sample_file <- function(x, ...) {
dplyr::sample_n(x, 1L) %>% dplyr::select(date, fullname)
}
as_tib_slab <- function(trans, slab, activ) {
tib <- list()
tib[[activ]] <- as.vector(slab)
tib <- as_tibble(tib)
prod_dims <- 1
total_prod <- prod(dim(slab))
for (i in seq_along(trans)) {
nm <- names(trans)[i]
nr <- nrow(trans[[i]])
tib[[nm]] <- rep(trans[[nm]][[nm]], each = prod_dims, length.out = total_prod)
prod_dims <- prod_dims * nr
}
tib
}
library(ncdump)
library(tibble)
library(dplyr, warn.conflicts = FALSE)
library(maps)
library(raadtools)
## Loading required package: raster
## Loading required package: sp
##
## Attaching package: 'raster'
## The following object is masked from 'package:dplyr':
##
## select
library(ggplot2)
set.seed(1)
files <- lapply(list(
sstfiles(time.resolution = "daily")
,sstfiles(time.resolution = "monthly")
,chlafiles()
,currentsfiles()
,derivicefiles()
,derivaadcfiles("si_200_interpolated_summer_climatology") %>% mutate(date = as.POSIXct(NA))
#,icefiles(product = "amsr")
,ocfiles(time.resolution = "monthly", product = "SeaWiFS", varname = "CHL", type= "L3m")
,sshfiles(ssha = TRUE)
,(windfiles() %>% dplyr::transmute(fullname = ufullname, date))
), sample_file) %>% bind_rows()
f <- files$fullname[1]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data/eclipse.ncdc.noaa.gov/pub/OI-daily-v2/NetCDF/1991/AVHRR/avhrr-only-v2.19910221.nc"
## the idea is to find out what's the file
NetCDF(f) %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 4 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 lat -89.875 89.875
## 2 lon 0.125 359.875
## 3 time 4799.000 4799.000
## 4 zlev 0.000 0.000
(x <- NetCDF(f) )
## Variables: sst, (anom, err, ice)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 4 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 sst 4 3 lon 1440
## 2 sst 4 2 lat 720
## 3 sst 4 1 zlev 1
## 4 sst 4 0 time 1
## then build up a query, returning a subsett-ed set of tables of each dimension's coordinates
trans <- x %>% filtrate(lon = lon > 100, lat = between(lat, -30, 20))
## Joining, by = ".variable_"
## extract the hyperslab index from the transform tables
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
## (this will be wrapped up in an easy function but it's important to keep things general)
image(trans$lon$lon, trans$lat$lat, slab, col = viridis::viridis(100))
map("world2", add = TRUE)
f <- files$fullname[2]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data/ftp.cdc.noaa.gov/Datasets/noaa.oisst.v2/sst.mnmean.nc"
(x <- NetCDF(f) )
## Variables: sst, (time_bnds)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 3 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 sst 2 1 lon 360
## 2 sst 2 0 lat 180
## 3 sst 2 2 time 424
## the idea is to find out what's the file
x %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 3 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 lat -89.5 89.5
## 2 lon 0.5 359.5
## 3 time 66443.0 79317.0
## then build up a query, returning a subsett-ed set of tables of each dimension's coordinates
(trans <- x %>% filtrate(lon = lon > 250, lat = lat < -10))
## Joining, by = ".variable_"
## $lon
## # A tibble: 110 × 5
## .dimension_ id lon name step
## <int> <int> <dbl> <chr> <int>
## 1 1 1 250.5 lon 251
## 2 1 1 251.5 lon 252
## 3 1 1 252.5 lon 253
## 4 1 1 253.5 lon 254
## 5 1 1 254.5 lon 255
## 6 1 1 255.5 lon 256
## 7 1 1 256.5 lon 257
## 8 1 1 257.5 lon 258
## 9 1 1 258.5 lon 259
## 10 1 1 259.5 lon 260
## # ... with 100 more rows
##
## $lat
## # A tibble: 80 × 5
## .dimension_ id lat name step
## <int> <int> <dbl> <chr> <int>
## 1 0 0 -10.5 lat 101
## 2 0 0 -11.5 lat 102
## 3 0 0 -12.5 lat 103
## 4 0 0 -13.5 lat 104
## 5 0 0 -14.5 lat 105
## 6 0 0 -15.5 lat 106
## 7 0 0 -16.5 lat 107
## 8 0 0 -17.5 lat 108
## 9 0 0 -18.5 lat 109
## 10 0 0 -19.5 lat 110
## # ... with 70 more rows
##
## $time
## # A tibble: 424 × 5
## .dimension_ id time name step
## <int> <int> <dbl> <chr> <int>
## 1 2 2 66443 time 1
## 2 2 2 66474 time 2
## 3 2 2 66505 time 3
## 4 2 2 66533 time 4
## 5 2 2 66564 time 5
## 6 2 2 66594 time 6
## 7 2 2 66625 time 7
## 8 2 2 66655 time 8
## 9 2 2 66686 time 9
## 10 2 2 66717 time 10
## # ... with 414 more rows
## extract the hyperslab index from the transform tables
## notice how this and the next block is completely devoid of references to the dimension or variable names
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
## (this will be wrapped up in an easy function but it's important to keep things general)
## note that this file uses south-up convention
image(trans$lon$lon, rev(trans$lat$lat), slab[,ncol(slab):1,1], col = viridis::viridis(100))
map("world2", add = TRUE)
## see how we got a 3D slab this time, because we didn't subset on time
dim(slab)
## [1] 110 80 424
## this time the slab will be 2D , in longitude ~ time
(trans <- x %>% filtrate(lon = lon == 147.5, lat = lat < 0 & lat > -70))
## Joining, by = ".variable_"
## $lon
## # A tibble: 1 × 5
## .dimension_ id lon name step
## <int> <int> <dbl> <chr> <int>
## 1 1 1 147.5 lon 148
##
## $lat
## # A tibble: 70 × 5
## .dimension_ id lat name step
## <int> <int> <dbl> <chr> <int>
## 1 0 0 -0.5 lat 91
## 2 0 0 -1.5 lat 92
## 3 0 0 -2.5 lat 93
## 4 0 0 -3.5 lat 94
## 5 0 0 -4.5 lat 95
## 6 0 0 -5.5 lat 96
## 7 0 0 -6.5 lat 97
## 8 0 0 -7.5 lat 98
## 9 0 0 -8.5 lat 99
## 10 0 0 -9.5 lat 100
## # ... with 60 more rows
##
## $time
## # A tibble: 424 × 5
## .dimension_ id time name step
## <int> <int> <dbl> <chr> <int>
## 1 2 2 66443 time 1
## 2 2 2 66474 time 2
## 3 2 2 66505 time 3
## 4 2 2 66533 time 4
## 5 2 2 66564 time 5
## 6 2 2 66594 time 6
## 7 2 2 66625 time 7
## 8 2 2 66655 time 8
## 9 2 2 66686 time 9
## 10 2 2 66717 time 10
## # ... with 414 more rows
## extract the hyperslab index from the transform tables
## notice how this and the next block is completely devoid of references to the dimension or variable names
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
image(rev(trans$lat$lat), trans$time$time, slab[nrow(slab):1, ], col = viridis::viridis(100))
f <- files$fullname[3]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data_local/chl/johnson/modis/8d/A20081852008192.L3m_8D_SO_Chl_9km.Johnson_SO_Chl.nc"
(x <- NetCDF(f) )
## Variables: chlorophyll, (originalFileAttributes)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 3 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 chlorophyll 0 0 longitude 4320
## 2 chlorophyll 0 1 latitude 720
## 3 chlorophyll 0 2 time 1
x %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 3 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 latitude -89.95834 -30.04167
## 2 longitude -179.95833 179.95835
## 3 time 6758.00000 6758.00000
(trans <- x %>% filtrate(longitude = longitude > 0, latitude = latitude > -50))
## Joining, by = ".variable_"
## $longitude
## # A tibble: 2,160 × 5
## .dimension_ id longitude name step
## <int> <int> <dbl> <chr> <int>
## 1 0 0 0.04167712 longitude 2161
## 2 0 0 0.12501045 longitude 2162
## 3 0 0 0.20834379 longitude 2163
## 4 0 0 0.29167712 longitude 2164
## 5 0 0 0.37501046 longitude 2165
## 6 0 0 0.45834380 longitude 2166
## 7 0 0 0.54167713 longitude 2167
## 8 0 0 0.62501047 longitude 2168
## 9 0 0 0.70834380 longitude 2169
## 10 0 0 0.79167714 longitude 2170
## # ... with 2,150 more rows
##
## $latitude
## # A tibble: 240 × 5
## .dimension_ id latitude name step
## <int> <int> <dbl> <chr> <int>
## 1 1 1 -49.95833 latitude 481
## 2 1 1 -49.87500 latitude 482
## 3 1 1 -49.79167 latitude 483
## 4 1 1 -49.70833 latitude 484
## 5 1 1 -49.62500 latitude 485
## 6 1 1 -49.54167 latitude 486
## 7 1 1 -49.45833 latitude 487
## 8 1 1 -49.37500 latitude 488
## 9 1 1 -49.29167 latitude 489
## 10 1 1 -49.20833 latitude 490
## # ... with 230 more rows
##
## $time
## # A tibble: 1 × 5
## .dimension_ id time name step
## <int> <int> <dbl> <chr> <int>
## 1 2 2 6758 time 1
## extract the hyperslab index from the transform tables
## notice how this and the next block is completely devoid of references to the dimension or variable names
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
## (this will be wrapped up in an easy function but it's important to keep things general)
## note that this file uses south-up convention
image(trans$longitude$longitude, trans$latitude$latitude, log(slab), col = viridis::viridis(100))
map("world2", add = TRUE)
f <- files$fullname[4]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data/ftp.aviso.altimetry.fr/global/delayed-time/grids/madt/all-sat-merged/uv/2015/dt_global_allsat_madt_uv_20150213_20150914.nc"
## now we really want to activate a different variable, because the default chosen is not of interest
(x <- NetCDF(f) )
## Variables: lat_bnds, (lon_bnds, crs, u, v)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 2 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 lat_bnds 2 3 nv 2
## 2 lat_bnds 2 1 lat 720
(x <- NetCDF(f) %>% activate("u"))
## Variables: u, (lat_bnds, lon_bnds, crs, v)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 3 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 u 7 2 lon 1440
## 2 u 7 1 lat 720
## 3 u 7 0 time 1
x %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 3 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 lat -89.875 89.875
## 2 lon 0.125 359.875
## 3 time 23784.000 23784.000
(trans <- x %>% filtrate(lon = between(lon, 100, 150), lat = lat > -50 & lat < 20))
## Joining, by = ".variable_"
## $lon
## # A tibble: 200 × 5
## .dimension_ id lon name step
## <int> <int> <dbl> <chr> <int>
## 1 2 2 100.125 lon 401
## 2 2 2 100.375 lon 402
## 3 2 2 100.625 lon 403
## 4 2 2 100.875 lon 404
## 5 2 2 101.125 lon 405
## 6 2 2 101.375 lon 406
## 7 2 2 101.625 lon 407
## 8 2 2 101.875 lon 408
## 9 2 2 102.125 lon 409
## 10 2 2 102.375 lon 410
## # ... with 190 more rows
##
## $lat
## # A tibble: 280 × 5
## .dimension_ id lat name step
## <int> <int> <dbl> <chr> <int>
## 1 1 1 -49.875 lat 161
## 2 1 1 -49.625 lat 162
## 3 1 1 -49.375 lat 163
## 4 1 1 -49.125 lat 164
## 5 1 1 -48.875 lat 165
## 6 1 1 -48.625 lat 166
## 7 1 1 -48.375 lat 167
## 8 1 1 -48.125 lat 168
## 9 1 1 -47.875 lat 169
## 10 1 1 -47.625 lat 170
## # ... with 270 more rows
##
## $time
## # A tibble: 1 × 5
## .dimension_ id time name step
## <int> <int> <dbl> <chr> <int>
## 1 0 0 23784 time 1
## extract the hyperslab index from the transform tables
## notice how this and the next block is completely devoid of references to the dimension or variable names
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
uslab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
## get the other slab while we are at it
x <- activate(x, "v")
vslab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
## (this will be wrapped up in an easy function but it's important to keep things general)
## note that this file uses south-up convention
image(trans$lon$lon, trans$lat$lat, uslab, col = viridis::viridis(100))
## use u and v together
image(trans$lon$lon, trans$lat$lat, sqrt(uslab^2 + vslab^2), col = viridis::viridis(100))
map("world2", add = TRUE)
f <- files$fullname[5]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data/webdav.data.aad.gov.au/data/environmental/smmr_ssmi_nasateam/time_since_melt/daily/1986/time_since_melt_19860514.nc"
(nc_object <- NetCDF(f) %>% activate("days_since_ice_melt"))
## Variables: days_since_ice_melt, (polar_stereographic)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 2 × 5
## variable_name .variable_ .dimension_ dimension_name
## <chr> <dbl> <int> <chr>
## 1 days_since_ice_melt 0 0 x
## 2 days_since_ice_melt 0 1 y
## # ... with 1 more variables: dimension_length <int>
nc_object %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 2 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 x -3937500 3937500
## 2 y -3937500 4337500
## what's wrong here??
#(trans <- nc_object %>% filtrate(x = x > 0, y = y < 0))
f <- files$fullname[6]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data/webdav.data.aad.gov.au/data/environmental/derived/antarctic/netcdf/si_200_interpolated_summer_climatology.nc"
## the idea is to find out what's the file
NetCDF(f) %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 2 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 lat -80 -30
## 2 lon -180 180
(x <- NetCDF(f) )
## Variables: silicate
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 2 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 silicate 2 1 lon 3601
## 2 silicate 2 0 lat 501
## then build up a query, returning a subsett-ed set of tables of each dimension's coordinates
trans <- x %>% filtrate(lon = lon > 100, lat = between(lat, -50, -30))
## Joining, by = ".variable_"
## extract the hyperslab index from the transform tables
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
## (this will be wrapped up in an easy function but it's important to keep things general)
image(trans$lon$lon, trans$lat$lat, slab, col = viridis::viridis(100))
map("world2", add = TRUE)
ggplot(as_tib_slab(trans, slab, "silicate"), aes(lon, lat, fill = silicate)) + geom_raster()
f <- files$fullname[7]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data/oceandata.sci.gsfc.nasa.gov/SeaWiFS/Mapped/Monthly/9km/chlor/S20100912010120.L3m_MO_CHL_chlor_a_9km.nc"
## the idea is to find out what's the file
NetCDF(f) %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 2 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 lat -89.95834 89.95834
## 2 lon -179.95833 179.95836
(x <- NetCDF(f) )
## Variables: chlor_a, (palette)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 2 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 chlor_a 0 1 lon 4320
## 2 chlor_a 0 0 lat 2160
## then build up a query, returning a subsett-ed set of tables of each dimension's coordinates
trans <- x %>% filtrate(lon = lon > 100, lat = between(lat, -50, -30))
## Joining, by = ".variable_"
## extract the hyperslab index from the transform tables
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
cpal <- palr::chlPal(palette = TRUE)
image(trans$lon$lon, rev(trans$lat$lat), slab[,ncol(slab):1], col = cpal$cols[-1], breaks = cpal$breaks)
map("world2", add = TRUE)
f <- files$fullname[8]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data/ftp.aviso.altimetry.fr/global/delayed-time/grids/msla/all-sat-merged/h/2008/dt_global_allsat_msla_h_20081024_20140106.nc"
## the idea is to find out what's the file
NetCDF(f) %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 2 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 lat -89.875 89.875
## 2 nv 0.000 1.000
(x <- NetCDF(f) %>% activate("sla") )
## Variables: sla, (lat_bnds, lon_bnds, crs, err)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 3 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 sla 7 2 lon 1440
## 2 sla 7 1 lat 720
## 3 sla 7 0 time 1
## then build up a query, returning a subsett-ed set of tables of each dimension's coordinates
trans <- x %>% filtrate(lon = lon > 100 & lon < 340, lat = between(lat, 0, 50))
## Joining, by = ".variable_"
## extract the hyperslab index from the transform tables
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
image(trans$lon$lon, trans$lat$lat, slab, col = viridis::viridis(100), asp = 1/cos(-40 * pi/180), xlim = c(0, 360))
map("world2", add = TRUE)
f <- files$fullname[9]
print(gsub(getOption("default.datadir"), "", f))
## [1] "/data/ftp.cdc.noaa.gov/Datasets/ncep.reanalysis2/gaussian_grid/uwnd.10m.gauss.2002.nc"
## the idea is to find out what's the file
NetCDF(f) %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 4 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 lat -88.542 88.542
## 2 level 10.000 10.000
## 3 lon 0.000 358.125
## 4 time 1770696.000 1779450.000
(x <- NetCDF(f) )
## Variables: uwnd
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 4 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 uwnd 4 0 lon 192
## 2 uwnd 4 1 lat 94
## 3 uwnd 4 2 level 1
## 4 uwnd 4 3 time 1460
## then build up a query, returning a subsett-ed set of tables of each dimension's coordinates
trans <- x %>% filtrate(lon = lon > 300, lat = between(lat, 0, 50), time = time < 1770732)
## Joining, by = ".variable_"
## extract the hyperslab index from the transform tables
hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step))))
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
for (i in seq(nrow(trans$time))) {
image(trans$lon$lon, rev(trans$lat$lat), slab[,ncol(slab):1,i], col = viridis::viridis(100), asp = 1/cos(-40 * pi/180))
map("world2", add = TRUE)
}
ggplot(as_tib_slab(trans, slab, "uwnd"), aes(lon, lat, fill = uwnd)) + geom_raster() + facet_wrap(~time)
f <- file.path(getOption("default.datadir"), "data_local/acecrc.org.au/ROMS/s_corney/cpolar/ocean_his_3101.nc")
(x <- NetCDF(f) %>% activate("salt"))
## Variables: salt, (ntimes, ndtfast, dt, dtfast, dstart, nHIS, ndefHIS, nRST, Falpha, Fbeta, Fgamma, nl_tnu2, nl_visc2, Akt_bak, Akv_bak, rdrg, rdrg2, Zob, Zos, Znudg, M2nudg, M3nudg, Tnudg, FSobc_in, FSobc_out, M2obc_in, M2obc_out, Tobc_in, Tobc_out, M3obc_in, M3obc_out, rho0, gamma2, spherical, xl, el, Vtransform, Vstretching, theta_s, theta_b, Tcline, hc, Cs_r, Cs_w, h, zice, f, pm, pn, lon_rho, lat_rho, lon_u, lat_u, lon_v, lat_v, lon_psi, lat_psi, angle, mask_rho, mask_u, mask_v, mask_psi, zeta, ubar, vbar, u, v, w, temp, rho, Hsbl, AKv, AKt, AKs, shflux, ssflux, swrad, sustr, svstr, bustr, bvstr)
## Dimensions:
## Joining, by = ".dimension_"
## # A tibble: 4 × 5
## variable_name .variable_ .dimension_ dimension_name dimension_length
## <chr> <dbl> <int> <chr> <int>
## 1 salt 72 0 xi_rho 1443
## 2 salt 72 4 eta_rho 392
## 3 salt 72 8 s_rho 31
## 4 salt 72 12 ocean_time 31
x %>% dimension_values() %>% group_by(name) %>% summarize_at("vals", funs(min, max))
## Joining, by = ".variable_"
## # A tibble: 4 × 3
## name min max
## <chr> <dbl> <dbl>
## 1 eta_rho 1.000000e+00 3.920000e+02
## 2 ocean_time 9.465984e+08 9.491904e+08
## 3 s_rho -9.838710e-01 -1.612903e-02
## 4 xi_rho 1.000000e+00 1.443000e+03
## then build up a query, returning a subsett-ed set of tables of each dimension's coordinates
trans <- x %>% filtrate(eta_rho = eta_rho < 100, xi_rho = between(xi_rho, 500, 600), ocean_time = ocean_time < 9.465984e+08 + 12, s_rho = s_rho > -0.17741935)
## Joining, by = ".variable_"
## extract the hyperslab index from the transform tables
(hslab <- bind_rows(lapply(trans, function(x) tibble(name = x$name[1], start = min(x$step), count = length(x$step)))))
## # A tibble: 4 × 3
## name start count
## <chr> <int> <int>
## 1 xi_rho 500 101
## 2 eta_rho 1 99
## 3 s_rho 27 5
## 4 ocean_time 1 1
## open the file and pull out that slab
con <- ncdf4::nc_open(x$file$filename[1])
slab <- ncdf4::ncvar_get(con, nctive(x),
start = hslab$start,
count = hslab$count)
## much more to think about ...