-
Notifications
You must be signed in to change notification settings - Fork 0
/
Fig 3a.R
325 lines (276 loc) · 16.6 KB
/
Fig 3a.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
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
library(readr)
library(deeptime)
library(ggplot2)
library(egg)
library(fANCOVA)
library(dplyr)
library(plyr)
library(reshape2)
# Set working directory (this will likely be the downloaded folder "Mo.U.Model")
setwd()
##==============================================
## NB - deeptime package required to reproduce these figures - if not yet installed need to use devtools to obtain from GitHub
## If not installed, comment out and run the lines of code directly below - install devtools if not installed already
##==============================================
# install.packages("devtools")
# devtools::install_github("willgearty/deeptime")
##==============================================
## Edit Deeptime timescale to fit 2012 geologic timescale and center labels wrt to x-axis
##==============================================
adapted_period_timescale <- deeptime::periods
adapted_period_timescale[10,2] <- 443.8
adapted_period_timescale[11,3] <- 443.8
adapted_period_timescale[10,3] <- 440.8
adapted_period_timescale[11,2] <- 447.5
adapted_stages_timescale <- deeptime::stages
adapted_stages_timescale[83,2] <- 443.8
adapted_stages_timescale[84,3] <- 443.8
adapted_stages_timescale[85,2] <- 447.5
##==============================================
## 3A - feux
##==============================================
round.factor <- 10
## Import Bartlett (with reconstructed age model)
load("Bartlett.full.RData")
## Omit any NA values
Bartlett <- filter(Bartlett, !is.na(U_iso))
## Fit LOESS regression to Bartlett
U.carb.fit <- loess.as(Bartlett$Age_est, Bartlett$U_iso, criterion = c("aicc", "gcv"))
## Predict time-dependent geochemical data from Bartlett data
U.carb.model <- predict(U.carb.fit, seq(444.7,443.2, -0.02))
## Generate rounding factor
U.carb.round <- (range(Bartlett$U_iso)[2]-range(Bartlett$U_iso)[1])/round.factor
## Import Murzuq Basin data (with age model)
load("Murzuq.RData")
## Fit cross-validated LOESS regression to Murzuq U and Mo
Murzuq <- filter(Murzuq, !is.na(d98Mo_auth_Al))
Murzuq <- filter(Murzuq, !is.na(dU238_auth_a))
Mo.shale.fit <- loess.as(Murzuq$Age_est, Murzuq$d98Mo_auth_Al, criterion = c("aicc"))
U.shale.fit <- loess.as(Murzuq$Age_est, Murzuq$dU238_auth_a, criterion = c("aicc"))
## Predict time-dependent geochemical data from Murzuq data
Mo.shale.model <- predict(Mo.shale.fit, seq(443.8,440.8, -0.02))
U.shale.model <- predict(U.shale.fit, seq(443.8,440.8, -0.02))
## Generate rounding factors
U.shale.round <- (range(Murzuq$dU238_auth_a)[2]-range(Murzuq$dU238_auth_a)[1])/round.factor
Mo.shale.round <- (range(Murzuq$d98Mo_auth_Al)[2]-range(Murzuq$d98Mo_auth_Al)[1])/round.factor
## Import combined shale and carbonate model
load("Mo-U.mass.balance.RData")
## Round model
mass.balance$dU.carb <- plyr::round_any(mass.balance$dU.carb, U.carb.round, f=round)
mass.balance$dMo <- plyr::round_any(mass.balance$dMo, Mo.shale.round, f=round)
mass.balance$dU <- plyr::round_any(mass.balance$dU, U.shale.round, f=round)
## Initiate data frame for summary
f.eux.sum <- data.frame(timebin = double(), mean.f.eux.value = double(),
median.f.eux.value = double(),
min.f.eux.value.05 = double(),
max.f.eux.value.95 = double(),
min.f.eux.value.10 = double(),
max.f.eux.value.90 = double(),
min.f.eux.value.20 = double(),
max.f.eux.value.80 = double(),
min.f.eux.value.30 = double(),
max.f.eux.value.70 = double(),
min.f.eux.value.40 = double(),
max.f.eux.value.60 = double())
## Round predictions to fit model rounding
## At 20kyr timesteps filter models based on LOESS models of geochemnical data for each percentile presented (plus some other statistics of interest)
# Bartlett data
for(timebin in seq(444.7, 443.2, -0.02)){
U.carb.value <- predict(U.carb.fit, timebin)
U.carb.value.rounded <- plyr::round_any(U.carb.value, U.carb.round, f=round)
f.eux.values <- filter(mass.balance, dU.carb == U.carb.value.rounded)
mean.f.eux.value <- mean(f.eux.values$f.eux)
median.f.eux.value <- median(f.eux.values$f.eux)
min.f.eux.value.05 <- quantile(f.eux.values$f.eux, probs=0.05)
max.f.eux.value.95 <- quantile(f.eux.values$f.eux, probs=0.95)
min.f.eux.value.10 <- quantile(f.eux.values$f.eux, probs=0.1)
max.f.eux.value.90 <- quantile(f.eux.values$f.eux, probs=0.9)
min.f.eux.value.20 <- quantile(f.eux.values$f.eux, probs=0.2)
max.f.eux.value.80 <- quantile(f.eux.values$f.eux, probs=0.8)
min.f.eux.value.30 <- quantile(f.eux.values$f.eux, probs=0.3)
max.f.eux.value.70 <- quantile(f.eux.values$f.eux, probs=0.7)
min.f.eux.value.40 <- quantile(f.eux.values$f.eux, probs=0.4)
max.f.eux.value.60 <- quantile(f.eux.values$f.eux, probs=0.6)
f.eux.sum <- rbind(f.eux.sum, c(timebin, mean.f.eux.value,
median.f.eux.value,
min.f.eux.value.05,
max.f.eux.value.95,
min.f.eux.value.10,
max.f.eux.value.90,
min.f.eux.value.20,
max.f.eux.value.80,
min.f.eux.value.30,
max.f.eux.value.70,
min.f.eux.value.40,
max.f.eux.value.60))
}
# Murzuq Basin data
for(timebin in seq(443.8, 440.8, -0.02)){
U.shale.value <- predict(U.shale.fit, timebin)
U.shale.value.rounded <- plyr::round_any(U.shale.value, U.shale.round, f=round)
Mo.shale.value <- predict(Mo.shale.fit, timebin)
Mo.shale.value.rounded <- plyr::round_any(Mo.shale.value, Mo.shale.round, f=round)
f.eux.values <- filter(mass.balance, dU == U.shale.value.rounded
& dMo == Mo.shale.value.rounded)
mean.f.eux.value <- mean(f.eux.values$f.eux)
median.f.eux.value <- median(f.eux.values$f.eux)
min.f.eux.value.05 <- quantile(f.eux.values$f.eux, probs=0.05)
max.f.eux.value.95 <- quantile(f.eux.values$f.eux, probs=0.95)
min.f.eux.value.10 <- quantile(f.eux.values$f.eux, probs=0.1)
max.f.eux.value.90 <- quantile(f.eux.values$f.eux, probs=0.9)
min.f.eux.value.20 <- quantile(f.eux.values$f.eux, probs=0.2)
max.f.eux.value.80 <- quantile(f.eux.values$f.eux, probs=0.8)
min.f.eux.value.30 <- quantile(f.eux.values$f.eux, probs=0.3)
max.f.eux.value.70 <- quantile(f.eux.values$f.eux, probs=0.7)
min.f.eux.value.40 <- quantile(f.eux.values$f.eux, probs=0.4)
max.f.eux.value.60 <- quantile(f.eux.values$f.eux, probs=0.6)
f.eux.sum <- rbind(f.eux.sum, c(timebin, mean.f.eux.value,
median.f.eux.value,
min.f.eux.value.05,
max.f.eux.value.95,
min.f.eux.value.10,
max.f.eux.value.90,
min.f.eux.value.20,
max.f.eux.value.80,
min.f.eux.value.30,
max.f.eux.value.70,
min.f.eux.value.40,
max.f.eux.value.60))
}
names(f.eux.sum) <- c("timebin", "mean.f.eux.value",
"median.f.eux.value",
"min.f.eux.value.05",
"max.f.eux.value.95",
"min.f.eux.value.10",
"max.f.eux.value.90",
"min.f.eux.value.20",
"max.f.eux.value.80",
"min.f.eux.value.30",
"max.f.eux.value.70",
"min.f.eux.value.40",
"max.f.eux.value.60")
f.eux.sum <- filter(f.eux.sum, !is.na(mean.f.eux.value))
# Fit cross-validated loess models to percentiles plotted (plus mean, median and confidence intervals for interest)
mean.f.eux.fit <- loess.as(f.eux.sum$timebin, f.eux.sum$mean.f.eux, criterion = c("aicc"))
median.f.eux.fit <- loess.as(f.eux.sum$timebin, f.eux.sum$median.f.eux, criterion = c("aicc"))
min.f.eux.fit.05 <- loess.as(f.eux.sum$timebin, f.eux.sum$min.f.eux.value.05, criterion = c("aicc"))
max.f.eux.fit.95 <- loess.as(f.eux.sum$timebin, f.eux.sum$max.f.eux.value.95, criterion = c("aicc"))
min.f.eux.fit.10 <- loess.as(f.eux.sum$timebin, f.eux.sum$min.f.eux.value.10, criterion = c("aicc"))
max.f.eux.fit.90 <- loess.as(f.eux.sum$timebin, f.eux.sum$max.f.eux.value.90, criterion = c("aicc"))
min.f.eux.fit.20 <- loess.as(f.eux.sum$timebin, f.eux.sum$min.f.eux.value.20, criterion = c("aicc"))
max.f.eux.fit.80 <- loess.as(f.eux.sum$timebin, f.eux.sum$max.f.eux.value.80, criterion = c("aicc"))
min.f.eux.fit.30 <- loess.as(f.eux.sum$timebin, f.eux.sum$min.f.eux.value.30, criterion = c("aicc"))
max.f.eux.fit.70 <- loess.as(f.eux.sum$timebin, f.eux.sum$max.f.eux.value.70, criterion = c("aicc"))
min.f.eux.fit.40 <- loess.as(f.eux.sum$timebin, f.eux.sum$min.f.eux.value.40, criterion = c("aicc"))
max.f.eux.fit.60 <- loess.as(f.eux.sum$timebin, f.eux.sum$max.f.eux.value.60, criterion = c("aicc"))
# Generate time dependent f.eux model based on loess models above
mean.f.eux.model <- predict(mean.f.eux.fit, seq(444.7,440.8, -0.02))
median.f.eux.model <- predict(median.f.eux.fit, seq(444.7,440.8, -0.02))
min.f.eux.model.05 <- predict(min.f.eux.fit.05, seq(444.7,440.8, -0.02))
max.f.eux.model.95 <- predict(max.f.eux.fit.95, seq(444.7,440.8, -0.02))
min.f.eux.model.10 <- predict(min.f.eux.fit.10, seq(444.7,440.8, -0.02))
max.f.eux.model.90 <- predict(max.f.eux.fit.90, seq(444.7,440.8, -0.02))
min.f.eux.model.20 <- predict(min.f.eux.fit.20, seq(444.7,440.8, -0.02))
max.f.eux.model.80 <- predict(max.f.eux.fit.80, seq(444.7,440.8, -0.02))
min.f.eux.model.30 <- predict(min.f.eux.fit.30, seq(444.7,440.8, -0.02))
max.f.eux.model.70 <- predict(max.f.eux.fit.70, seq(444.7,440.8, -0.02))
min.f.eux.model.40 <- predict(min.f.eux.fit.40, seq(444.7,440.8, -0.02))
max.f.eux.model.60 <- predict(max.f.eux.fit.60, seq(444.7,440.8, -0.02))
# summary data frame (for viewing only)
f.eux.sum.model <- as.data.frame(cbind(seq(444.7,440.8, -0.02), mean.f.eux.model,
median.f.eux.model,
min.f.eux.model.05,
max.f.eux.model.95,
min.f.eux.model.10,
max.f.eux.model.90,
min.f.eux.model.20,
max.f.eux.model.80,
min.f.eux.model.30,
max.f.eux.model.70,
min.f.eux.model.40,
max.f.eux.model.60))
# summary of percentile bracket minima (for plotting as envelope minima)
f.eux.sum.model.mins <- as.data.frame(cbind(seq(444.7,440.8, -0.02),
min.f.eux.model.05,
min.f.eux.model.10,
min.f.eux.model.20,
min.f.eux.model.30,
min.f.eux.model.40))
# summary of percentile bracket maxima (for plotting as envelope maxima)
f.eux.sum.model.maxs <- as.data.frame(cbind(seq(444.7,440.8, -0.02),
max.f.eux.model.95,
max.f.eux.model.90,
max.f.eux.model.80,
max.f.eux.model.70,
max.f.eux.model.60))
# rename timebins for plotting
names(f.eux.sum.model)[1] <- "timebin"
names(f.eux.sum.model.mins)[1] <- "timebin"
names(f.eux.sum.model.maxs)[1] <- "timebin"
# melt dataframe for plottimng
f.eux.sum.model.melted <- reshape2::melt(f.eux.sum.model, id.vars="timebin")
f.eux.sum.model.mins.melted <- reshape2::melt(f.eux.sum.model.mins, id.vars="timebin")
f.eux.sum.model.maxs.melted <- reshape2::melt(f.eux.sum.model.maxs, id.vars="timebin")
#annotate melted dataframe for generating fill scale when plotted
f.eux.sum.model.envelopes <- cbind(f.eux.sum.model.mins.melted,f.eux.sum.model.maxs.melted,
as.factor(c(rep("5", nrow(f.eux.sum.model.mins)), rep("10", nrow(f.eux.sum.model.mins)), rep("20", nrow(f.eux.sum.model.mins)),
rep("30", nrow(f.eux.sum.model.mins)), rep("40", nrow(f.eux.sum.model.mins)))))
names(f.eux.sum.model.envelopes)<- c("timebin", "Min.model", "Min", "timebin2", "Max.model", "Max", "low.bound")
# Relevel for plot fill scale
f.eux.sum.model.envelopes$low.bound <- relevel(f.eux.sum.model.envelopes$low.bound, 5)
f.eux.sum.model.envelopes$low.bound <- relevel(f.eux.sum.model.envelopes$low.bound, 10)
f.eux.sum.model.envelopes$low.bound <- relevel(f.eux.sum.model.envelopes$low.bound, 20)
f.eux.sum.model.envelopes$low.bound <- relevel(f.eux.sum.model.envelopes$low.bound, 30)
f.eux.sum.model.envelopes$low.bound <- relevel(f.eux.sum.model.envelopes$low.bound, 40)
f.eux.sum.model.envelopes <- filter(f.eux.sum.model.envelopes, !(is.na(Min) | is.na(Max)))
# plot time-dependent mass balance confidence envelopes
eux <- ggplot(f.eux.sum.model.envelopes, aes(x=timebin, ymax=Max*100, ymin=Min*100))+
annotate(geom="rect", xmin=-Inf, xmax=Inf, ymin=0.11, ymax=0.3, fill="grey95", color="grey75")+
annotate(geom="text", y=.15, x=441.52, label=expression("Modern f"[eux]), size=6)+
geom_ribbon(aes(fill=low.bound),alpha=.8)+
scale_y_log10(breaks = c(0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9,1.0,2,3,4,5,6,7,8,9,10,20,30,40,50,60,70,80,90,100),
labels = c("0.1", rep("",8),"1.0", rep("",8),"10", rep("",8), "100"))+
theme_minimal()+scale_x_reverse()+
ylab(expression("f"[eux]*" (%)"))+
xlab("Time (Ma)")+
theme(plot.margin = margin(.1,1,.1,1,"cm"),panel.border = element_rect(fill=NA,color="black", size=2,linetype="solid"),
axis.ticks = element_line(size=1, color="black"),
axis.line = element_line(lineend = 'square'),
axis.title = element_text(size=34),
axis.text = element_text(size=26, color="black"),
legend.title = element_text(size=24),
legend.text = element_text( size=18),
axis.ticks.length = unit(5, "points"),
legend.justification=c(1,1), legend.position=c(.2,.58),
legend.key = element_rect(color="grey20", size=1.5),
legend.background = element_rect(fill=alpha('white', 1),color=alpha('white', 1)),
panel.grid.major = element_blank(),panel.grid.minor = element_blank())+
scale_fill_brewer(palette= "YlGnBu", direction=-1, "Percentile", labels=c(expression("5 to 95"^"th"), expression("10 to 90"^"th"), expression("20 to 80"^"th"),expression("30 to 70"^"th"), expression("40 to 60"^"th")) )+
annotate(geom="text", label="A", x=447.25, y=60, size=10)+
coord_geo(xlim=c(447.5,440.8), ylim=c(.03,130),expand=FALSE, # Geologic timescale added for clarity
pos = as.list(rep("bottom", 2)),
abbrv=list(FALSE, FALSE),
dat = list(adapted_stages_timescale, adapted_period_timescale),
height = list(unit(2, "lines"), unit(2, "lines")),
size=list(8,8),
bord=list(c("left", "bottom", "right"),c("left", "bottom", "right")), lwd=as.list(c(1,1)))
# plot temporal extent of geochemical data
top <- ggplot(f.eux.sum.model.envelopes)+
annotate(geom="rect", ymin=2.5, ymax=4, xmax=444.7, xmin=443.2, fill="skyblue", colour="grey60")+
annotate(geom="rect", ymin=0.5, ymax=2, xmax=443.8, xmin=440.8, fill="goldenrod2", colour="grey60")+
annotate(geom="text", label=expression(delta^238*"U"[carb]), x=445.1, y=3.25, size=4.5)+
annotate(geom="text", label=expression(delta^238*"U"[eux]*" + "*delta^98*"Mo"[eux]), x=444.78, y=1.25, size=4.5)+
theme_minimal()+scale_x_reverse()+
coord_cartesian(xlim=c(447.5,440.8),ylim= c(0,4.5), expand = c(0,0))+
theme(plot.margin = margin(1,1,.1,1,"cm"),panel.border = element_rect(fill=NA,color=NA),
axis.ticks = element_blank(),
axis.line = element_blank(),
axis.title = element_blank(),
axis.text = element_blank(),
panel.grid.major = element_blank(),panel.grid.minor = element_blank())
# rerun if "polygon edge not found" error is returned
fig3a <- ggarrange2(top, eux, ncol=1, heights=c(.17,1), margin = unit(0.1, "line"))
##==============================================
## Save as pdf
##==============================================
ggsave("Figure 3a.pdf",fig3a, width = 11, height = 7)