# Author: An-Chi Ho # Date: 13rd July 2021 #--------------------------------------------------------------------- # This script shows how to use a value array as the inner dimension selector to express # dependency on a file dimension. By this means, we don't need to specify the *_across # parameter and Start() can recognize this dependecy relationship. # In the first case, 'time' is dependent on 'sdate'. We create the actual time values # for each sdate beforehand. The time array is two-dimensional with the names 'time' # and 'sdate'. # In the second case, 'region' is dependent on 'sdate'. The two files have different # index for Nino3. sdate 1993 has 'Nino3' at index 9 while sdate 2013 has 'Nino3' at # index 11. Create a value array for region selector so Start() can look for 'Nino3' in # each file. #--------------------------------------------------------------------- library(startR) library(lubridate) # Case 1: 'time' depends on 'sdate' repos <- '/esarchive/exp/ecmwf/system4_m1/daily_mean/$var$_f24h/$var$_$sdate$.nc' yr1 <- "2001" sdates <- ymd("20010501") + rep(years(0:2), each = 1) times <- array(ymd("20010501") + days(0:30) + rep(years(0:2), each = 31), dim = c(time = 31, sdate = 3)) times <- as.POSIXct(times * 86400, tz = 'UTC', origin = '1970-01-01') exp <- Start(dat = repos, var = 'tos', sdate = format(sdates, "%Y%m%d"), time = times, #dim: [time = 31, sdate = 3]. time is corresponding to each sdate ensemble = indices(1:5), lat = 'all', lon = 'all', synonims = list(lat = c('lat', 'latitude'), lon = c('lon', 'longitude')), return_vars = list(lon = NULL, lat = NULL, time = 'sdate'), retrieve = T) #============================================================================= # Case 2: 'region' depends on 'sdate' path <- paste0('/esarchive/exp/ecearth/a35b/diags/DCPP/EC-Earth-Consortium/', 'EC-Earth3-HR/dcppA-hindcast/r1i1p1f1/Omon/$var$_mixed/gn/v20201107/', '$var$_Omon_EC-Earth3-HR_dcppA-hindcast_s$sdate$-r1i1p1f1_gn_$chunk$.nc') region <- array('Nino3', dim = c(sdate = 2, region = 1)) data <- Start(dat = path, var = 'tosmean', sdate = c('1993', '2013'), chunk = indices(1:2), chunk_depends = 'sdate', region = region, time = 'all', time_across = 'chunk', merge_across_dims = TRUE, return_vars = list(time = c('sdate', 'chunk'), region = 'sdate'), retrieve = T) dim(data) # dat var sdate region time # 1 1 2 1 2 data[1, 1, , 1, ] # [,1] [,2] #[1,] 24.98788 24.46488 # --> region index 9 in original file #[2,] 24.47482 24.75953 # --> region index 11 in orginal file