From d079c794cab02a1cd1fdc509e14234e412111d4f Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Thu, 27 Jul 2023 19:56:16 +0530
Subject: [PATCH 01/12] added tests for land_water_mask

---
 .../test.integration.download.CRUNCEP.R       | 47 +++++++++++++++++++
 1 file changed, 47 insertions(+)
 create mode 100644 modules/data.atmosphere/tests/testthat/test.integration.download.CRUNCEP.R

diff --git a/modules/data.atmosphere/tests/testthat/test.integration.download.CRUNCEP.R b/modules/data.atmosphere/tests/testthat/test.integration.download.CRUNCEP.R
new file mode 100644
index 00000000000..a81cd2387c3
--- /dev/null
+++ b/modules/data.atmosphere/tests/testthat/test.integration.download.CRUNCEP.R
@@ -0,0 +1,47 @@
+
+skip_on_ci()
+skip_if_offline()
+
+tmpdir <- tempfile(pattern = "CRUNCEPdata")
+dir.create(tmpdir)
+teardown(unlink(tmpdir, recursive = TRUE))
+
+outfolder  <- tmpdir
+start_date <- "2000-01-01"
+end_date   <- "2000-12-31"
+lat.in     <- 40
+lon.in     <- -88
+
+lat.in <- as.numeric(lat.in)
+lon.in <- as.numeric(lon.in)
+
+lat_grid <- floor(2 * (90 - lat.in)) + 1
+lon_grid <- floor(2 * (lon.in + 180)) + 1
+
+maskfile <- file.path(outfolder, "cruncep_landwater_mask.nc")
+mask_url <- paste0(
+  "https://thredds.daac.ornl.gov/thredds/ncss/ornldaac/1220/",
+  "mstmip_driver_global_hd_landwatermask_v1.nc4",
+  "?var=land_water_mask&disableLLSubset=on&disableProjSubset=on&horizStride=1&",
+  "accept=netcdf"
+)
+utils::download.file(mask_url, maskfile)
+
+mask_nc <- ncdf4::nc_open(maskfile)
+on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
+
+test_that("File exists at desired location", {
+  # Set the desired file path
+  file_path <- paste0(tmpdir, "/cruncep_landwater_mask.nc")
+  
+  # Check if file exists at desired location
+  expect_true(file.exists(file_path))
+})
+
+test_that("NetCDF file contains lat and lon variables", {
+  expect_true("land_water_mask" %in% names(mask_nc$var))
+  
+  # Check the dimensions of "land_water_mask" variable
+  expect_equal(mask_nc$var$land_water_mask$dim[[1]]$name, "lon")
+  expect_equal(mask_nc$var$land_water_mask$dim[[2]]$name, "lat")
+})
\ No newline at end of file

From 28b056f48c859c15d48a07200037e21fda8b01d0 Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Fri, 28 Jul 2023 20:27:43 +0530
Subject: [PATCH 02/12] moved tests

---
 .../integrationTests/test.download.CRUNCEP.R  | 246 ++++++++++++++++++
 .../test.integration.download.CRUNCEP.R       |  47 ----
 2 files changed, 246 insertions(+), 47 deletions(-)
 create mode 100644 modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
 delete mode 100644 modules/data.atmosphere/tests/testthat/test.integration.download.CRUNCEP.R

diff --git a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
new file mode 100644
index 00000000000..8970f867608
--- /dev/null
+++ b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
@@ -0,0 +1,246 @@
+library(testthat)
+library(ncdf4)
+library(lubridate)
+
+skip_on_ci()
+skip_if_offline()
+
+tmpdir <- tempfile(pattern = "CRUNCEPdata")
+dir.create(tmpdir)
+on.exit(teardown(unlink(tmpdir, recursive = TRUE)))
+
+outfolder  <- tmpdir
+start_date <- "2000-01-01"
+end_date   <- "2000-12-31"
+lat.in     <- 40
+lon.in     <- -88
+method     <- "ncss"
+overwrite  <- TRUE
+verbose <- FALSE
+maxErrors <- 10
+sleep <- 2
+
+lat.in <- as.numeric(lat.in)
+lon.in <- as.numeric(lon.in)
+
+lat_grid <- floor(2 * (90 - lat.in)) + 1
+lon_grid <- floor(2 * (lon.in + 180)) + 1
+
+maskfile <- file.path(outfolder, "cruncep_landwater_mask.nc")
+mask_url <- paste0(
+  "https://thredds.daac.ornl.gov/thredds/ncss/ornldaac/1220/",
+  "mstmip_driver_global_hd_landwatermask_v1.nc4",
+  "?var=land_water_mask&disableLLSubset=on&disableProjSubset=on&horizStride=1&",
+  "accept=netcdf"
+)
+utils::download.file(mask_url, maskfile)
+
+mask_nc <- ncdf4::nc_open(maskfile)
+on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
+
+test_that("File exists at desired location", {
+  # Set the desired file path
+  file_path <- paste0(tmpdir, "/cruncep_landwater_mask.nc")
+  
+  # Check if file exists at desired location
+  expect_true(file.exists(file_path))
+})
+
+test_that("NetCDF file contains lat and lon variables", {
+  expect_true("land_water_mask" %in% names(mask_nc$var))
+  
+  # Check the dimensions of "land_water_mask" variable
+  expect_equal(mask_nc$var$land_water_mask$dim[[1]]$name, "lon")
+  expect_equal(mask_nc$var$land_water_mask$dim[[2]]$name, "lat")
+})
+
+
+# Set search radius to up to 2 pixels (1 degree) in any direction
+mask_minlat <- lat_grid - 2
+mask_minlon <- lon_grid - 2
+mask_lats <- ncdf4::ncvar_get(mask_nc, "lat", start = mask_minlat, count = 5)
+mask_lons <- ncdf4::ncvar_get(mask_nc, "lon", start = mask_minlon, count = 5)
+mask_values <- ncdf4::ncvar_get(
+  mask_nc,
+  "land_water_mask",
+  start = c(mask_minlon, mask_minlat),
+  count = c(5, 5)
+)
+
+# Build a lat-lon grid matrix and calculate distance from target coords
+mask_grid <- as.matrix(expand.grid(lon = mask_lons, lat = mask_lats))
+mask_igrid <- as.matrix(expand.grid(lon = seq_along(mask_lons), lats = seq_along(mask_lats)))
+mask_dist <- (lon.in - mask_grid[, 1])^2 + (lat.in - mask_grid[, 2])^2
+
+# Order by increasing distance (closest first)
+mask_order <- order(mask_dist)
+mask_igrido <- mask_igrid[mask_order,]
+on_land <- as.logical(mask_values[mask_igrido])
+igrid <- which(on_land)[1]
+if (igrid > 1) {
+  lon.in.orig <- lon.in
+  lat.in.orig <- lat.in
+  lon.in <- mask_grid[mask_order[igrid], 1]
+  lat.in <- mask_grid[mask_order[igrid], 2]
+}
+
+if (method == "opendap") {
+  # Convert lat-lon to grid row and column
+  lat_grid <- floor(2 * (90 - lat.in)) + 1
+  lon_grid <- floor(2 * (lon.in + 180)) + 1
+} else if (method == "ncss") {
+  # Only downloading point data, so only one point
+  lat_grid <- 1
+  lon_grid <- 1
+}
+
+start_date <- as.POSIXlt(start_date, tz = "UTC")
+end_date <- as.POSIXlt(end_date, tz = "UTC")
+
+# taking out year from dates
+start_year <- lubridate::year(start_date)
+end_year <- lubridate::year(end_date)
+
+# year - 2000
+ylist <- seq(start_year, end_year, by = 1)
+rows <- length(ylist)
+results <- data.frame(file = character(rows),
+                      host = character(rows),
+                      mimetype = character(rows),
+                      formatname = character(rows),
+                      startdate = character(rows),
+                      enddate = character(rows),
+                      dbfile.name = "CRUNCEP",
+                      stringsAsFactors = FALSE)
+
+var <- tibble::tribble(
+  ~DAP.name, ~CF.name, ~units,
+  "tair", "air_temperature", "Kelvin",
+  "lwdown", "surface_downwelling_longwave_flux_in_air", "W/m2",
+  "press", "air_pressure", "Pascal",
+  "swdown", "surface_downwelling_shortwave_flux_in_air", "W/m2",
+  "uwind", "eastward_wind", "m/s",
+  "vwind", "northward_wind", "m/s",
+  "qair", "specific_humidity", "g/g",
+  "rain", "precipitation_flux", "kg/m2/s"
+)
+
+for (i in seq_len(rows)) {
+  year <- ylist[i]
+  ntime <- 1464
+  
+  loc.file <- file.path(outfolder, paste("CRUNCEP", year, "nc", sep = "."))
+  results$file[i] <- loc.file
+  results$host[i] <- "localhost"
+  results$startdate[i] <- paste0(year, "-01-01 00:00:00")
+  results$enddate[i] <- paste0(year, "-12-31 23:59:59")
+  results$mimetype[i] <- "application/x-netcdf"
+  results$formatname[i] <- "CF Meteorology"
+  
+  ## Create dimensions
+  lat <- ncdf4::ncdim_def(name = "latitude", units = "degree_north", vals = lat.in, create_dimvar = TRUE)
+  lon <- ncdf4::ncdim_def(name = "longitude", units = "degree_east", vals = lon.in, create_dimvar = TRUE)
+  
+  days_elapsed <- (1:ntime) * 6/24 - 3/24 # data are 6-hourly, with timestamp at center of interval
+  time <- ncdf4::ncdim_def(name = "time", units = paste0("days since ", year, "-01-01T00:00:00Z"),
+                           vals = as.array(days_elapsed), create_dimvar = TRUE, unlim = TRUE)
+  
+  dim <- list(lat, lon, time)
+  
+  var.list <- list()
+  dat.list <- list()
+  
+  file_name <- "mstmip_driver_global_hd_climate_%1$s_%2$d_v1.nc4"
+  ## get data off OpenDAP
+  
+  dap_base <- switch(
+    method,
+    opendap = paste0("https://thredds.daac.ornl.gov/thredds/dodsC/ornldaac/1220/", file_name),
+    ncss = paste0("https://thredds.daac.ornl.gov/thredds/ncss/grid/ornldaac/1220/", file_name, "/dataset.html")
+  )
+  
+  for (j in seq_len(nrow(var))) {
+    current_var <- var$DAP.name[j]
+    url <- sprintf(dap_base, current_var, year)
+    print(paste0("Attempting to access file at: ", url))
+    if (method == "opendap") {
+      dap <- PEcAn.utils::retry.func(ncdf4::nc_open(url, verbose=verbose), maxErrors=maxErrors, sleep=sleep)
+    } else if (method == "ncss") {
+      ncss_query <- glue::glue(
+        url, "?",
+        "var={current_var}&",
+        "south={lat.in}&",
+        "west={lon.in}&",
+        # Add tiny amount to latitude and longitude to satisfy
+        # non-point condition, but still be within grid cell.
+        "north={lat.in + 5e-6}&",
+        "east={lon.in + 5e-6}&",
+        # Year starts at 00:00:00 and ends at 21:00:00
+        "time_start={year}-01-01T00:00:00Z&",
+        "time_end={year}-12-31T21:00:00Z&",
+        "accept=netcdf"
+      )
+      # Cache raw CRUNCEP files so that later workflows don't have to download
+      # them (even if they do have to do some reprocessing).
+      raw_file <- file.path(
+        outfolder,
+        glue::glue("cruncep-raw-{year}-{lat.in}-{lon.in}-{current_var}.nc")
+      )
+      utils::download.file(ncss_query, raw_file)
+      dap <- ncdf4::nc_open(raw_file)
+    }
+    
+    # confirm that timestamps match
+    if (dap$dim$time$len != ntime) {
+      print(paste0("Expected", ntime, "observations, but", url,  "contained", dap$dim$time$len))
+    }
+    dap_time <- PEcAn.utils::ud_convert(dap$dim$time$vals,
+                                        dap$dim$time$units,
+                                        time$units)
+    if (!isTRUE(all.equal(dap_time, time$vals))){
+      print(paste0("Timestamp mismatch.",
+                   "Expected", min(time$vals), '..', max(time$vals), time$units,
+                   "but got", min(dap_time), "..", max(dap_time)))
+    }
+    
+    dat.list[[j]] <- PEcAn.utils::retry.func(
+      ncdf4::ncvar_get(
+        dap,
+        as.character(var$DAP.name[j]),
+        c(lon_grid, lat_grid, 1),
+        c(1, 1, ntime)
+      ),
+      maxErrors=maxErrors, sleep=sleep)
+    
+    var.list[[j]] <- ncdf4::ncvar_def(name = as.character(var$CF.name[j]),
+                                      units = as.character(var$units[j]),
+                                      dim = dim,
+                                      missval = -999,
+                                      verbose = verbose)
+    ncdf4::nc_close(dap)
+  }
+  ## change units of precip to kg/m2/s instead of 6 hour accumulated precip
+  dat.list[[8]] <- dat.list[[8]] / 21600
+  
+  ## put data in new file
+  loc <- ncdf4::nc_create(filename = loc.file, vars = var.list, verbose = verbose)
+  for (j in seq_len(nrow(var))) {
+    ncdf4::ncvar_put(nc = loc, varid = as.character(var$CF.name[j]), vals = dat.list[[j]])
+  }
+  ncdf4::nc_close(loc)
+}
+
+test_that("All the required files are downloaded and stored at desired location", {
+  # Cached raw CRUNCEP files 
+  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc")))
+  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc")))
+  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc")))
+  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc")))
+  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc")))
+  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc")))
+  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc")))
+  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc")))
+  
+  # CRUNCEP file
+  expect_true(file.exists(paste0(tmpdir, "/CRUNCEP.2000.nc")))
+})
\ No newline at end of file
diff --git a/modules/data.atmosphere/tests/testthat/test.integration.download.CRUNCEP.R b/modules/data.atmosphere/tests/testthat/test.integration.download.CRUNCEP.R
deleted file mode 100644
index a81cd2387c3..00000000000
--- a/modules/data.atmosphere/tests/testthat/test.integration.download.CRUNCEP.R
+++ /dev/null
@@ -1,47 +0,0 @@
-
-skip_on_ci()
-skip_if_offline()
-
-tmpdir <- tempfile(pattern = "CRUNCEPdata")
-dir.create(tmpdir)
-teardown(unlink(tmpdir, recursive = TRUE))
-
-outfolder  <- tmpdir
-start_date <- "2000-01-01"
-end_date   <- "2000-12-31"
-lat.in     <- 40
-lon.in     <- -88
-
-lat.in <- as.numeric(lat.in)
-lon.in <- as.numeric(lon.in)
-
-lat_grid <- floor(2 * (90 - lat.in)) + 1
-lon_grid <- floor(2 * (lon.in + 180)) + 1
-
-maskfile <- file.path(outfolder, "cruncep_landwater_mask.nc")
-mask_url <- paste0(
-  "https://thredds.daac.ornl.gov/thredds/ncss/ornldaac/1220/",
-  "mstmip_driver_global_hd_landwatermask_v1.nc4",
-  "?var=land_water_mask&disableLLSubset=on&disableProjSubset=on&horizStride=1&",
-  "accept=netcdf"
-)
-utils::download.file(mask_url, maskfile)
-
-mask_nc <- ncdf4::nc_open(maskfile)
-on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
-
-test_that("File exists at desired location", {
-  # Set the desired file path
-  file_path <- paste0(tmpdir, "/cruncep_landwater_mask.nc")
-  
-  # Check if file exists at desired location
-  expect_true(file.exists(file_path))
-})
-
-test_that("NetCDF file contains lat and lon variables", {
-  expect_true("land_water_mask" %in% names(mask_nc$var))
-  
-  # Check the dimensions of "land_water_mask" variable
-  expect_equal(mask_nc$var$land_water_mask$dim[[1]]$name, "lon")
-  expect_equal(mask_nc$var$land_water_mask$dim[[2]]$name, "lat")
-})
\ No newline at end of file

From d4721c03ef8d5e3b342c9313bd4854733ab4f201 Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Thu, 3 Aug 2023 16:58:02 +0530
Subject: [PATCH 03/12] added checks for variable units

---
 .../integrationTests/test.download.CRUNCEP.R  | 38 +++++++++++++++++--
 1 file changed, 35 insertions(+), 3 deletions(-)

diff --git a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
index 8970f867608..98310cf6545 100644
--- a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
+++ b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
@@ -2,9 +2,6 @@ library(testthat)
 library(ncdf4)
 library(lubridate)
 
-skip_on_ci()
-skip_if_offline()
-
 tmpdir <- tempfile(pattern = "CRUNCEPdata")
 dir.create(tmpdir)
 on.exit(teardown(unlink(tmpdir, recursive = TRUE)))
@@ -125,6 +122,7 @@ var <- tibble::tribble(
   "rain", "precipitation_flux", "kg/m2/s"
 )
 
+## NCSS
 for (i in seq_len(rows)) {
   year <- ylist[i]
   ntime <- 1464
@@ -243,4 +241,38 @@ test_that("All the required files are downloaded and stored at desired location"
   
   # CRUNCEP file
   expect_true(file.exists(paste0(tmpdir, "/CRUNCEP.2000.nc")))
+})
+
+test_that("All cruncep raw data files have the correct variable units", {
+  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc"))
+  expect_equal(nc$var$tair$units, "K")
+  nc_close(nc)
+
+  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc"))
+  expect_equal(nc$var$lwdown$units, "W/m2")
+  nc_close(nc)
+
+  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc"))
+  expect_equal(nc$var$press$units, "Pa")
+  nc_close(nc)
+
+  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc"))
+  expect_equal(nc$var$swdown$units, "W/m2")
+  nc_close(nc)
+
+  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc"))
+  expect_equal(nc$var$uwind$units, "m/s")
+  nc_close(nc)
+
+  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc"))
+  expect_equal(nc$var$vwind$units, "m/s")
+  nc_close(nc)
+
+  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc"))
+  expect_equal(nc$var$qair$units, "g/g")
+  nc_close(nc)
+
+  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc"))
+  expect_equal(nc$var$rain$units, "mm/6h")
+  nc_close(nc)
 })
\ No newline at end of file

From 641297aa6ebf06cea1a1fda89fff4a85dc16a26d Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Thu, 3 Aug 2023 19:01:08 +0530
Subject: [PATCH 04/12] clean up of some unwanted code segments

---
 .../integrationTests/test.download.CRUNCEP.R  | 38 ++++++++-----------
 1 file changed, 15 insertions(+), 23 deletions(-)

diff --git a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
index 98310cf6545..750fb2f9ec4 100644
--- a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
+++ b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
@@ -101,14 +101,6 @@ end_year <- lubridate::year(end_date)
 # year - 2000
 ylist <- seq(start_year, end_year, by = 1)
 rows <- length(ylist)
-results <- data.frame(file = character(rows),
-                      host = character(rows),
-                      mimetype = character(rows),
-                      formatname = character(rows),
-                      startdate = character(rows),
-                      enddate = character(rows),
-                      dbfile.name = "CRUNCEP",
-                      stringsAsFactors = FALSE)
 
 var <- tibble::tribble(
   ~DAP.name, ~CF.name, ~units,
@@ -122,19 +114,12 @@ var <- tibble::tribble(
   "rain", "precipitation_flux", "kg/m2/s"
 )
 
-## NCSS
 for (i in seq_len(rows)) {
   year <- ylist[i]
   ntime <- 1464
   
   loc.file <- file.path(outfolder, paste("CRUNCEP", year, "nc", sep = "."))
-  results$file[i] <- loc.file
-  results$host[i] <- "localhost"
-  results$startdate[i] <- paste0(year, "-01-01 00:00:00")
-  results$enddate[i] <- paste0(year, "-12-31 23:59:59")
-  results$mimetype[i] <- "application/x-netcdf"
-  results$formatname[i] <- "CF Meteorology"
-  
+
   ## Create dimensions
   lat <- ncdf4::ncdim_def(name = "latitude", units = "degree_north", vals = lat.in, create_dimvar = TRUE)
   lon <- ncdf4::ncdim_def(name = "longitude", units = "degree_east", vals = lon.in, create_dimvar = TRUE)
@@ -149,7 +134,6 @@ for (i in seq_len(rows)) {
   dat.list <- list()
   
   file_name <- "mstmip_driver_global_hd_climate_%1$s_%2$d_v1.nc4"
-  ## get data off OpenDAP
   
   dap_base <- switch(
     method,
@@ -169,17 +153,13 @@ for (i in seq_len(rows)) {
         "var={current_var}&",
         "south={lat.in}&",
         "west={lon.in}&",
-        # Add tiny amount to latitude and longitude to satisfy
-        # non-point condition, but still be within grid cell.
         "north={lat.in + 5e-6}&",
         "east={lon.in + 5e-6}&",
-        # Year starts at 00:00:00 and ends at 21:00:00
         "time_start={year}-01-01T00:00:00Z&",
         "time_end={year}-12-31T21:00:00Z&",
         "accept=netcdf"
       )
-      # Cache raw CRUNCEP files so that later workflows don't have to download
-      # them (even if they do have to do some reprocessing).
+
       raw_file <- file.path(
         outfolder,
         glue::glue("cruncep-raw-{year}-{lat.in}-{lon.in}-{current_var}.nc")
@@ -188,7 +168,6 @@ for (i in seq_len(rows)) {
       dap <- ncdf4::nc_open(raw_file)
     }
     
-    # confirm that timestamps match
     if (dap$dim$time$len != ntime) {
       print(paste0("Expected", ntime, "observations, but", url,  "contained", dap$dim$time$len))
     }
@@ -275,4 +254,17 @@ test_that("All cruncep raw data files have the correct variable units", {
   nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc"))
   expect_equal(nc$var$rain$units, "mm/6h")
   nc_close(nc)
+})
+
+test_that("Combined data file has the correct variable units", {
+  nc <- nc_open(paste0(tmpdir, "/CRUNCEP.2000.nc"))
+  expect_equal(nc$var$air_temperature$units, "Kelvin")
+  expect_equal(nc$var$surface_downwelling_longwave_flux_in_air$units, "W/m2")
+  expect_equal(nc$var$air_pressure$units, "Pascal")
+  expect_equal(nc$var$surface_downwelling_shortwave_flux_in_air$units, "W/m2")
+  expect_equal(nc$var$eastward_wind$units, "m/s")
+  expect_equal(nc$var$northward_wind$units, "m/s")
+  expect_equal(nc$var$specific_humidity$units, "g/g")
+  expect_equal(nc$var$precipitation_flux$units, "kg/m2/s")
+  nc_close(nc)
 })
\ No newline at end of file

From 7ac83a30bbb5df8388f9b4e7da6557950b0860fc Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Fri, 18 Aug 2023 19:05:48 +0530
Subject: [PATCH 05/12] removed duplicate need_packages

---
 modules/data.atmosphere/R/download.ERA5.R |  2 +-
 modules/data.atmosphere/R/need_packages.R | 22 ----------------------
 2 files changed, 1 insertion(+), 23 deletions(-)
 delete mode 100644 modules/data.atmosphere/R/need_packages.R

diff --git a/modules/data.atmosphere/R/download.ERA5.R b/modules/data.atmosphere/R/download.ERA5.R
index 59d1d134a5f..533aeac96fc 100644
--- a/modules/data.atmosphere/R/download.ERA5.R
+++ b/modules/data.atmosphere/R/download.ERA5.R
@@ -51,7 +51,7 @@ download.ERA5.old <- function(outfolder, start_date, end_date, lat.in, lon.in,
     "This function is an incomplete prototype! Use with caution!"
   )
 
-  need_packages("reticulate")
+  PEcAn.utils::need_packages("reticulate")
   
   if (!is.null(reticulate_python)) {
     reticulate::use_python(reticulate_python)
diff --git a/modules/data.atmosphere/R/need_packages.R b/modules/data.atmosphere/R/need_packages.R
deleted file mode 100644
index eb895acd6e8..00000000000
--- a/modules/data.atmosphere/R/need_packages.R
+++ /dev/null
@@ -1,22 +0,0 @@
-#' Check if required packages are installed, and throw an informative
-#' error if not.
-#'
-#' @param ... Package names, as characters. Can be passed as
-#'   individual arguments, character vectors, or any combination thereof.
-#' @return `pkgs`, invisibly
-#' @author Alexey Shiklomanov
-#' @examples
-#' PEcAn.data.atmosphere:::need_packages("stats", "methods") # Always works 
-#' try(PEcAn.data.atmosphere:::need_packages("notapackage"))
-need_packages <- function(...) {
-  pkgs <- unlist(list(...), recursive = TRUE)
-  have <- vapply(pkgs, requireNamespace, logical(1), quietly = TRUE)
-  if (any(!have)) {
-    msg <- sprintf(
-      "The following packages are required but not installed: %s",
-      paste0("`", pkgs, "`", collapse = ", ")
-    )
-    PEcAn.logger::logger.severe(msg)
-  }
-  invisible(pkgs)
-}

From 0dbbdf8560e690ac3c71772c059555d846e8ac21 Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Fri, 18 Aug 2023 19:09:47 +0530
Subject: [PATCH 06/12] removed .Rd

---
 modules/data.atmosphere/man/need_packages.Rd | 27 --------------------
 1 file changed, 27 deletions(-)
 delete mode 100644 modules/data.atmosphere/man/need_packages.Rd

diff --git a/modules/data.atmosphere/man/need_packages.Rd b/modules/data.atmosphere/man/need_packages.Rd
deleted file mode 100644
index 6a40181ad96..00000000000
--- a/modules/data.atmosphere/man/need_packages.Rd
+++ /dev/null
@@ -1,27 +0,0 @@
-% Generated by roxygen2: do not edit by hand
-% Please edit documentation in R/need_packages.R
-\name{need_packages}
-\alias{need_packages}
-\title{Check if required packages are installed, and throw an informative
-error if not.}
-\usage{
-need_packages(...)
-}
-\arguments{
-\item{...}{Package names, as characters. Can be passed as
-individual arguments, character vectors, or any combination thereof.}
-}
-\value{
-`pkgs`, invisibly
-}
-\description{
-Check if required packages are installed, and throw an informative
-error if not.
-}
-\examples{
-PEcAn.data.atmosphere:::need_packages("stats", "methods") # Always works 
-try(PEcAn.data.atmosphere:::need_packages("notapackage"))
-}
-\author{
-Alexey Shiklomanov
-}

From 08a8793aca74ba3d57f87084ca976c7f281b938b Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Mon, 21 Aug 2023 15:01:06 +0530
Subject: [PATCH 07/12] updated integration test

---
 .../integrationTests/test.download.CRUNCEP.R  | 215 +++---------------
 1 file changed, 31 insertions(+), 184 deletions(-)

diff --git a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
index 750fb2f9ec4..dbabb91be5c 100644
--- a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
+++ b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
@@ -1,39 +1,39 @@
 library(testthat)
 library(ncdf4)
-library(lubridate)
+
+# putting logger to debug mode
+PEcAn.logger::logger.setUseConsole(TRUE, FALSE)
+on.exit(PEcAn.logger::logger.setUseConsole(TRUE, TRUE), add = TRUE)
+PEcAn.logger::logger.setLevel("DEBUG")
+
+# mocking functions
+mockery::stub(PEcAn.DB::convert_input, 'dbfile.input.check', data.frame())
+mockery::stub(PEcAn.DB::convert_input, 'db.query', data.frame(id = 1))
 
 tmpdir <- tempfile(pattern = "CRUNCEPdata")
 dir.create(tmpdir)
 on.exit(teardown(unlink(tmpdir, recursive = TRUE)))
 
-outfolder  <- tmpdir
-start_date <- "2000-01-01"
-end_date   <- "2000-12-31"
-lat.in     <- 40
-lon.in     <- -88
-method     <- "ncss"
-overwrite  <- TRUE
-verbose <- FALSE
-maxErrors <- 10
-sleep <- 2
-
-lat.in <- as.numeric(lat.in)
-lon.in <- as.numeric(lon.in)
-
-lat_grid <- floor(2 * (90 - lat.in)) + 1
-lon_grid <- floor(2 * (lon.in + 180)) + 1
-
-maskfile <- file.path(outfolder, "cruncep_landwater_mask.nc")
-mask_url <- paste0(
-  "https://thredds.daac.ornl.gov/thredds/ncss/ornldaac/1220/",
-  "mstmip_driver_global_hd_landwatermask_v1.nc4",
-  "?var=land_water_mask&disableLLSubset=on&disableProjSubset=on&horizStride=1&",
-  "accept=netcdf"
+PEcAn.DB::convert_input(
+  input.id = NA,
+  outfolder = tmpdir,
+  formatname = NULL,
+  mimetype = NULL,
+  site.id = 1,
+  start_date = "2000-01-01",
+  end_date = "2000-12-31",
+  pkg = 'PEcAn.data.atmosphere',
+  fcn = 'download.CRUNCEP',
+  con = NULL,
+  host = data.frame(name = "localhost"),
+  browndog = NULL,
+  write = FALSE,
+  lat.in = 40,
+  lon.in = -88,
+  method = "ncss",
+  maxErrors = 10,
+  sleep = 2
 )
-utils::download.file(mask_url, maskfile)
-
-mask_nc <- ncdf4::nc_open(maskfile)
-on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
 
 test_that("File exists at desired location", {
   # Set the desired file path
@@ -44,6 +44,9 @@ test_that("File exists at desired location", {
 })
 
 test_that("NetCDF file contains lat and lon variables", {
+
+  mask_nc <- ncdf4::nc_open(paste0(tmpdir, "/cruncep_landwater_mask.nc"))
+  on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
   expect_true("land_water_mask" %in% names(mask_nc$var))
   
   # Check the dimensions of "land_water_mask" variable
@@ -51,162 +54,6 @@ test_that("NetCDF file contains lat and lon variables", {
   expect_equal(mask_nc$var$land_water_mask$dim[[2]]$name, "lat")
 })
 
-
-# Set search radius to up to 2 pixels (1 degree) in any direction
-mask_minlat <- lat_grid - 2
-mask_minlon <- lon_grid - 2
-mask_lats <- ncdf4::ncvar_get(mask_nc, "lat", start = mask_minlat, count = 5)
-mask_lons <- ncdf4::ncvar_get(mask_nc, "lon", start = mask_minlon, count = 5)
-mask_values <- ncdf4::ncvar_get(
-  mask_nc,
-  "land_water_mask",
-  start = c(mask_minlon, mask_minlat),
-  count = c(5, 5)
-)
-
-# Build a lat-lon grid matrix and calculate distance from target coords
-mask_grid <- as.matrix(expand.grid(lon = mask_lons, lat = mask_lats))
-mask_igrid <- as.matrix(expand.grid(lon = seq_along(mask_lons), lats = seq_along(mask_lats)))
-mask_dist <- (lon.in - mask_grid[, 1])^2 + (lat.in - mask_grid[, 2])^2
-
-# Order by increasing distance (closest first)
-mask_order <- order(mask_dist)
-mask_igrido <- mask_igrid[mask_order,]
-on_land <- as.logical(mask_values[mask_igrido])
-igrid <- which(on_land)[1]
-if (igrid > 1) {
-  lon.in.orig <- lon.in
-  lat.in.orig <- lat.in
-  lon.in <- mask_grid[mask_order[igrid], 1]
-  lat.in <- mask_grid[mask_order[igrid], 2]
-}
-
-if (method == "opendap") {
-  # Convert lat-lon to grid row and column
-  lat_grid <- floor(2 * (90 - lat.in)) + 1
-  lon_grid <- floor(2 * (lon.in + 180)) + 1
-} else if (method == "ncss") {
-  # Only downloading point data, so only one point
-  lat_grid <- 1
-  lon_grid <- 1
-}
-
-start_date <- as.POSIXlt(start_date, tz = "UTC")
-end_date <- as.POSIXlt(end_date, tz = "UTC")
-
-# taking out year from dates
-start_year <- lubridate::year(start_date)
-end_year <- lubridate::year(end_date)
-
-# year - 2000
-ylist <- seq(start_year, end_year, by = 1)
-rows <- length(ylist)
-
-var <- tibble::tribble(
-  ~DAP.name, ~CF.name, ~units,
-  "tair", "air_temperature", "Kelvin",
-  "lwdown", "surface_downwelling_longwave_flux_in_air", "W/m2",
-  "press", "air_pressure", "Pascal",
-  "swdown", "surface_downwelling_shortwave_flux_in_air", "W/m2",
-  "uwind", "eastward_wind", "m/s",
-  "vwind", "northward_wind", "m/s",
-  "qair", "specific_humidity", "g/g",
-  "rain", "precipitation_flux", "kg/m2/s"
-)
-
-for (i in seq_len(rows)) {
-  year <- ylist[i]
-  ntime <- 1464
-  
-  loc.file <- file.path(outfolder, paste("CRUNCEP", year, "nc", sep = "."))
-
-  ## Create dimensions
-  lat <- ncdf4::ncdim_def(name = "latitude", units = "degree_north", vals = lat.in, create_dimvar = TRUE)
-  lon <- ncdf4::ncdim_def(name = "longitude", units = "degree_east", vals = lon.in, create_dimvar = TRUE)
-  
-  days_elapsed <- (1:ntime) * 6/24 - 3/24 # data are 6-hourly, with timestamp at center of interval
-  time <- ncdf4::ncdim_def(name = "time", units = paste0("days since ", year, "-01-01T00:00:00Z"),
-                           vals = as.array(days_elapsed), create_dimvar = TRUE, unlim = TRUE)
-  
-  dim <- list(lat, lon, time)
-  
-  var.list <- list()
-  dat.list <- list()
-  
-  file_name <- "mstmip_driver_global_hd_climate_%1$s_%2$d_v1.nc4"
-  
-  dap_base <- switch(
-    method,
-    opendap = paste0("https://thredds.daac.ornl.gov/thredds/dodsC/ornldaac/1220/", file_name),
-    ncss = paste0("https://thredds.daac.ornl.gov/thredds/ncss/grid/ornldaac/1220/", file_name, "/dataset.html")
-  )
-  
-  for (j in seq_len(nrow(var))) {
-    current_var <- var$DAP.name[j]
-    url <- sprintf(dap_base, current_var, year)
-    print(paste0("Attempting to access file at: ", url))
-    if (method == "opendap") {
-      dap <- PEcAn.utils::retry.func(ncdf4::nc_open(url, verbose=verbose), maxErrors=maxErrors, sleep=sleep)
-    } else if (method == "ncss") {
-      ncss_query <- glue::glue(
-        url, "?",
-        "var={current_var}&",
-        "south={lat.in}&",
-        "west={lon.in}&",
-        "north={lat.in + 5e-6}&",
-        "east={lon.in + 5e-6}&",
-        "time_start={year}-01-01T00:00:00Z&",
-        "time_end={year}-12-31T21:00:00Z&",
-        "accept=netcdf"
-      )
-
-      raw_file <- file.path(
-        outfolder,
-        glue::glue("cruncep-raw-{year}-{lat.in}-{lon.in}-{current_var}.nc")
-      )
-      utils::download.file(ncss_query, raw_file)
-      dap <- ncdf4::nc_open(raw_file)
-    }
-    
-    if (dap$dim$time$len != ntime) {
-      print(paste0("Expected", ntime, "observations, but", url,  "contained", dap$dim$time$len))
-    }
-    dap_time <- PEcAn.utils::ud_convert(dap$dim$time$vals,
-                                        dap$dim$time$units,
-                                        time$units)
-    if (!isTRUE(all.equal(dap_time, time$vals))){
-      print(paste0("Timestamp mismatch.",
-                   "Expected", min(time$vals), '..', max(time$vals), time$units,
-                   "but got", min(dap_time), "..", max(dap_time)))
-    }
-    
-    dat.list[[j]] <- PEcAn.utils::retry.func(
-      ncdf4::ncvar_get(
-        dap,
-        as.character(var$DAP.name[j]),
-        c(lon_grid, lat_grid, 1),
-        c(1, 1, ntime)
-      ),
-      maxErrors=maxErrors, sleep=sleep)
-    
-    var.list[[j]] <- ncdf4::ncvar_def(name = as.character(var$CF.name[j]),
-                                      units = as.character(var$units[j]),
-                                      dim = dim,
-                                      missval = -999,
-                                      verbose = verbose)
-    ncdf4::nc_close(dap)
-  }
-  ## change units of precip to kg/m2/s instead of 6 hour accumulated precip
-  dat.list[[8]] <- dat.list[[8]] / 21600
-  
-  ## put data in new file
-  loc <- ncdf4::nc_create(filename = loc.file, vars = var.list, verbose = verbose)
-  for (j in seq_len(nrow(var))) {
-    ncdf4::ncvar_put(nc = loc, varid = as.character(var$CF.name[j]), vals = dat.list[[j]])
-  }
-  ncdf4::nc_close(loc)
-}
-
 test_that("All the required files are downloaded and stored at desired location", {
   # Cached raw CRUNCEP files 
   expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc")))

From b25e18688815d2397b884cfcd2c9b06d2b7bf187 Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Mon, 21 Aug 2023 18:30:05 +0530
Subject: [PATCH 08/12] fixed typos

---
 modules/data.atmosphere/R/download.US_WCr.R     | 2 +-
 modules/data.atmosphere/R/download.US_Wlef.R    | 2 +-
 modules/data.atmosphere/R/extract_ERA5.R        | 2 +-
 modules/data.atmosphere/man/download.US_WCr.Rd  | 2 +-
 modules/data.atmosphere/man/download.US_Wlef.Rd | 2 +-
 modules/data.atmosphere/man/extract.nc.ERA5.Rd  | 2 +-
 6 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/modules/data.atmosphere/R/download.US_WCr.R b/modules/data.atmosphere/R/download.US_WCr.R
index 3924311f526..66615e6ef44 100644
--- a/modules/data.atmosphere/R/download.US_WCr.R
+++ b/modules/data.atmosphere/R/download.US_WCr.R
@@ -2,7 +2,7 @@
 ##' 
 ##' @section General Description:
 ##' Obtains data from Ankur Desai's Willow Creek flux tower, and selects certain variables (NEE and LE) to return
-##' Data is retruned at the given timestep in the given range.
+##' Data is returned at the given timestep in the given range.
 ##' 
 ##' This data includes information on a number of flux variables.
 ##' 
diff --git a/modules/data.atmosphere/R/download.US_Wlef.R b/modules/data.atmosphere/R/download.US_Wlef.R
index a92e7599e31..4d44c880f1a 100644
--- a/modules/data.atmosphere/R/download.US_Wlef.R
+++ b/modules/data.atmosphere/R/download.US_Wlef.R
@@ -2,7 +2,7 @@
 ##' 
 ##' @section General Description:
 ##' Obtains data from Ankur Desai's WLEF/ Parks Fall flux tower, and selects certain variables (NEE and LE) to return
-##' Data is retruned at the given timestep in the given range.
+##' Data is returned at the given timestep in the given range.
 ##' 
 ##' This data includes information on a number of flux variables.
 ##' 
diff --git a/modules/data.atmosphere/R/extract_ERA5.R b/modules/data.atmosphere/R/extract_ERA5.R
index bf219b4a509..f85b43cdec8 100644
--- a/modules/data.atmosphere/R/extract_ERA5.R
+++ b/modules/data.atmosphere/R/extract_ERA5.R
@@ -8,7 +8,7 @@
 #' @param outfolder Path to directory where nc files need to be saved.
 #' @param in.prefix initial portion of the filename that does not vary by date. Does not include directory; specify that as part of in.path.
 #' @param newsite site name.
-#' @param vars variables to be extarcted. If NULL all the variables will be returned.
+#' @param vars variables to be extracted. If NULL all the variables will be returned.
 #' @param overwrite Logical if files needs to be overwritten.
 #' @details For the list of variables check out the documentation at \link{https://confluence.ecmwf.int/display/CKB/ERA5+data+documentation#ERA5datadocumentation-Spatialgrid}
 #'
diff --git a/modules/data.atmosphere/man/download.US_WCr.Rd b/modules/data.atmosphere/man/download.US_WCr.Rd
index 3c3fcfa99d1..c09d4eda31a 100644
--- a/modules/data.atmosphere/man/download.US_WCr.Rd
+++ b/modules/data.atmosphere/man/download.US_WCr.Rd
@@ -19,7 +19,7 @@ download.US-WCr
 \section{General Description}{
 
 Obtains data from Ankur Desai's Willow Creek flux tower, and selects certain variables (NEE and LE) to return
-Data is retruned at the given timestep in the given range.
+Data is returned at the given timestep in the given range.
 
 This data includes information on a number of flux variables.
 
diff --git a/modules/data.atmosphere/man/download.US_Wlef.Rd b/modules/data.atmosphere/man/download.US_Wlef.Rd
index fd461d65299..f20ffdc1fc7 100644
--- a/modules/data.atmosphere/man/download.US_Wlef.Rd
+++ b/modules/data.atmosphere/man/download.US_Wlef.Rd
@@ -19,7 +19,7 @@ download.US_Wlef
 \section{General Description}{
 
 Obtains data from Ankur Desai's WLEF/ Parks Fall flux tower, and selects certain variables (NEE and LE) to return
-Data is retruned at the given timestep in the given range.
+Data is returned at the given timestep in the given range.
 
 This data includes information on a number of flux variables.
 }
diff --git a/modules/data.atmosphere/man/extract.nc.ERA5.Rd b/modules/data.atmosphere/man/extract.nc.ERA5.Rd
index a5c6ce66081..8b82e4490ba 100644
--- a/modules/data.atmosphere/man/extract.nc.ERA5.Rd
+++ b/modules/data.atmosphere/man/extract.nc.ERA5.Rd
@@ -35,7 +35,7 @@ extract.nc.ERA5(
 
 \item{newsite}{site name.}
 
-\item{vars}{variables to be extarcted. If NULL all the variables will be returned.}
+\item{vars}{variables to be extracted. If NULL all the variables will be returned.}
 
 \item{overwrite}{Logical if files needs to be overwritten.}
 }

From 5484b780ec7088d654d0dd452f627019941d0d45 Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Mon, 21 Aug 2023 19:16:57 +0530
Subject: [PATCH 09/12] fixed typos

---
 modules/data.atmosphere/R/download.AmerifluxLBL.R            | 2 +-
 modules/data.atmosphere/R/extract_local_CMIP5.R              | 2 +-
 modules/data.atmosphere/R/extract_local_NLDAS.R              | 2 +-
 modules/data.atmosphere/R/tdm_predict_subdaily_met.R         | 2 +-
 modules/data.atmosphere/R/tdm_temporal_downscale_functions.R | 2 +-
 modules/data.atmosphere/man/download.AmerifluxLBL.Rd         | 2 +-
 modules/data.atmosphere/man/extract.local.CMIP5.Rd           | 2 +-
 modules/data.atmosphere/man/extract.local.NLDAS.Rd           | 2 +-
 modules/data.atmosphere/man/predict_subdaily_met.Rd          | 2 +-
 modules/data.atmosphere/man/temporal.downscale.functions.Rd  | 2 +-
 10 files changed, 10 insertions(+), 10 deletions(-)

diff --git a/modules/data.atmosphere/R/download.AmerifluxLBL.R b/modules/data.atmosphere/R/download.AmerifluxLBL.R
index 2bb5b2b65b1..4fe706a1f33 100644
--- a/modules/data.atmosphere/R/download.AmerifluxLBL.R
+++ b/modules/data.atmosphere/R/download.AmerifluxLBL.R
@@ -5,7 +5,7 @@
 ##' to the given outfolder. Details about amf_download_base function can be found here:
 ##' https://github.com/chuhousen/amerifluxr/blob/master/R/amf_download_base.R
 ##' 
-##' Uses Ameirflux LBL JSON API to download met data from Ameriflux towers in CSV format
+##' Uses Ameriflux LBL JSON API to download met data from Ameriflux towers in CSV format
 ##' 
 ##' @export
 ##' @param sitename the Ameriflux ID of the site to be downloaded, used as file name prefix. 
diff --git a/modules/data.atmosphere/R/extract_local_CMIP5.R b/modules/data.atmosphere/R/extract_local_CMIP5.R
index 45952de9d8a..8cadc9bf16b 100644
--- a/modules/data.atmosphere/R/extract_local_CMIP5.R
+++ b/modules/data.atmosphere/R/extract_local_CMIP5.R
@@ -24,7 +24,7 @@
 ##' @param date.origin (optional) specify the date of origin for timestamps in the files being read.  
 ##'                    If NULL defaults to 1850 for historical simulations (except MPI-ESM-P) and 
 ##'                    850 for p1000 simulations (plus MPI-ESM-P historical).  Format: YYYY-MM-DD
-##' @param adjust.pr - adjustment factor fore preciptiation when the extracted values seem off
+##' @param adjust.pr - adjustment factor fore precipitation when the extracted values seem off
 ##' @param overwrite logical. Download a fresh version even if a local file with the same name already exists?
 ##' @param verbose logical. to control printing of debug info
 ##' @param ... Other arguments, currently ignored
diff --git a/modules/data.atmosphere/R/extract_local_NLDAS.R b/modules/data.atmosphere/R/extract_local_NLDAS.R
index a23c4d74531..2846dafca09 100644
--- a/modules/data.atmosphere/R/extract_local_NLDAS.R
+++ b/modules/data.atmosphere/R/extract_local_NLDAS.R
@@ -7,7 +7,7 @@
 ##' @description This function extracts NLDAS data from grids that have been downloaded and stored locally.
 ##'              Once upon a time, you could query these files directly from the internet, but now they're 
 ##'              behind a tricky authentication wall. Files are saved as a netCDF file in CF conventions. 
-##'              These files are ready to be used in the general PEcAn workflow or fed into the downscalign 
+##'              These files are ready to be used in the general PEcAn workflow or fed into the downscaling 
 ##'              workflow.
 # ----------------------------------- 
 # Parameters
diff --git a/modules/data.atmosphere/R/tdm_predict_subdaily_met.R b/modules/data.atmosphere/R/tdm_predict_subdaily_met.R
index 58c3d72845e..825bd203672 100644
--- a/modules/data.atmosphere/R/tdm_predict_subdaily_met.R
+++ b/modules/data.atmosphere/R/tdm_predict_subdaily_met.R
@@ -30,7 +30,7 @@
 ##' @param ens.labs - vector containing the labels (suffixes) for each ensemble member; this allows you to add to your 
 ##'                   ensemble rather than overwriting with a default naming scheme
 ##' @param resids - logical stating whether to pass on residual data or not
-##' @param adjust.pr - adjustment factor fore preciptiation when the extracted values seem off
+##' @param adjust.pr - adjustment factor fore precipitation when the extracted values seem off
 ##' @param force.sanity - (logical) do we force the data to meet sanity checks?                             
 ##' @param sanity.tries - how many time should we try to predict a reasonable value before giving up?  We don't want to end up in an infinite loop
 ##' @param overwrite logical: replace output file if it already exists?
diff --git a/modules/data.atmosphere/R/tdm_temporal_downscale_functions.R b/modules/data.atmosphere/R/tdm_temporal_downscale_functions.R
index 3866ad36e99..911b1ba4014 100644
--- a/modules/data.atmosphere/R/tdm_temporal_downscale_functions.R
+++ b/modules/data.atmosphere/R/tdm_temporal_downscale_functions.R
@@ -19,7 +19,7 @@
 # -----------------------------------
 ##' @param dat.train - training data generated by tdm_nc2dat.train.R
 ##' @param n.beta - number of betas to generate
-##' @param resids - whether or not to propogate residuals, set to FALSE
+##' @param resids - whether or not to propagate residuals, set to FALSE
 ##' @param parallel - whether or not to run in parallel. this is a feature
 ##'                   still being worked on, set to FALSE
 ##' @param n.cores - number of cores to use parallel processing on, set to NULL
diff --git a/modules/data.atmosphere/man/download.AmerifluxLBL.Rd b/modules/data.atmosphere/man/download.AmerifluxLBL.Rd
index 753bcd5457f..8b30000484f 100644
--- a/modules/data.atmosphere/man/download.AmerifluxLBL.Rd
+++ b/modules/data.atmosphere/man/download.AmerifluxLBL.Rd
@@ -50,7 +50,7 @@ to the given outfolder. Details about amf_download_base function can be found he
 https://github.com/chuhousen/amerifluxr/blob/master/R/amf_download_base.R
 }
 \details{
-Uses Ameirflux LBL JSON API to download met data from Ameriflux towers in CSV format
+Uses Ameriflux LBL JSON API to download met data from Ameriflux towers in CSV format
 }
 \examples{
 \dontrun{
diff --git a/modules/data.atmosphere/man/extract.local.CMIP5.Rd b/modules/data.atmosphere/man/extract.local.CMIP5.Rd
index fec601e2925..14eb0142c7e 100644
--- a/modules/data.atmosphere/man/extract.local.CMIP5.Rd
+++ b/modules/data.atmosphere/man/extract.local.CMIP5.Rd
@@ -45,7 +45,7 @@ extract.local.CMIP5(
 If NULL defaults to 1850 for historical simulations (except MPI-ESM-P) and 
 850 for p1000 simulations (plus MPI-ESM-P historical).  Format: YYYY-MM-DD}
 
-\item{adjust.pr}{- adjustment factor fore preciptiation when the extracted values seem off}
+\item{adjust.pr}{- adjustment factor fore precipitation when the extracted values seem off}
 
 \item{overwrite}{logical. Download a fresh version even if a local file with the same name already exists?}
 
diff --git a/modules/data.atmosphere/man/extract.local.NLDAS.Rd b/modules/data.atmosphere/man/extract.local.NLDAS.Rd
index e98ba47e292..b9a9f2e88b8 100644
--- a/modules/data.atmosphere/man/extract.local.NLDAS.Rd
+++ b/modules/data.atmosphere/man/extract.local.NLDAS.Rd
@@ -41,7 +41,7 @@ to control printing of debug info}
 This function extracts NLDAS data from grids that have been downloaded and stored locally.
              Once upon a time, you could query these files directly from the internet, but now they're 
              behind a tricky authentication wall. Files are saved as a netCDF file in CF conventions. 
-             These files are ready to be used in the general PEcAn workflow or fed into the downscalign 
+             These files are ready to be used in the general PEcAn workflow or fed into the downscaling 
              workflow.
 }
 \author{
diff --git a/modules/data.atmosphere/man/predict_subdaily_met.Rd b/modules/data.atmosphere/man/predict_subdaily_met.Rd
index 3d4898e3e6f..09a98b8ce0c 100644
--- a/modules/data.atmosphere/man/predict_subdaily_met.Rd
+++ b/modules/data.atmosphere/man/predict_subdaily_met.Rd
@@ -47,7 +47,7 @@ ensemble rather than overwriting with a default naming scheme}
 
 \item{resids}{- logical stating whether to pass on residual data or not}
 
-\item{adjust.pr}{- adjustment factor fore preciptiation when the extracted values seem off}
+\item{adjust.pr}{- adjustment factor fore precipitation when the extracted values seem off}
 
 \item{force.sanity}{- (logical) do we force the data to meet sanity checks?}
 
diff --git a/modules/data.atmosphere/man/temporal.downscale.functions.Rd b/modules/data.atmosphere/man/temporal.downscale.functions.Rd
index 77e86d6c3a7..668fe14e60a 100644
--- a/modules/data.atmosphere/man/temporal.downscale.functions.Rd
+++ b/modules/data.atmosphere/man/temporal.downscale.functions.Rd
@@ -25,7 +25,7 @@ temporal.downscale.functions(
 \item{day.window}{- number of days surrounding current day we want to pull
 statistics from}
 
-\item{resids}{- whether or not to propogate residuals, set to FALSE}
+\item{resids}{- whether or not to propagate residuals, set to FALSE}
 
 \item{parallel}{- whether or not to run in parallel. this is a feature
 still being worked on, set to FALSE}

From e68f54060a67b29189f6b9289f340f880306c83c Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Tue, 22 Aug 2023 14:34:47 +0530
Subject: [PATCH 10/12] fixed typos

---
 modules/data.atmosphere/R/metgapfill.R    | 4 ++--
 modules/data.atmosphere/man/metgapfill.Rd | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/modules/data.atmosphere/R/metgapfill.R b/modules/data.atmosphere/R/metgapfill.R
index 2d3eba7f096..db53eebe875 100644
--- a/modules/data.atmosphere/R/metgapfill.R
+++ b/modules/data.atmosphere/R/metgapfill.R
@@ -5,14 +5,14 @@
 ##' Future version will first downscale and fill with NARR, then REddyProc
 ##'
 ##' @export
-##' @param in.path location on disk where inputs are stord
+##' @param in.path location on disk where inputs are stored
 ##' @param in.prefix prefix of input and output files
 ##' @param outfolder location on disk where outputs will be stored
 ##' @param start_date the start date of the data to be downloaded (will only use the year part of the date)
 ##' @param end_date the end date of the data to be downloaded (will only use the year part of the date)
 ##' @param overwrite should existing files be overwritten
 ##' @param verbose should the function be very verbose
-##' @param lst is timezone offset from UTC, if timezone is available in time:units atribute in file, it will use that, default is to assume UTC
+##' @param lst is timezone offset from UTC, if timezone is available in time:units attribute in file, it will use that, default is to assume UTC
 ##' @author Ankur Desai
 metgapfill <- function(in.path, in.prefix, outfolder, start_date, end_date, lst = 0,
                        overwrite = FALSE, verbose = FALSE, ...) {
diff --git a/modules/data.atmosphere/man/metgapfill.Rd b/modules/data.atmosphere/man/metgapfill.Rd
index e5a7d37a385..5c96abb1525 100644
--- a/modules/data.atmosphere/man/metgapfill.Rd
+++ b/modules/data.atmosphere/man/metgapfill.Rd
@@ -21,7 +21,7 @@ metgapfill(
 )
 }
 \arguments{
-\item{in.path}{location on disk where inputs are stord}
+\item{in.path}{location on disk where inputs are stored}
 
 \item{in.prefix}{prefix of input and output files}
 
@@ -31,7 +31,7 @@ metgapfill(
 
 \item{end_date}{the end date of the data to be downloaded (will only use the year part of the date)}
 
-\item{lst}{is timezone offset from UTC, if timezone is available in time:units atribute in file, it will use that, default is to assume UTC}
+\item{lst}{is timezone offset from UTC, if timezone is available in time:units attribute in file, it will use that, default is to assume UTC}
 
 \item{overwrite}{should existing files be overwritten}
 

From 6b1c13a1088e4c41d45d647533d1a61ead220e74 Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Tue, 22 Aug 2023 23:07:11 +0530
Subject: [PATCH 11/12] removed teardown and on.exit

---
 .../integrationTests/test.download.CRUNCEP.R  | 208 +++++++++---------
 1 file changed, 103 insertions(+), 105 deletions(-)

diff --git a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
index dbabb91be5c..e6e0a9a6356 100644
--- a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
+++ b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
@@ -3,115 +3,113 @@ library(ncdf4)
 
 # putting logger to debug mode
 PEcAn.logger::logger.setUseConsole(TRUE, FALSE)
-on.exit(PEcAn.logger::logger.setUseConsole(TRUE, TRUE), add = TRUE)
 PEcAn.logger::logger.setLevel("DEBUG")
 
 # mocking functions
 mockery::stub(PEcAn.DB::convert_input, 'dbfile.input.check', data.frame())
 mockery::stub(PEcAn.DB::convert_input, 'db.query', data.frame(id = 1))
 
-tmpdir <- tempfile(pattern = "CRUNCEPdata")
-dir.create(tmpdir)
-on.exit(teardown(unlink(tmpdir, recursive = TRUE)))
-
-PEcAn.DB::convert_input(
-  input.id = NA,
-  outfolder = tmpdir,
-  formatname = NULL,
-  mimetype = NULL,
-  site.id = 1,
-  start_date = "2000-01-01",
-  end_date = "2000-12-31",
-  pkg = 'PEcAn.data.atmosphere',
-  fcn = 'download.CRUNCEP',
-  con = NULL,
-  host = data.frame(name = "localhost"),
-  browndog = NULL,
-  write = FALSE,
-  lat.in = 40,
-  lon.in = -88,
-  method = "ncss",
-  maxErrors = 10,
-  sleep = 2
-)
-
-test_that("File exists at desired location", {
-  # Set the desired file path
-  file_path <- paste0(tmpdir, "/cruncep_landwater_mask.nc")
-  
-  # Check if file exists at desired location
-  expect_true(file.exists(file_path))
-})
-
-test_that("NetCDF file contains lat and lon variables", {
-
-  mask_nc <- ncdf4::nc_open(paste0(tmpdir, "/cruncep_landwater_mask.nc"))
-  on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
-  expect_true("land_water_mask" %in% names(mask_nc$var))
-  
-  # Check the dimensions of "land_water_mask" variable
-  expect_equal(mask_nc$var$land_water_mask$dim[[1]]$name, "lon")
-  expect_equal(mask_nc$var$land_water_mask$dim[[2]]$name, "lat")
-})
-
-test_that("All the required files are downloaded and stored at desired location", {
-  # Cached raw CRUNCEP files 
-  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc")))
-  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc")))
-  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc")))
-  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc")))
-  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc")))
-  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc")))
-  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc")))
-  expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc")))
-  
-  # CRUNCEP file
-  expect_true(file.exists(paste0(tmpdir, "/CRUNCEP.2000.nc")))
-})
-
-test_that("All cruncep raw data files have the correct variable units", {
-  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc"))
-  expect_equal(nc$var$tair$units, "K")
-  nc_close(nc)
-
-  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc"))
-  expect_equal(nc$var$lwdown$units, "W/m2")
-  nc_close(nc)
-
-  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc"))
-  expect_equal(nc$var$press$units, "Pa")
-  nc_close(nc)
-
-  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc"))
-  expect_equal(nc$var$swdown$units, "W/m2")
-  nc_close(nc)
-
-  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc"))
-  expect_equal(nc$var$uwind$units, "m/s")
-  nc_close(nc)
-
-  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc"))
-  expect_equal(nc$var$vwind$units, "m/s")
-  nc_close(nc)
-
-  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc"))
-  expect_equal(nc$var$qair$units, "g/g")
-  nc_close(nc)
-
-  nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc"))
-  expect_equal(nc$var$rain$units, "mm/6h")
-  nc_close(nc)
-})
-
-test_that("Combined data file has the correct variable units", {
-  nc <- nc_open(paste0(tmpdir, "/CRUNCEP.2000.nc"))
-  expect_equal(nc$var$air_temperature$units, "Kelvin")
-  expect_equal(nc$var$surface_downwelling_longwave_flux_in_air$units, "W/m2")
-  expect_equal(nc$var$air_pressure$units, "Pascal")
-  expect_equal(nc$var$surface_downwelling_shortwave_flux_in_air$units, "W/m2")
-  expect_equal(nc$var$eastward_wind$units, "m/s")
-  expect_equal(nc$var$northward_wind$units, "m/s")
-  expect_equal(nc$var$specific_humidity$units, "g/g")
-  expect_equal(nc$var$precipitation_flux$units, "kg/m2/s")
-  nc_close(nc)
+withr::with_dir(tempdir(), {
+  tmpdir <- getwd()
+  PEcAn.DB::convert_input(
+    input.id = NA,
+    outfolder = tmpdir,
+    formatname = NULL,
+    mimetype = NULL,
+    site.id = 1,
+    start_date = "2000-01-01",
+    end_date = "2000-12-31",
+    pkg = 'PEcAn.data.atmosphere',
+    fcn = 'download.CRUNCEP',
+    con = NULL,
+    host = data.frame(name = "localhost"),
+    browndog = NULL,
+    write = FALSE,
+    lat.in = 40,
+    lon.in = -88,
+    method = "ncss",
+    maxErrors = 10,
+    sleep = 2
+  )
+
+  test_that("File exists at desired location", {
+    # Set the desired file path
+    file_path <- paste0(tmpdir, "/cruncep_landwater_mask.nc")
+    
+    # Check if file exists at desired location
+    expect_true(file.exists(file_path))
+  })
+
+  test_that("NetCDF file contains lat and lon variables", {
+
+    mask_nc <- ncdf4::nc_open(paste0(tmpdir, "/cruncep_landwater_mask.nc"))
+    on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
+    expect_true("land_water_mask" %in% names(mask_nc$var))
+    
+    # Check the dimensions of "land_water_mask" variable
+    expect_equal(mask_nc$var$land_water_mask$dim[[1]]$name, "lon")
+    expect_equal(mask_nc$var$land_water_mask$dim[[2]]$name, "lat")
+  })
+
+  test_that("All the required files are downloaded and stored at desired location", {
+    # Cached raw CRUNCEP files 
+    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc")))
+    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc")))
+    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc")))
+    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc")))
+    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc")))
+    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc")))
+    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc")))
+    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc")))
+    
+    # CRUNCEP file
+    expect_true(file.exists(paste0(tmpdir, "/CRUNCEP.2000.nc")))
+  })
+
+  test_that("All cruncep raw data files have the correct variable units", {
+    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc"))
+    expect_equal(nc$var$tair$units, "K")
+    nc_close(nc)
+
+    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc"))
+    expect_equal(nc$var$lwdown$units, "W/m2")
+    nc_close(nc)
+
+    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc"))
+    expect_equal(nc$var$press$units, "Pa")
+    nc_close(nc)
+
+    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc"))
+    expect_equal(nc$var$swdown$units, "W/m2")
+    nc_close(nc)
+
+    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc"))
+    expect_equal(nc$var$uwind$units, "m/s")
+    nc_close(nc)
+
+    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc"))
+    expect_equal(nc$var$vwind$units, "m/s")
+    nc_close(nc)
+
+    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc"))
+    expect_equal(nc$var$qair$units, "g/g")
+    nc_close(nc)
+
+    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc"))
+    expect_equal(nc$var$rain$units, "mm/6h")
+    nc_close(nc)
+  })
+
+  test_that("Combined data file has the correct variable units", {
+    nc <- nc_open(paste0(tmpdir, "/CRUNCEP.2000.nc"))
+    expect_equal(nc$var$air_temperature$units, "Kelvin")
+    expect_equal(nc$var$surface_downwelling_longwave_flux_in_air$units, "W/m2")
+    expect_equal(nc$var$air_pressure$units, "Pascal")
+    expect_equal(nc$var$surface_downwelling_shortwave_flux_in_air$units, "W/m2")
+    expect_equal(nc$var$eastward_wind$units, "m/s")
+    expect_equal(nc$var$northward_wind$units, "m/s")
+    expect_equal(nc$var$specific_humidity$units, "g/g")
+    expect_equal(nc$var$precipitation_flux$units, "kg/m2/s")
+    nc_close(nc)
+  })
 })
\ No newline at end of file

From 4090ddc5787f51ff0b0ae9d76ae59ef140379727 Mon Sep 17 00:00:00 2001
From: meetagrawal09 <agrawalmeet91@gmail.com>
Date: Wed, 23 Aug 2023 16:28:01 +0530
Subject: [PATCH 12/12] moved tests to a function

---
 .../integrationTests/test.download.CRUNCEP.R  | 236 ++++++++++--------
 1 file changed, 125 insertions(+), 111 deletions(-)

diff --git a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
index e6e0a9a6356..8512b49f34d 100644
--- a/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
+++ b/modules/data.atmosphere/inst/integrationTests/test.download.CRUNCEP.R
@@ -1,115 +1,129 @@
 library(testthat)
 library(ncdf4)
 
-# putting logger to debug mode
-PEcAn.logger::logger.setUseConsole(TRUE, FALSE)
-PEcAn.logger::logger.setLevel("DEBUG")
-
-# mocking functions
-mockery::stub(PEcAn.DB::convert_input, 'dbfile.input.check', data.frame())
-mockery::stub(PEcAn.DB::convert_input, 'db.query', data.frame(id = 1))
-
-withr::with_dir(tempdir(), {
-  tmpdir <- getwd()
-  PEcAn.DB::convert_input(
-    input.id = NA,
-    outfolder = tmpdir,
-    formatname = NULL,
-    mimetype = NULL,
-    site.id = 1,
-    start_date = "2000-01-01",
-    end_date = "2000-12-31",
-    pkg = 'PEcAn.data.atmosphere',
-    fcn = 'download.CRUNCEP',
-    con = NULL,
-    host = data.frame(name = "localhost"),
-    browndog = NULL,
-    write = FALSE,
-    lat.in = 40,
-    lon.in = -88,
-    method = "ncss",
-    maxErrors = 10,
-    sleep = 2
-  )
-
-  test_that("File exists at desired location", {
-    # Set the desired file path
-    file_path <- paste0(tmpdir, "/cruncep_landwater_mask.nc")
-    
-    # Check if file exists at desired location
-    expect_true(file.exists(file_path))
+test_download_CRUNCEP <- function(start_date, end_date, lat.in, lon.in, method, maxErrors, sleep) {
+  # putting logger to debug mode
+  PEcAn.logger::logger.setUseConsole(TRUE, FALSE)
+  on.exit(PEcAn.logger::logger.setUseConsole(TRUE, TRUE), add = TRUE)
+  PEcAn.logger::logger.setLevel("DEBUG")
+
+  # mocking functions
+  mockery::stub(PEcAn.DB::convert_input, 'dbfile.input.check', data.frame())
+  mockery::stub(PEcAn.DB::convert_input, 'db.query', data.frame(id = 1))
+
+  withr::with_dir(tempdir(), {
+    tmpdir <- getwd()
+    PEcAn.DB::convert_input(
+      input.id = NA,
+      outfolder = tmpdir,
+      formatname = NULL,
+      mimetype = NULL,
+      site.id = 1,
+      start_date = start_date,
+      end_date = end_date,
+      pkg = 'PEcAn.data.atmosphere',
+      fcn = 'download.CRUNCEP',
+      con = NULL,
+      host = data.frame(name = "localhost"),
+      browndog = NULL,
+      write = FALSE,
+      lat.in = lat.in,
+      lon.in = lon.in,
+      method = method,
+      maxErrors = maxErrors,
+      sleep = sleep
+    )
+
+    test_that("File exists at desired location", {
+      # Set the desired file path
+      file_path <- paste0(tmpdir, "/cruncep_landwater_mask.nc")
+      
+      # Check if file exists at desired location
+      expect_true(file.exists(file_path))
+    })
+
+    test_that("NetCDF file contains lat and lon variables", {
+
+      mask_nc <- ncdf4::nc_open(paste0(tmpdir, "/cruncep_landwater_mask.nc"))
+      on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
+      expect_true("land_water_mask" %in% names(mask_nc$var))
+      
+      # Check the dimensions of "land_water_mask" variable
+      expect_equal(mask_nc$var$land_water_mask$dim[[1]]$name, "lon")
+      expect_equal(mask_nc$var$land_water_mask$dim[[2]]$name, "lat")
+    })
+
+    test_that("All the required files are downloaded and stored at desired location", {
+      # Cached raw CRUNCEP files 
+      expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc")))
+      expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc")))
+      expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc")))
+      expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc")))
+      expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc")))
+      expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc")))
+      expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc")))
+      expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc")))
+      
+      # CRUNCEP file
+      expect_true(file.exists(paste0(tmpdir, "/CRUNCEP.2000.nc")))
+    })
+
+    test_that("All cruncep raw data files have the correct variable units", {
+      nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc"))
+      expect_equal(nc$var$tair$units, "K")
+      nc_close(nc)
+
+      nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc"))
+      expect_equal(nc$var$lwdown$units, "W/m2")
+      nc_close(nc)
+
+      nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc"))
+      expect_equal(nc$var$press$units, "Pa")
+      nc_close(nc)
+
+      nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc"))
+      expect_equal(nc$var$swdown$units, "W/m2")
+      nc_close(nc)
+
+      nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc"))
+      expect_equal(nc$var$uwind$units, "m/s")
+      nc_close(nc)
+
+      nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc"))
+      expect_equal(nc$var$vwind$units, "m/s")
+      nc_close(nc)
+
+      nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc"))
+      expect_equal(nc$var$qair$units, "g/g")
+      nc_close(nc)
+
+      nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc"))
+      expect_equal(nc$var$rain$units, "mm/6h")
+      nc_close(nc)
+    })
+
+    test_that("Combined data file has the correct variable units", {
+      nc <- nc_open(paste0(tmpdir, "/CRUNCEP.2000.nc"))
+      expect_equal(nc$var$air_temperature$units, "Kelvin")
+      expect_equal(nc$var$surface_downwelling_longwave_flux_in_air$units, "W/m2")
+      expect_equal(nc$var$air_pressure$units, "Pascal")
+      expect_equal(nc$var$surface_downwelling_shortwave_flux_in_air$units, "W/m2")
+      expect_equal(nc$var$eastward_wind$units, "m/s")
+      expect_equal(nc$var$northward_wind$units, "m/s")
+      expect_equal(nc$var$specific_humidity$units, "g/g")
+      expect_equal(nc$var$precipitation_flux$units, "kg/m2/s")
+      nc_close(nc)
+    })
   })
-
-  test_that("NetCDF file contains lat and lon variables", {
-
-    mask_nc <- ncdf4::nc_open(paste0(tmpdir, "/cruncep_landwater_mask.nc"))
-    on.exit(ncdf4::nc_close(mask_nc), add = TRUE)
-    expect_true("land_water_mask" %in% names(mask_nc$var))
-    
-    # Check the dimensions of "land_water_mask" variable
-    expect_equal(mask_nc$var$land_water_mask$dim[[1]]$name, "lon")
-    expect_equal(mask_nc$var$land_water_mask$dim[[2]]$name, "lat")
-  })
-
-  test_that("All the required files are downloaded and stored at desired location", {
-    # Cached raw CRUNCEP files 
-    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc")))
-    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc")))
-    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc")))
-    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc")))
-    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc")))
-    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc")))
-    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc")))
-    expect_true(file.exists(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc")))
-    
-    # CRUNCEP file
-    expect_true(file.exists(paste0(tmpdir, "/CRUNCEP.2000.nc")))
-  })
-
-  test_that("All cruncep raw data files have the correct variable units", {
-    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-tair.nc"))
-    expect_equal(nc$var$tair$units, "K")
-    nc_close(nc)
-
-    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-lwdown.nc"))
-    expect_equal(nc$var$lwdown$units, "W/m2")
-    nc_close(nc)
-
-    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-press.nc"))
-    expect_equal(nc$var$press$units, "Pa")
-    nc_close(nc)
-
-    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-swdown.nc"))
-    expect_equal(nc$var$swdown$units, "W/m2")
-    nc_close(nc)
-
-    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-uwind.nc"))
-    expect_equal(nc$var$uwind$units, "m/s")
-    nc_close(nc)
-
-    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-vwind.nc"))
-    expect_equal(nc$var$vwind$units, "m/s")
-    nc_close(nc)
-
-    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-qair.nc"))
-    expect_equal(nc$var$qair$units, "g/g")
-    nc_close(nc)
-
-    nc <- nc_open(paste0(tmpdir, "/cruncep-raw-2000-40--88-rain.nc"))
-    expect_equal(nc$var$rain$units, "mm/6h")
-    nc_close(nc)
-  })
-
-  test_that("Combined data file has the correct variable units", {
-    nc <- nc_open(paste0(tmpdir, "/CRUNCEP.2000.nc"))
-    expect_equal(nc$var$air_temperature$units, "Kelvin")
-    expect_equal(nc$var$surface_downwelling_longwave_flux_in_air$units, "W/m2")
-    expect_equal(nc$var$air_pressure$units, "Pascal")
-    expect_equal(nc$var$surface_downwelling_shortwave_flux_in_air$units, "W/m2")
-    expect_equal(nc$var$eastward_wind$units, "m/s")
-    expect_equal(nc$var$northward_wind$units, "m/s")
-    expect_equal(nc$var$specific_humidity$units, "g/g")
-    expect_equal(nc$var$precipitation_flux$units, "kg/m2/s")
-    nc_close(nc)
-  })
-})
\ No newline at end of file
+}
+
+
+test_download_CRUNCEP(
+  start_date = "2000-01-01",
+  end_date = "2000-12-31",
+  lat.in = 40,
+  lon.in = -88,
+  method = "ncss",
+  maxErrors = 10,
+  sleep = 2
+)
\ No newline at end of file