comments added to guide new developers

podaac/subsetter/gpm_cleanup.py

@@ -1,6 +1,11 @@
 """
-Module designed for mapping the dimensions in GPM. Phony dimensions are changed
-to nscan, nbin, nfreq by using the DimensionNames variable attribute
+Module designed for mapping the dimensions in GPM, and changing calendar time type
+to seconds since 1980-1-6 T00:00:00Z
+Problem: each variable will have unique phony dims. When time for subset_with_bbox
+is called, each unique group will have to have common dimensions. 
+Solution: Phony dimensions are changed to nscan, nbin, nfreq and will have the same name for each
+unique group that will be subsetted.
+nscan, nbin, and nfreq are named in the variable attributes.
 """
 
 import datetime
@@ -14,9 +19,10 @@ def compute_new_time_data(time_group, nc_dataset):
  create a time variable, timeMidScan, that is present in other
  GPM collections but not the ENV collections.
  """
- # set the time unit for GPM
+ # set the time unit for GPM - this is the case for all GPM collections
  time_unit_out = "seconds since 1980-01-06 00:00:00"
- # conver to a float, seconds variable
+ # convert to a float variable, seconds since the above date
+ # this list will be a new time variable
  new_time_list = [date2num(datetime.datetime(
  nc_dataset[time_group+'__Year'][:][i],
  nc_dataset[time_group+'__Month'][:][i],
@@ -37,7 +43,7 @@ def change_var_dims(nc_dataset, variables=None, time_name="__timeMidScan"):
  dimensions to have the name in the DimensionName attribute rather than phony_dim
  """
  var_list = list(nc_dataset.variables.keys())
- # loop through variable list to avoid netcdf4 runtime error
+ # loop through the entire variable list to avoid netcdf4 runtime error
  for var_name in var_list:
  # GPM will always need to be cleaned up via netCDF
  # generalizing coordinate variables in netCDF file to speed variable subsetting up
@@ -57,14 +63,16 @@ def change_var_dims(nc_dataset, variables=None, time_name="__timeMidScan"):
  dim_prefix = var_name.split('__')[1]
  # new dimension name
  new_dim = '__'+dim_prefix+'__'+dim
- length = var.shape[count]
+ # get dim size of the newly created variable
+ dim_size = var.shape[count]
  # check if the dimension name created has already been created in the dataset
  if new_dim not in dim_dict:
  # create the new dimension
- nc_dataset.createDimension(new_dim, length)
- dim_dict[new_dim] = length
+ nc_dataset.createDimension(new_dim, dim_size)
+ dim_dict[new_dim] = dim_size
  # utilized from Dimension Cleanup module
  attrs_contents = {}
+ # new variable that will point to the updated dimensions
  new_mapped_var = {}
  # if the variable has attributes, get the attributes to then be copied to the new variable
  if len(var.ncattrs()) > 0:
@@ -99,7 +107,7 @@ def change_var_dims(nc_dataset, variables=None, time_name="__timeMidScan"):
  comp_args = {"zlib": True, "complevel": 1}
  nc_dataset.createVariable(new_time_var_name, 'f8', var_dims, **comp_args)
  nc_dataset.variables[new_time_var_name].setncattr('unit', time_unit)
- # copy the data in
+ # copy the new time list into the new time variable
  nc_dataset.variables[new_time_var_name][:] = time_data
 
  return nc_dataset

podaac/subsetter/group_handling.py

@@ -108,7 +108,7 @@ def recombine_grouped_datasets(datasets: List[xr.Dataset], output_file: str, sta
  Name of the output file to write the resulting NetCDF file to.
  TODO: add docstring and type hint for `start_date` parameter.
  """
-
+ # 
  base_dataset = nc.Dataset(output_file, mode='w')
  for dataset in datasets:
  group_lst = []
@@ -179,7 +179,7 @@ def _rename_variables(dataset: xr.Dataset, base_dataset: nc.Dataset, start_date,
  comp_args = {"zlib": True, "complevel": 1}
 
  var_data = variable.data
-
+ # create variable based upon the original data type
  if variable.dtype in [object, '|S27']:
  comp_args = {"zlib": False, "complevel": 1}
  var_group.createVariable(new_var_name, 'S4', var_dims, fill_value=fill_value, **comp_args)

podaac/subsetter/time_converting.py

@@ -22,7 +22,8 @@
 
 def get_start_date(instrument_type):
  """
- returns the start date based on the instrument type
+ returns the start date based on the instrument type. Start date will calculate
+ seconds since this start date.
  """
  if instrument_type in ['OMI', 'MLS']:
  start_date = datetime.datetime.strptime("1993-01-01T00:00:00.00", "%Y-%m-%dT%H:%M:%S.%f")
@@ -51,6 +52,7 @@ def convert_to_datetime(dataset: xr.Dataset, time_vars: list, instrument_type: s
  dataset[var].values = date_time_array.astype("datetime64[ns]") + dataset[var].astype('timedelta64[s]').values
  continue
  # if there isn't a start_date, get it from the UTC variable
+ # subtracts the seconds from the first UTC time to get the date that seconds are from
  utc_var_name = subset.compute_utc_name(dataset)
  if utc_var_name:
  start_seconds = dataset[var].values[0]