-
Notifications
You must be signed in to change notification settings - Fork 0
/
validation_tmn.R
237 lines (196 loc) · 10 KB
/
validation_tmn.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
#--------------------------------
# Name: validation_tmn.R
# Purpose: Validate the tmn predictions using the NevCAN data
# Author: Andrew Vitale vitale232@gmail.com
# Created 2014/11/13
# R: 3.1.2
#--------------------------------
library(raster)
setwd('~/Dropbox/UNR/UNR-Thesis/Data/')
#### Define functions
mae = function(preds, obs, ...){
mean(abs(preds - obs), ...)
}
rmse = function(preds, obs, ...){
err = preds - obs
sqrt(mean(err^2, ...))
}
TP = function(x, y, ...){
t = t.test(x, y, ...)$statistic
p = t.test(x, y, ...)$p.value
out = c(t, p)
names(out) = c('t', 'p')
out
}
#### Read in the prediction maps for Tmn from cross16 preds
lf = list.files(file.path(getwd(), 'Temperature-Maps', 'Tmn', 'tmn_mod'),
pattern='.tif$', full.names=TRUE)
tmn_stack = stack(lf)
#### Read in the NevCAN data that's been cleaned up in Python
validation = read.csv('./Validation/validation_nevcan-daily.csv')
names(validation)[1] = 'date'
validation$date = as.Date(validation$date)
#### Read in the station coordinates
stations = read.csv('./Validation/stations.csv')
stations$long = stations$long * -1
coordinates(stations) = ~long+lat
proj4string(stations) = CRS('+proj=longlat +datum=WGS84 +ellps=WGS84 +towgs84=0,0,0')
stations = spTransform(stations, CRS(projection(tmn_stack)))
#### Read in the road shapefile for mapping
roads = readOGR(dsn='/home/vitale232/Dropbox/UNR/UNR-Thesis/SiteMap', layer='DirtRoads')
#### Extract the predicted Tmn values
preds = extract(tmn_stack, stations)
preds = t(preds)
preds = as.data.frame(preds)
names(preds) = stations$station
preds$date = as.Date(row.names(preds), format='X%Y.%m.%d_tmn')
#### Make a data.frame from validation and preds
df = merge(validation[ , c('date', 'sage_west_tmn', 'pj_west_tmn',
'montane_west_tmn', 'subalpine_west_tmn')],
preds[ , c('date', 'sagebrush_west', 'pj_west',
'montane_west', 'subalpine_west')])
df$sage = df$sagebrush_west - df$sage_west_tmn
df$pj = df$pj_west - df$sage_west_tmn
df$montane = df$montane_west - df$montane_west_tmn
df$subalpine = df$subalpine_west - df$subalpine_west_tmn
tp = matrix(rep(NA, 8), ncol=2)
rownames(tp) = c('sage', 'pj', 'montane', 'subalpine')
colnames(tp) = c('t', 'p')
tp['sage', c('t' ,'p')] = TP(df$sage_west_tmn, df$sagebrush_west)
tp['pj', c('t' ,'p')] = TP(df$pj_west_tmn, df$pj_west)
tp['montane', c('t' ,'p')] = TP(df$montane_west_tmn, df$montane_west)
tp['subalpine', c('t' ,'p')] = TP(df$subalpine_west_tmn, df$subalpine_west)
m = matrix(rep(NA, 15), ncol=3)
rownames(m) = c('sage', 'pj', 'montane', 'subalpine', 'overall')
colnames(m) = c('MAE', 'RMSE', 'R2')
m['sage', 'MAE'] = mae(df$sagebrush_west, df$sage_west_tmn, na.rm=TRUE)
m['pj', 'MAE'] = mae(df$pj_west, df$pj_west_tmn, na.rm=TRUE)
m['montane', 'MAE'] = mae(df$montane_west, df$montane_west_tmn, na.rm=TRUE)
m['subalpine', 'MAE'] = mae(df$subalpine_west, df$subalpine_west_tmn, na.rm=TRUE)
m['overall', 'MAE'] = with(df, {
mae(c(sagebrush_west, pj_west, montane_west, subalpine_west),
c(sage_west_tmn, pj_west_tmn, montane_west_tmn, subalpine_west_tmn),
na.rm=TRUE)})
m['sage', 'RMSE'] = rmse(df$sagebrush_west, df$sage_west_tmn, na.rm=TRUE)
m['pj', 'RMSE'] = rmse(df$pj_west, df$pj_west_tmn, na.rm=TRUE)
m['montane', 'RMSE'] = rmse(df$montane_west, df$montane_west_tmn, na.rm=TRUE)
m['subalpine', 'RMSE'] = rmse(df$subalpine_west, df$subalpine_west_tmn, na.rm=TRUE)
m['overall', 'RMSE'] = with(df, {
rmse(c(sagebrush_west, pj_west, montane_west, subalpine_west),
c(sage_west_tmn, pj_west_tmn, montane_west_tmn, subalpine_west_tmn),
na.rm=TRUE)})
m['sage', 'R2'] = cor(df$sagebrush_west, df$sage_west_tmn, use='complete.obs')
m['pj', 'R2'] = cor(df$pj_west, df$pj_west_tmn, use='complete.obs')
m['montane', 'R2'] = cor(df$montane_west, df$montane_west_tmn, use='complete.obs')
m['subalpine', 'R2'] = cor(df$subalpine_west, df$subalpine_west_tmn, use='complete.obs')
m['overall', 'R2'] = with(df, {
cor(c(sagebrush_west, pj_west, montane_west, subalpine_west),
c(sage_west_tmn, pj_west_tmn, montane_west_tmn, subalpine_west_tmn),
use='complete.obs')})
write.csv(m, './Validation/accuracy-table_tmn_mod.csv')
write.csv(tp, './Validation/t-test-table_tmn_mod-tmn.csv')
#### Calculate bias for table 5
tmn_by_month_raw = df[, c('date', 'sage', 'pj', 'montane', 'subalpine')]
tmn_by_month_raw$month = factor(format(tmn_by_month_raw$date, '%b'),
levels=c('Jun', 'Jul', 'Aug', 'Sep',
'Oct', 'Nov', 'Dec', 'Jan',
'Feb', 'Mar', 'Apr', 'May'))
tmn_by_month = aggregate(. ~ month,
data=tmn_by_month_raw[ , -which(names(tmn_by_month_raw) == 'date')],
mean, na.rm=TRUE)
tmn_by_month$month = as.character(tmn_by_month$month)
#### Add row for overall and take the mean for each site
tmn_by_month[13, 'month'] = 'Overall'
tmn_by_month[13, 'sage'] = mean(df$sage, na.rm=TRUE)
tmn_by_month[13, 'pj'] = mean(df$pj, na.rm=TRUE)
tmn_by_month[13, 'montane'] = mean(df$montane, na.rm=TRUE)
tmn_by_month[13, 'subalpine'] = mean(df$subalpine, na.rm=TRUE)
#### add a column for monthly overalls and calculate overall overall
tmn_by_month_raw$overall = apply(tmn_by_month_raw[ , c('sage', 'pj', 'montane', 'subalpine')], 1,
mean, na.rm=TRUE)
tmn_by_month$overall = c(aggregate(overall ~ month, tmn_by_month_raw, mean, na.rm=TRUE)$overall, NA)
tmn_by_month[which(tmn_by_month$month == 'Overall'), 'overall'] = with(tmn_by_month_raw, {
mean(c(sage, pj, montane, subalpine), na.rm=TRUE)})
row.names(tmn_by_month) = tmn_by_month$month
tmn_out = t(tmn_by_month[, -which(names(tmn_by_month) == 'month')])
row.names(tmn_out) = c('Sage', 'PJ', 'Montane', 'Subalpine', 'Overall')
write.csv(signif(tmn_out, 3),
'./Tables/Tmn/overall-bias.csv')
#### Caclulate the MAE by month and by site, including overalls for table 6 ####
tmn_mae = df
tmn_mae$month = format(tmn_mae$date, '%b')
months = c('Jun', 'Jul', 'Aug', 'Sep',
'Oct', 'Nov', 'Dec', 'Jan',
'Feb', 'Mar', 'Apr', 'May')
tmn_mae_out = as.data.frame(matrix(nrow=5, ncol=14))
colnames(tmn_mae_out) = c('Site', months, 'Overall')
tmn_mae_out$Site = c('Sage', 'PJ', 'Montane',
'Subalpine', 'Overall')
for(i in 1:length(months)){
tmn_mae_out[tmn_mae_out$Site == 'Sage', months[i]] = with(tmn_mae[tmn_mae$month == months[i], ], {
mae(sagebrush_west, sage_west_tmn, na.rm=TRUE)})
tmn_mae_out[tmn_mae_out$Site == 'PJ', months[i]] = with(tmn_mae[tmn_mae$month == months[i], ], {
mae(pj_west, pj_west_tmn, na.rm=TRUE)})
tmn_mae_out[tmn_mae_out$Site == 'Montane', months[i]] = with(tmn_mae[tmn_mae$month == months[i], ], {
mae(montane_west, montane_west_tmn, na.rm=TRUE)})
tmn_mae_out[tmn_mae_out$Site == 'Subalpine', months[i]] = with(tmn_mae[tmn_mae$month == months[i], ], {
mae(subalpine_west, subalpine_west_tmn, na.rm=TRUE)})
tmn_mae_out[tmn_mae_out$Site == 'Overall', months[i]] = with(tmn_mae[tmn_mae$month == months[i], ], {
mae(c(sagebrush_west, pj_west, montane_west, subalpine_west),
c(sage_west_tmn, pj_west_tmn, montane_west_tmn, subalpine_west_tmn),
na.rm=TRUE)})
}
tmn_mae_out$Overall = m[ , 'MAE']
write.csv(tmn_mae_out, './Tables/Tmn/MAE_by-site_by-month.csv')
#### Calculate RMSE table 7 for paper
tmn_rmse = df
tmn_rmse$month = format(tmn_rmse$date, '%b')
months = c('Jun', 'Jul', 'Aug', 'Sep',
'Oct', 'Nov', 'Dec', 'Jan',
'Feb', 'Mar', 'Apr', 'May')
tmn_rmse_out = as.data.frame(matrix(nrow=5, ncol=14))
colnames(tmn_rmse_out) = c('Site', months, 'Overall')
tmn_rmse_out$Site = c('Sage', 'PJ', 'Montane',
'Subalpine', 'Overall')
for(i in 1:length(months)){
tmn_rmse_out[tmn_rmse_out$Site == 'Sage', months[i]] = with(tmn_rmse[tmn_rmse$month == months[i], ], {
rmse(sagebrush_west, sage_west_tmn, na.rm=TRUE)})
tmn_rmse_out[tmn_rmse_out$Site == 'PJ', months[i]] = with(tmn_rmse[tmn_rmse$month == months[i], ], {
rmse(pj_west, pj_west_tmn, na.rm=TRUE)})
tmn_rmse_out[tmn_rmse_out$Site == 'Montane', months[i]] = with(tmn_rmse[tmn_rmse$month == months[i], ], {
rmse(montane_west, montane_west_tmn, na.rm=TRUE)})
tmn_rmse_out[tmn_rmse_out$Site == 'Subalpine', months[i]] = with(tmn_rmse[tmn_rmse$month == months[i], ], {
rmse(subalpine_west, subalpine_west_tmn, na.rm=TRUE)})
tmn_rmse_out[tmn_rmse_out$Site == 'Overall', months[i]] = with(tmn_rmse[tmn_rmse$month == months[i], ], {
rmse(c(sagebrush_west, pj_west, montane_west, subalpine_west),
c(sage_west_tmn, pj_west_tmn, montane_west_tmn, subalpine_west_tmn),
na.rm=TRUE)})
}
tmn_rmse_out$Overall = m[ , 'RMSE']
write.csv(tmn_rmse_out, './Tables/Tmn/RMSE_by-site_by-month.csv')
#
# plot(df$date, df$sage, type='l', ylim=c(-10, 20),
# xlab='Date', ylab='prediction - observation (°C)')
# lines(df$date, df$pj, col='orange')
# lines(df$date, df$montane, col='darkgreen')
# lines(df$date, df$subalpine, col='blue')
# abline(0, 0, lty=2)
# legend('topleft', legend=c('Sagebrush West', 'Pinyon-Juniper',
# 'Montane', 'Subalpine West'),
# col=c('black', 'orange', 'darkgreen', 'blue'),
# lty=1)
#
# setEPS(height=7, width=11)
# postscript('~/Dropbox/UNR/UNR-Thesis/Figures/validation_tmn_pred-obs.eps')
# plot(df$date, df$sage, ylim=c(-10, 20),
# xlab='Date', ylab='prediction - observation (°C)',
# cex.lab=1.25, cex.axis=1.25)
# points(df$date, df$pj, col='orange')
# points(df$date, df$montane, col='darkgreen')
# points(df$date, df$subalpine, col='blue')
# abline(0, 0, lty=2)
# legend('topleft', legend=c('Sagebrush West', 'Pinyon-Juniper',
# 'Montane', 'Subalpine West'),
# col=c('black', 'orange', 'darkgreen', 'blue'),
# pch=1)
# dev.off()