Hard coded exception in 'sw_SiteClimate_Ambient' for year 1942

[dschlaep]

The function 'sw_SiteClimate_Ambient' has a hard-coded exception for the year 1942:
- If the year is 1942, then an additional day is added to the sequence 'month_forEachDoy'

I don't see the reason why that this is done. It appears that the code believes that the year 1942 is a leap year and that the input argument 'weatherList' would treat it as non-leap year? In any case, 1942 was not a leap year.

Please explain code in a comment or remove.

Thanks, Daniel

[gopherchucks]

Interesting, I'm not sure why it's there. I don't think it's my code. Maybe someone was writing a test case, or something? Anyway, I'll delete it since it doesn't look like it has any effect on the Wrapper.

[CaitlinA]

In retrospect, I think I am the one who hardcoded the 1942 exception into the sw_SiteClimate_Ambient function. I ran into the 1942 'problem' again today. For some reason, in 1942, the month_forEachDoy statement, only creates a sequence that is a length of 364, and ends on 12-30-1942 (excludes 12-31-1942). The hardcoded exception added an additional day-month to account for 12-31-1942. Do you have a more elegant solution to fix this? I will send along weather files in a separate file so you can see for yourself.

@dschlaep

[dschlaep]

Replicate problem described by Caitlin for year 1942

sessionInfo()

R version 3.3.0 (2016-05-03)
Platform: x86_64-apple-darwin14.5.0 (64-bit)
Running under: OS X 10.10.5 (Yosemite)

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

attached base packages:
[1] stats graphics grDevices utils datasets methods base

loaded via a namespace (and not attached):
[1] tools_3.3.0 Rsoilwat31_1.1.1

## Load weather data in question
weath <- Rsoilwat31::getWeatherData_folders(
  LookupWeatherFolder = ".", weatherDirName = "Weather_SheepStation",
  filebasename = "weath", startYear = 1926, endYear = 2015)

# Check structure
str(weath[["1942"]])
wdims <- sapply(weath, function(x) dim(x@data))

isLeapYear <- function(y) {
  #from package: tis
  y %% 4 == 0 & (y %% 100 != 0 | y %% 400 == 0)
}

identical(as.vector(wdims[1, ]), ifelse(isLeapYear(1926:2015), 366L, 365L))

Sys.timezone()

## Formal class 'swWeatherData' [package "Rsoilwat31"] with 2 slots
##   ..@ data: num [1:365, 1:4] 1 2 3 4 5 6 7 8 9 10 ...
##   .. ..- attr(*, "dimnames")=List of 2
##   .. .. ..$ : NULL
##   .. .. ..$ : chr [1:4] "DOY" "Tmax_C" "Tmin_C" "PPT_cm"
##   ..@ year: int 1942
## [1] TRUE
## [1] "Europe/Zurich"

I tested this for "Europe/Berlin" and "Europe/Zurich". I copy-pasted results below for "Europe/Berlin":

# Code from 'sw_SiteClimate_Ambient'
weatherList <- weath
year.start <- 1926
year.end <- 2015
do.C4vars <- FALSE

  sw.weather.suffix <- as.numeric(names(weatherList))
  itemp <- year.start <= sw.weather.suffix & year.end >= sw.weather.suffix
  years <- sw.weather.suffix[itemp]

  no.yrs <- length(years)
  y <- which(years == 1942)

      x <- Rsoilwat31::get_swWeatherData(weatherList, years[y])@data[, c("Tmax_C", "Tmin_C", "PPT_cm"), drop = FALSE]
      temp.dailyTempMean <- rowMeans(x[, c("Tmax_C", "Tmin_C")])

temp_seq <- seq(from = as.POSIXlt(paste0(years[y], "-01-01")),
                                         to = as.POSIXlt(paste0(years[y], "-12-31")),
                                         by = "1 day")
      month_forEachDoy <- as.POSIXlt(temp_seq)$mon + 1


print(data.frame(temp_seq, month_forEachDoy)[c(305:308, 360:365), ])

##       temp_seq month_forEachDoy
## 305 1942-11-01               11
## 306 1942-11-02               11
## 307 1942-11-03               11
## 308 1942-11-04               11
## 360 1942-12-26               12
## 361 1942-12-27               12
## 362 1942-12-28               12
## 363 1942-12-29               12
## 364 1942-12-30               12
## 365 1942-12-31               12

For timezone "Europe/Berlin", I get

305 1942-11-01 00:00:00 11
306 1942-11-02 00:00:00 11
307 1942-11-02 23:00:00 11
308 1942-11-03 23:00:00 11
360 1942-12-25 23:00:00 12
361 1942-12-26 23:00:00 12
362 1942-12-27 23:00:00 12
363 1942-12-28 23:00:00 12
364 1942-12-29 23:00:00 12
365 1942-12-30 23:00:00 12

==> there is no date for Dec 31, 1942 even though there are 365 entries

print(temp_seq[305:308])

## [1] "1942-11-01 CET" "1942-11-02 CET" "1942-11-03 CET" "1942-11-04 CET"

For timezone "Europe/Berlin", I get:

[1] "1942-11-01 00:00:00 CEST" "1942-11-02 00:00:00 CEST" "1942-11-02 23:00:00 CET" "1942-11-03 23:00:00 CET"

==> date Nov 2, 1942 occurs twice due to switch from CEST to CET

Explanation of issue:

Note: see Regression in strptime:
"In R-3.1.3 that returned "1942-01-01 CEST" which, paradoxically, is correct as they
evidently did strange things in Germany during the war period. Java also returns the
same. However, R-3.2.4 returns "1942-01-01 CET"."

See History of Daylight Saving Time — DST:
"Year-round DST, also called “War Time”, was in force during World War II,
from February 9, 1942, to September 30, 1945"

==> If I run the code on my system where Sys.timezone() returns "Europe/Zurich",
then I don't experience issue #17.

==> Thus, issue 17 should appear for US, Canadian, and German/Hitler time zones
during WW I and II which had DST all year round, but not for time zones which didn't
have DST year round in 1942 (such as Europe/Zurich).

Suggested solution: fix time zone to UTC during time sequence calculation

temp_seq2 <- seq(from = ISOdate(years[y], 1, 1, tz = "UTC"),
               to = ISOdate(years[y], 12, 31, tz = "UTC"),
               by = "1 day")
month_forEachDoy2 <- as.POSIXlt(temp_seq2)$mon + 1

print(data.frame(temp_seq2, month_forEachDoy2)[c(305:308, 360:365), ])

print(temp_seq2[305:308])

##               temp_seq2 month_forEachDoy2
## 305 1942-11-01 12:00:00                11
## 306 1942-11-02 12:00:00                11
## 307 1942-11-03 12:00:00                11
## 308 1942-11-04 12:00:00                11
## 360 1942-12-26 12:00:00                12
## 361 1942-12-27 12:00:00                12
## 362 1942-12-28 12:00:00                12
## 363 1942-12-29 12:00:00                12
## 364 1942-12-30 12:00:00                12
## 365 1942-12-31 12:00:00                12
## [1] "1942-11-01 12:00:00 UTC" "1942-11-02 12:00:00 UTC"
## [3] "1942-11-03 12:00:00 UTC" "1942-11-04 12:00:00 UTC"

[CaitlinA]

How interesting! Who knew that programming could lead to a quick history
lesson? Thanks for looking into this.

On Wed, Sep 28, 2016 at 2:52 AM, daniel [email protected] wrote:

Closed #17
#17 via
eab32aa
eab32aa
.

—
You are receiving this because you were assigned.
Reply to this email directly, view it on GitHub
#17 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AM0e_3_Kfyj8iFjjS5OsacQ1PAdiAikOks5qujjZgaJpZM4JxaLA
.

Caitlin Andrews
Ecologist
Colorado Plateau Research Station
Southwest Biological Science Center
U.S. Geological Survey
Mobile: 802-922-3494