)
latlon_dim_obs <- which(names_obs == lat_name | names_obs == lon_name)
if (any(unlist(lapply(names_exp[-latlon_dim_exp],
function(x){x == names_obs[-latlon_dim_obs]})))) {
original_pos <- lapply(names_exp, function(x) which(x == names_obs[-latlon_dim_obs]))
original_pos <- lapply(original_pos, length) > 0
names_exp[original_pos] <- paste0(names_exp[original_pos], "_exp")
}
return(names_exp)
## Improvements: other dimensions to avoid replacement for more flexibility.
}