-
Notifications
You must be signed in to change notification settings - Fork 20
/
fetchHenry.R
459 lines (370 loc) · 16.5 KB
/
fetchHenry.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
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
# # TODO: finish this
# .summarizeSoilVWC <- function(soilVWC.data) {
#
# # TODO: strip-out bogus data
#
#
# d <- ddply(soilVWC.data, c('sid', 'year'), .progress='text', .fun = function(i) {
#
#
# i.q <- quantile(i$sensor_value, probs = c(0.05, 0.25, 0.5, 0.75, 0.95), na.rm = TRUE)
# days.gt.q.crit <- length(which(i$sensor_value >= i.q['50%']))
# days.of.data <- length(na.omit(i$sensor_value))
#
# # TODO: these summaries are only useful when you have an entire year of data...
# res <- data.frame(days.gt.q.crit, days.of.data)
# return(res)
# })
#
# return(d)
# }
# summarize daily values by Julian day
#' @param soiltemp.data A `data.frame` containing soil temperature data
#' @export
#' @rdname fetchHenry
summarizeSoilTemperature <- function(soiltemp.data) {
# hacks to make R CMD check --as-cran happy:
sensor_value <- NULL
V1 <- NULL
.SD <- NULL
season <- NULL
# determine number of complete years of data
# proceed with data.table aggregation / joins
dd <- as.data.table(soiltemp.data)
# days of real data / site / year
cr.1 <- dd[, sum(!is.na(sensor_value)), by = c('sid', 'year')]
# complete yrs of real data / site
cr.2 <- cr.1[, sum(V1 >= 365), by = 'sid']
names(cr.2)[2] <- 'complete.yrs'
# determine functional years of data
# number of complete years after accounting for overlap
.functionalYrs <- function(i) {
# convert current sensor's data to wide format, first row is the year
# note: when all data are NA, dcast will perform an aggregate
# take the first value (NA) in this case
w <- data.table::dcast(i, year ~ doy, value.var = 'sensor_value', fun.aggregate = function(i) i[1])
# on DOY 1-365, count total number of non-NA records over all years
non.na.doy <- apply(w[, 2:366], 2, function(j) length(na.omit(j)))
# the minimum value is the number of functional years
res <- data.frame(functional.yrs = min(non.na.doy))
return(res)
}
fy <- dd[, .functionalYrs(.SD), by = 'sid']
# compute summaries by DOY:
# n: number of non-NA records
# daily.mean: mean of non-NA values
.doySummary <- function(i) {
res <- data.frame(
year = i$year[1],
n.total = length(i$sensor_value),
n = length(na.omit(i$sensor_value)),
daily.mean = mean(i$sensor_value, na.rm = TRUE)
)
return(res)
}
d <- dd[, .doySummary(.SD), by = c('sid', 'doy')]
# convert DOY -> month
d$month <- format(as.Date(paste0(d$year, "-", d$doy), format = "%Y-%j"), "%b")
d$season <- month2season(d$month)
# compute unbiased MAST, number of obs, complete records per average no. days in year
.unbiasedMAST <- function(i) {
res <- data.frame(
gap.index = round(1 - (sum(i$n) / sum(i$n.total)), 2),
days.of.data = sum(i$n),
MAST = round(mean(i$daily.mean, na.rm = TRUE), 2)
)
return(res)
}
d.mast <- d[, .unbiasedMAST(.SD), by = 'sid']
# compute unbiased seasonal averages
.seasonalMeanTemp <- function(i) {
res <- data.frame(
seasonal.mean.temp = round(mean(i$daily.mean, na.rm = TRUE), 2)
)
return(res)
}
d.seasonal.long <- d[season %in% c('Winter', 'Summer'), .seasonalMeanTemp(.SD), by = c('season', 'sid')]
# convert seasonal avgs to wide format
d.season <- data.table::dcast(d.seasonal.long, sid ~ season, value.var = 'seasonal.mean.temp')
# combine columns
d.summary <- merge.data.table(d.mast, d.season, by = 'sid', all.x = TRUE, sort = FALSE)
d.summary <- merge.data.table(d.summary, cr.2, by = 'sid', all.x = TRUE, sort = FALSE)
d.summary <- merge.data.table(d.summary, fy, by = 'sid', all.x = TRUE, sort = FALSE)
# estimate STR
# note that gelic / cryic assignment is problematic when missing O horizon / saturation details
d.summary$STR <- estimateSTR(d.summary$MAST, d.summary$Summer, d.summary$Winter)
# downgrade to data.frame
d.summary <- as.data.frame(d.summary)
# re-shuffle columns and return
return(d.summary[, c('sid', 'days.of.data', 'gap.index', 'functional.yrs', 'complete.yrs', 'MAST', 'Winter', 'Summer', 'STR')])
}
#' @export
#' @param x character vector containing month abbreviation e.g. `c('Jun', 'Dec', 'Sep')`
#' @rdname fetchHenry
month2season <- function(x) {
season <- rep(NA, times = length(x))
season[x %in% c('Jun', 'Jul', 'Aug')] <- 'Summer'
season[x %in% c('Dec', 'Jan', 'Feb')] <- 'Winter'
season[x %in% c('Mar', 'Apr', 'May')] <- 'Spring'
season[x %in% c('Sep', 'Oct', 'Nov')] <- 'Fall'
# fix factor levels for season
season <- factor(season, levels = c('Winter', 'Spring', 'Summer', 'Fall'))
return(season)
}
## function for padding daily time-series with NA in the presence of missing days
## must be run on subsets defined by year
.fill_missing_days <- function(x) {
## TODO this doesn't account for leap-years
# ID missing days
missing.days <- which(is.na(match(1:365, x$doy)))
# short-circuit
if (length(missing.days) < 1) {
return(x)
}
# get constants
this.sid <- x$sid[1]
this.year <- x$year[1]
# make fake date-times for missing data
fake.datetimes <- paste0(this.year, ' ', missing.days, ' 00:00')
# TODO: this will result in timezone specific to locale;
# especially an issue when granularity is less than daily or for large extents
fake.datetimes <- as.POSIXct(as.Date(fake.datetimes, format = "%Y %j %H:%M"))
# generate DF with missing information
fake.data <- data.frame(
sid = this.sid,
date_time = fake.datetimes,
year = this.year,
doy = missing.days,
month = format(fake.datetimes, "%b")
)
fill.cols <- which(!colnames(x) %in% colnames(fake.data))
if (length(fill.cols) > 0) {
na.data <- as.data.frame(x)[, fill.cols, drop = FALSE][0,, drop = FALSE][1:nrow(fake.data),, drop = FALSE]
fake.data <- cbind(fake.data, na.data)
}
# make datatypes for time match
x$date_time <- as.POSIXct(as.Date(x$date_time, format = "%Y-%m-%d %H:%M:%S"))
# splice in missing data
y <- rbind(x, fake.data)
# re-order by DOY and return
return(y[order(y$doy), ])
}
# .formatDates
#
# @param sensor.data a data.frame containing columns `"sid"` and `"date_time"`
# @param gran granularity, common usage is `'day'`
# @param pad.missing.days pad missing days with `NA` rows?
# @param tz Used in POSIXct conversion for custom timezone. Default `""` is current locale
# @param format Used in POSIXct conversion. Default format for Henry date times `"%Y-%m-%d %H:%M:%S"`
.formatDates <- function(sensor.data, gran, pad.missing.days, tz = "UTC", format = "%Y-%m-%d %H:%M:%S") {
.SD <- NULL
# must have data, otherwise do nothing
# when sensor data are missing, sensor.data is a list of length 0
if (length(sensor.data) > 0) {
sensor.data$date_time <- as.POSIXct(sensor.data$date_time, format = format, tz = tz)
sensor.data$year <- as.integer(format(sensor.data$date_time, "%Y"))
sensor.data$doy <- as.integer(format(sensor.data$date_time, "%j"))
sensor.data$month <- format(sensor.data$date_time, "%b")
# re-level months
sensor.data$month <- factor(sensor.data$month, levels = c('Jan','Feb','Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Dec'))
# optionally pad daily data with NA
if (gran == 'day' & pad.missing.days) {
sensor.data <- as.data.table(sensor.data)
cnm <- colnames(sensor.data)
sensor.data <- sensor.data[, .fill_missing_days(.SD), by = c('sid', 'year'), .SDcols = cnm]
sensor.data <- sensor.data[, .SD, .SDcols = cnm]
sensor.data <- as.data.frame(sensor.data)
}
# add-in seasons
sensor.data$season <- month2season(sensor.data$month)
# water year/day: October 1st -- September 30th
w <- waterDayYear(sensor.data$date_time, tz = tz)
# row-order is preserved
sensor.data$water_year <- w$wy
sensor.data$water_day <- w$wd
}
return(sensor.data)
}
# this loads and packages the data into a list of objects
#' Get data from Henry Mount Soil Temperature and Water Database
#'
#' This function is a front-end to the REST query functionality of the Henry
#' Mount Soil Temperature and Water Database.
#'
#' Filling missing days with NA is useful for computing and index of how
#' complete the data are, and for estimating (mostly) unbiased MAST and
#' seasonal mean soil temperatures. Summaries are computed by first averaging
#' over Julian day, then averaging over all days of the year (MAST) or just
#' those days that occur within "summer" or "winter". This approach makes it
#' possible to estimate summaries in the presence of missing data. The quality
#' of summaries should be weighted by the number of "functional years" (number
#' of years with non-missing data after combining data by Julian day) and
#' "complete years" (number of years of data with >= 365 days of non-missing
#' data).
#'
#' @aliases fetchHenry month2season summarizeSoilTemperature
#' @param what type of data to return: 'sensors': sensor metadata only |
#' 'soiltemp': sensor metadata + soil temperature data | 'soilVWC': sensor
#' metadata + soil moisture data | 'airtemp': sensor metadata + air temperature
#' data | 'waterlevel': sensor metadata + water level data |'all': sensor
#' metadata + all sensor data
#' @param usersiteid (optional) filter results using a NASIS user site ID
#' @param project (optional) filter results using a project ID
#' @param sso (optional) filter results using a soil survey office code
#' @param gran data granularity: "hour" (if available), "day", "week", "month", "year"; returned data
#' are averages
#' @param start.date (optional) starting date filter
#' @param stop.date (optional) ending date filter
#' @param pad.missing.days should missing data ("day" granularity) be filled
#' with NA? see details
#' @param soiltemp.summaries should soil temperature ("day" granularity only)
#' be summarized? see details
#' @param tz Used for custom timezone. Default `""` is current locale
#' @return a list containing: \item{sensors}{a \code{sf} \code{data.frame}
#' object containing site-level information} \item{soiltemp}{a
#' \code{data.frame} object containing soil temperature timeseries data}
#' \item{soilVWC}{a \code{data.frame} object containing soil moisture
#' timeseries data} \item{airtemp}{a \code{data.frame} object containing air
#' temperature timeseries data} \item{waterlevel}{a \code{data.frame} object
#' containing water level timeseries data}
#' @note This function and the back-end database are very much a work in
#' progress.
#' @author D.E. Beaudette
#' @seealso \code{\link{fetchSCAN}}
#' @keywords manip
#'
#' @details See:
#' - [Henry Mount Soil Climate Database](http://soilmap2-1.lawr.ucdavis.edu/henry/)
#' - [`fetchHenry` Tutorial](http://ncss-tech.github.io/AQP/soilDB/Henry-demo.html)
#'
#' @export fetchHenry
fetchHenry <- function(what='all', usersiteid=NULL, project=NULL, sso=NULL, gran='day', start.date=NULL, stop.date=NULL, pad.missing.days=TRUE, soiltemp.summaries=TRUE, tz='') {
# check for required packages
if (!requireNamespace('jsonlite', quietly = TRUE))
stop('please install the `jsonlite` package', call. = FALSE)
if (!requireNamespace('sf', quietly = TRUE))
stop('please install the `sf` package', call. = FALSE)
# important: backward compatibility R <4.0
opt.original <- options(stringsAsFactors = FALSE)
# sanity-check: `what` should be within the legal set of options
if (!what %in% c('all', 'sensors', 'soiltemp', 'soilVWC', 'airtemp', 'waterlevel'))
stop("`what` must be either: 'all', 'sensors', 'soiltemp', 'soilVWC', 'airtemp', or 'waterlevel'", call. = FALSE)
# sanity-check: user must supply some kind of criteria
if (what != 'sensors') {
if (missing(usersiteid) & missing(project) & missing(sso))
stop('you must provide some filtering criteria', call. = FALSE)
}
# init empty filter
f <- vector()
# init empty pieces
s <- NULL
# process filter components
if (!is.null(usersiteid)) {
f <- c(f, paste('&usersiteid=', usersiteid, sep = ''))
}
if (!is.null(project)) {
project <- paste(project, collapse = ',')
f <- c(f, paste('&project=', project, sep = ''))
}
if (!is.null(sso)) {
sso <- paste(sso, collapse = ',')
f <- c(f, paste('&sso=', sso, sep = ''))
}
if (!is.null(gran)) {
f <- c(f, paste('&gran=', gran, sep = ''))
}
if (!is.null(start.date)) {
f <- c(f, paste('&start=', start.date, sep = ''))
}
if (!is.null(stop.date)) {
f <- c(f, paste('&stop=', stop.date, sep = ''))
}
# combine filters
f <- paste(f, collapse = '')
# everything in one URL / JSON package
json.url <- URLencode(paste('http://soilmap2-1.lawr.ucdavis.edu/henry/query.php?what=', what, f, sep = ''))
# this is a little noisy, but people like to see progress
tf.json <- tempfile()
curl::curl_download(url = json.url, destfile = tf.json, mode = 'wb', handle = .soilDB_curl_handle(), quiet = FALSE)
## TODO: check NA handling
# parse JSON into list of DF
try({
s <- jsonlite::fromJSON(gzfile(tf.json))
})
# report query that returns no data and stop
if (length(s$sensors) == 0 ) {
stop('query returned no data', call. = FALSE)
}
# post-process data, if there are some
if (length(s$soiltemp) > 0 | length(s$soilVWC) > 0 | length(s$airtemp) > 0 | length(s$waterlevel) > 0 ) {
.SD <- NULL
# period of record over all sensors
.POR <- function(i) {
# date range
start.date <- min(i$date_time, na.rm = TRUE)
end.date <- max(i$date_time, na.rm = TRUE)
# compute days since last visit
dslv <- round(as.numeric(difftime(Sys.Date(), end.date, units = 'days')))
res <- data.frame(start.date, end.date, dslv)
return(res)
}
por <- as.data.table(na.omit(rbind(s$soiltemp, s$soilVWC, s$airtemp, s$waterlevel)))[, .POR(.SD), by = 'sid']
por <- as.data.frame(por)
# convert dates and add helper column
s$soiltemp <- .formatDates(s$soiltemp, gran = gran, pad.missing.days = pad.missing.days, tz = tz)
s$soilVWC <- .formatDates(s$soilVWC, gran = gran, pad.missing.days = pad.missing.days, tz = tz)
s$airtemp <- .formatDates(s$airtemp, gran = gran, pad.missing.days = pad.missing.days, tz = tz)
s$waterlevel <- .formatDates(s$waterlevel, gran = gran, pad.missing.days = pad.missing.days, tz = tz)
# optionally compute summaries, requires padded NA values and, daily granularity
if (soiltemp.summaries & pad.missing.days & (length(s$soiltemp) > 0)) {
message('computing un-biased soil temperature summaries')
if (gran != 'day')
stop('soil temperature summaries can only be computed from daily data', call. = FALSE)
# compute unbiased estimates of MAST and summer/winter temp
soiltemp.summary <- summarizeSoilTemperature(s$soiltemp)
# combine summaries and join to sensors data
por <- merge(por, soiltemp.summary, by = 'sid', all.x = TRUE, sort = FALSE)
}
# splice-into sensors data by = 'sid', all.x = TRUE, sort = FALSE)
s$sensors <- merge(s$sensors, por, by = 'sid', all.x = TRUE, sort = FALSE)
}
# copy over sensor name + depth to all sensor tables--if present
# "name" is the sensor name - depth for plotting
if (length(s$soiltemp) > 0) {
name.idx <- match(s$soiltemp$sid, s$sensors$sid)
s$soiltemp$name <- paste0(s$sensors$name[name.idx], '-', s$sensors$sensor_depth[name.idx])
s$soiltemp$sensor_name <- s$sensors$name[name.idx]
s$soiltemp$sensor_depth <- s$sensors$sensor_depth[name.idx]
}
if (length(s$soilVWC) > 0) {
name.idx <- match(s$soilVWC$sid, s$sensors$sid)
s$soilVWC$name <- paste0(s$sensors$name[name.idx], '-', s$sensors$sensor_depth[name.idx])
s$soilVWC$sensor_name <- s$sensors$name[name.idx]
s$soilVWC$sensor_depth <- s$sensors$sensor_depth[name.idx]
}
if (length(s$airtemp) > 0) {
name.idx <- match(s$airtemp$sid, s$sensors$sid)
s$airtemp$name <- paste0(s$sensors$name[name.idx], '-', s$sensors$sensor_depth[name.idx])
s$airtemp$sensor_name <- s$sensors$name[name.idx]
s$airtemp$sensor_depth <- s$sensors$sensor_depth[name.idx]
}
if (length(s$waterlevel) > 0) {
name.idx <- match(s$waterlevel$sid, s$sensors$sid)
s$waterlevel$name <- paste0(s$sensors$name[name.idx], '-', s$sensors$sensor_depth[name.idx])
s$waterlevel$sensor_name <- s$sensors$name[name.idx]
s$waterlevel$sensor_depth <- s$sensors$sensor_depth[name.idx]
}
# init coordinates
if (!is.null(s$sensors)) {
s$sensors <- sf::st_as_sf(s$sensors,
coords = c("wgs84_longitude", "wgs84_latitude"),
crs = 'EPSG:4326')
}
# reset options:
options(opt.original)
# some feedback via message:
s.size <- round(object.size(s) / 1024 / 1024, 2)
message(paste(nrow(s$sensors), ' sensors loaded (', s.size, ' Mb transferred)', sep = ''))
# done
return(s)
}