-
Notifications
You must be signed in to change notification settings - Fork 105
/
bias_correction_isimip3.R
206 lines (176 loc) · 11.3 KB
/
bias_correction_isimip3.R
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
# bias_correction_isimip3.R
#
# Copyright (C) 2021 Santander Meteorology Group (http://meteo.unican.es)
#
# This work is licensed under a Creative Commons Attribution 4.0 International
# License (CC BY 4.0 - http://creativecommons.org/licenses/by/4.0)
#' @title Script to bias-correct CMIP6 with ISIMIP3ISIMIP3
#' @description Script to bias-correct CMIP6 with the ISIMIP3 method. ISIMIP3 (Lange 2019, https://doi.org/10.5194/gmd-12-3055-2019)
#' is a parametric quantile mapping which has been designed to robustly adjust biases in all percentiles of a distribution whilst
#' preserving their trends. The observational reference used for calibration is W5E5 (Cucchi et al. 2020, https://doi.org/10.5194/essd-12-2097-2020),
#' which was previously conservatively remapped onto a 1ºx1º regular grid. Note that spatial chunking is required to alleviate computationally costly calculations.
#' @author S. Herrera
#' @author M. Iturbide
#' @author A. Casanueva
# GMS 2020. Script to bias-correct CMIP6 with isimip3
# S. Herrera, 25-07-2020. Prepare auxiliary function.
# M. Iturbide, 27-07-2020. First version.
# M. Iturbide, 28-08-2020. Allow lat-lon chunking and set number of chunks based on memory resources.
# A. Casanueva, 25-09-2020. Allow parse options in .sh
# A. Casanueva, 02-01-2021. Split test period
library(downscaleR)
library(loadeR)
library(loadeR.2nc)
library(climate4R.UDG)
# Source chunking function
source("https://raw.githubusercontent.com/SantanderMetGroup/climate4R/devel/R/climate4R.chunk.R")
# ***************************************
## Argument setting for the C4R function:
years.hist <- 1980:2005
#years.ssp <- 2015:2100
years.ssp <- 2015:2057 # 2058-2100
max.size <- 700 #Mb
memory.offset <- 360
# Select SSP
# ssp <- "ssp126"
# ssp <- "ssp245"
ssp <- "ssp585"
# ssp <- "ssp370"
message("Starting bias adjustment of CMIP6 for ", ssp, " with isimip3 at ", Sys.time())
out.dir <- paste0("/oceano/gmeteo/WORK/PROYECTOS/2018_IPCC/data/BA_DATA/CMIP6/temperatures/")
# ***************************************
# ***************************************
## Datasets
dataset.obs <- "/oceano/gmeteo/WORK/PROYECTOS/2018_IPCC/data/OBSERVATIONS/W5E5/deg1/w5e5_v1.0.ncml"
di.obs <- dataInventory(dataset.obs)
datasets.hist <- UDG.datasets("historical")[["CMIP6"]]
datasets.ssp <- UDG.datasets(ssp)[["CMIP6"]]
hist.members <- gsub("CMIP6_|historical_", "", datasets.hist)
fut.members <- gsub(paste0("CMIP6_|", ssp, "_"), "", datasets.ssp)
members <- intersect(hist.members, fut.members)
ind.h <- sapply(members, function(x) grep(x, hist.members))
ind.f <- sapply(members, function(x) grep(x, fut.members))
datasets.hist <- datasets.hist[ind.h]
datasets.ssp <- datasets.ssp[ind.f]
# ***************************************
# ***************************************
# aux.fun.isimip3 ------------------
aux.fun.isimip3 <- function(y.tas, y.tasmin, y.tasmax,
x.tas, x.tasmin, x.tasmax,
newdata.tas, newdata.tasmin, newdata.tasmax,
isimip3.args = list(lower_bound = c(NULL),
lower_threshold = c(NULL),
upper_bound = c(NULL),
upper_threshold = c(NULL),
randomization_seed = NULL,
detrend = array(data = TRUE, dim = 1),
rotation_matrices = c(NULL),
n_quantiles = 50,
distribution = c("normal"),
trend_preservation = array(data = "additive", dim = 1),
adjust_p_values = array(data = FALSE, dim = 1),
if_all_invalid_use = c(NULL),
invalid_value_warnings = FALSE),
isimip3.range.args = list(lower_bound = c(0),
lower_threshold = c(0.01),
upper_bound = c(NULL),
upper_threshold = c(NULL),
randomization_seed = NULL,
detrend = array(data = FALSE, dim = 1),
rotation_matrices = c(NULL),
n_quantiles = 50,
distribution = c("rice"),
trend_preservation = array(data = "mixed", dim=1),
adjust_p_values = array(data = FALSE, dim = 1),
if_all_invalid_use = c(NULL),
invalid_value_warnings = FALSE),
isimip3.skew.args = list(lower_bound = c(0),
lower_threshold = c(0.0001),
upper_bound = c(1),
upper_threshold = c(0.9999),
randomization_seed = NULL,
detrend = array(data = FALSE, dim = 1),
rotation_matrices = c(NULL),
n_quantiles = 50,
distribution = c("beta"),
trend_preservation = array(data = "bounded", dim = 1),
adjust_p_values = array(data = FALSE, dim = 1),
if_all_invalid_use = c(NULL),
invalid_value_warnings = FALSE)){
# Calculate range and skewness
y.range <- gridArithmetics(y.tasmax, y.tasmin, operator = c("-"))
x.range <- gridArithmetics(x.tasmax, x.tasmin, operator = c("-"))
newdata.range <- gridArithmetics(newdata.tasmax,newdata.tasmin, operator = c("-"))
y.skew <- gridArithmetics(gridArithmetics(y.tas, y.tasmin, operator = "-"), y.range, operator = "/")
x.skew <- gridArithmetics(gridArithmetics(x.tas,x.tasmin, operator = "-"), x.range, operator = "/")
newdata.skew <- gridArithmetics(gridArithmetics(newdata.tas, newdata.tasmin, operator = "-"), newdata.range, operator = "/")
attr.tasmin <- y.tasmin$Variable
attr.tasmax <- y.tasmax$Variable
y.tasmax <- NULL; y.tasmin <- NULL; x.tasmax <- NULL; x.tasmin <- NULL; newdata.tasmax <- NULL;newdata.tasmin <- NULL
# tas
message("Starting bias adjustment of mean temperature at ", Sys.time())
bc.tas.args <- list("y" = y.tas, "x" = x.tas, "newdata" = newdata.tas, "precipitation" = FALSE, "isimip3.args" = isimip3.args, "method"="isimip3")
bc.tas <- do.call("biasCorrection", bc.tas.args)
bc.tas.args <- NULL; y.tas <- NULL; x.tas <- NULL; newdata.tas <- NULL
# range
message("Starting bias adjustment of temperature range at ", Sys.time())
bc.range.args <- list("y" = y.range, "x" = x.range, "newdata" = newdata.range, "precipitation" = FALSE, "isimip3.args" = isimip3.range.args, "method"="isimip3")
bc.range <- do.call("biasCorrection", bc.range.args)
bc.range.args <- NULL; y.range <- NULL; x.range <- NULL; newdata.range <- NULL
# skewness
message("Starting bias adjustment of temperature skewness at ", Sys.time())
bc.skew.args <- list("y" = y.skew, "x" = x.skew, "newdata" = newdata.skew, "precipitation" = FALSE, "isimip3.args" = isimip3.skew.args, "method"="isimip3")
bc.skew <- do.call("biasCorrection", bc.skew.args)
bc.skew.args <- NULL; y.skew <- NULL; x.skew <- NULL; newdata.skew <- NULL
message("Calculating bias-adjusted minimum temperature at ", Sys.time())
bc.tasmin <- gridArithmetics(bc.tas, gridArithmetics(bc.range, bc.skew, operator = c("*")), operator = c("-"))
# put right attributes
bc.tasmin$Variable <- attr.tasmin
attr(bc.tasmin$Variable, "correction") <- "isimip3"
message("Calculating bias-adjusted maximum temperature at ", Sys.time())
bc.tasmax <- gridArithmetics(bc.tasmin, bc.range, operator = c("+"))
# put right attributes
bc.tasmax$Variable <- attr.tasmax
attr(bc.tasmax$Variable, "correction") <- "isimip3"
bc.range <- NULL; bc.skew <- NULL
makeMultiGrid(bc.tas, bc.tasmin, bc.tasmax)
}
# ***************************************
# ***************************************
# apply bias correction -----------------
models <- 1:length(datasets.hist)
message("Ready to start models:\n ",paste(datasets.ssp[models], collapse="\n "))
lapply(models, function(x) {
if(datasets.ssp[x]=="CMIP6_AWI-CM-1-1-MR_ssp585_r1i1p1f1"){
n.chunks <- 60; chunk.horiz <- FALSE
} else if(datasets.ssp[x]=="CMIP6_CNRM-CM6-1-HR_ssp585_r1i1p1f2"){
n.chunks <- 90; chunk.horiz <- FALSE
} else{ n.chunks <- 45; chunk.horiz <- FALSE}
if(!file.exists(paste0(out.dir,"/",datasets.ssp[x],"_chunk0",n.chunks,".nc"))){
message("Starting GCM ",datasets.ssp[x], " at ", Sys.time())
di <- dataInventory(datasets.hist[x])
di2 <- dataInventory(datasets.ssp[x])
if (any(names(di) %in% "tas") & any(names(di) %in% "tasmax") & any(names(di) %in% "tasmin")) {
if (any(names(di2) %in% "tas") & any(names(di2) %in% "tasmax") & any(names(di2) %in% "tasmin")) {
###COMPUTE BC:
index <- climate4R.chunk(n.chunks = n.chunks,
chunk.horizontally = chunk.horiz,
C4R.FUN.args = list(FUN = "aux.fun.isimip3",
y.tas = list(dataset = dataset.obs, var = "tas", years = years.hist),
y.tasmin = list(dataset = dataset.obs, var = "tasmin", years = years.hist),
y.tasmax = list(dataset = dataset.obs, var = "tasmax", years = years.hist),
x.tas = list(dataset = datasets.hist[x], var = "tas", years = years.hist),
x.tasmin = list(dataset = datasets.hist[x], var = "tasmin", years = years.hist),
x.tasmax = list(dataset = datasets.hist[x], var = "tasmax", years = years.hist),
newdata.tas = list(dataset = datasets.ssp[x], var = "tas", years = years.ssp),
newdata.tasmin = list(dataset = datasets.ssp[x], var = "tasmin", years = years.ssp),
newdata.tasmax = list(dataset = datasets.ssp[x], var = "tasmax", years = years.ssp)),
output.path = out.dir,
filename = paste0(datasets.ssp[x],'_', years.ssp[1], '-',years.ssp[length(years.ssp)]) )
index <- NULL
message("Finished GCM ",datasets.ssp[x], " at ", Sys.time())
} else{message("Variable missing in ", datasets.ssp[x])}
} else{message("Variable missing in ", datasets.hist[x])}
} else{message("Skipping GCM ",datasets.ssp[x], ", already available")}
})
# ***************************************