hagelslag.data package

Submodules

hagelslag.data.HRRRModelGrid module

class hagelslag.data.HRRRModelGrid.HRRRModelGrid(run_date, variable, start_date, end_date, path, frequency='1H')

Bases: ModelGrid

hagelslag.data.HailForecastGrid module

class hagelslag.data.HailForecastGrid.HailForecastGrid(run_date, start_date, end_date, ensemble_name, ml_model, members, variable, message_number, path)

Bases: object

HailForecastGrid loads and stores gridded machine learning hail forecasts from GRIB2 files. It can load an arbitrary number of members and timesteps at once.

run_date

Date of the initial time of the model run

Type:

datetime.datetime

start_date

Date of the initial forecast time being loaded

Type:

datetime.datetime

end_date

Date of the final forecast time being loaded

Type:

datetime.datetime

forecast_dates

All forecast times

Type:

pandas.DatetimeIndex

ensemble_name

Name of the NWP ensemble being used

Type:

str

ml_model

Name of the machine learning model being loaded

Type:

str

variable

Name of the machine learning model variable being forecast

Type:

str

message_number

Field in the GRIB2 file to load. The first field in the file has message number 1.

Type:

int

path

Path to top-level GRIB2 directory. Assumes files are stored in directories by run_date

Type:

str

data

Hail forecast data with dimensions (member, time, y, x)

Type:

ndarray

lon

2D array of longitudes

Type:

ndarray

lat

2D array of latitudes

Type:

ndarray

x

2d array of x-coordinate values in km

Type:

ndarray

y

2d array of y-coordinate values in km

Type:

ndarray

i

2d array of row indices

Type:

ndarray

j

2d array of column indices

Type:

ndarray

dx

distance between grid points

Type:

float

proj

a pyproj projection object used for converting lat-lon points to x-y coordinate values

Type:

Proj

projparams

PROJ4 parameters describing map projection

Type:

dict

load_data()
period_neighborhood_probability(radius, smoothing, threshold, stride, start_time, end_time)

Calculate the neighborhood probability over the full period of the forecast

Parameters:

  • radius – circular radius from each point in km

  • smoothing – width of Gaussian smoother in km

  • threshold – intensity of exceedance

  • stride – number of grid points to skip for reduced neighborhood grid

Returns:

(neighborhood probabilities)

hagelslag.data.MRMSGrid module

class hagelslag.data.MRMSGrid.MRMSGrid(start_date, end_date, variable, path, freq='1H')

Bases: object

An interface to the NOAA National Severe Storms Lab Multi-Radar Multi-Sensor (MRMS) dataset.

MRMSGrid assumes that the data are in netCDF format and have been interpolated to match the grid being used for forecasting.

Parameters:

  • start_date (datetime.datetime or time str) – Date of first time step to be loaded.

  • end_date (datetime.datetime or str in timestamp format) – Date of last time step to be loaded.

  • variable (str) – MRMS variable name

  • path (str) – Path to the directory containing MRMS files.

  • freq (str, optional (default="1H")) – Time frequency of the data being loaded. Uses pandas time syntax.

start_date

Date of first time step to be loaded.

Type:

datetime.datetime or time str

end_date

Date of last time step to be loaded.

Type:

datetime.datetime or str in timestamp format

variable

MRMS variable name

Type:

str

path

Path to the directory containing MRMS files.

Type:

str

freq

Time frequency of the data being loaded. Uses pandas time syntax.

Type:

str, optional (default=”1H”)

all_dates

pandas.DatetimeIndex

List of dates being loaded
data

Array of gridded observations after load_data is called. None otherwise.

Type:

ndarray or None

valid_dates

Contains the dates where data loaded successfully.

Type:

ndarray

load_data()

Loads data files and stores the output in the data attribute.

period_neighborhood_probability(radius, smoothing, threshold, stride, x, y, dx)

Calculate the neighborhood probability over the full period of the forecast

Parameters:

  • radius – circular radius from each point in km

  • smoothing – width of Gaussian smoother in km

  • threshold – intensity of exceedance

  • stride – number of grid points to skip for reduced neighborhood grid

  • x – x-coordinate array in km

  • y – y-coordinate array in km

  • dx – distance between grid points in km

Returns:

neighborhood probablities

hagelslag.data.ModelGrid module

class hagelslag.data.ModelGrid.ModelGrid(filenames, run_date, start_date, end_date, variable, frequency='1H')

Bases: object

Base class for reading 2D model output grids from netCDF files.

Given a list of file names, loads the values of a single variable from a model run. Supports model output in netCDF format.

filenames

List of netCDF files containing model output

Type:

list of str

run_date

Date of the initialization time of the model run.

Type:

ISO date string or datetime.datetime object

start_date

Date of the first timestep extracted.

Type:

ISO date string or datetime.datetime object

end_date

Date of the last timestep extracted.

Type:

ISO date string or datetime.datetime object

freqency

spacing between model time steps.

Type:

str

valid_dates

DatetimeIndex of all model timesteps

forecast_hours

array of all hours in the forecast

file_objects

List of the file objects for each model time step

Type:

list

close()

Close links to all open file objects and delete the objects.

static format_var_name(variable, var_list)

Searches var list for variable name, checks other variable name format options.

Parameters:

  • variable (str) – Variable being loaded

  • var_list (list) – List of variables in file.

Returns:

Name of variable in file containing relevant data, and index of variable z-level if multiple variables contained in same array in file.

load_data()

Load data from netCDF file objects or list of netCDF file objects. Handles special variable name formats.

Returns:

Array of data loaded from files in (time, y, x) dimensions, Units

load_data_old()

Loads time series of 2D data grids from each opened file. The code handles loading a full time series from one file or individual time steps from multiple files. Missing files are supported.

hagelslag.data.ModelOutput module

hagelslag.data.NCARModelGrid module

class hagelslag.data.NCARModelGrid.NCARModelGrid(member, run_date, variable, start_date, end_date, path, single_step=False)

Bases: ModelGrid

Extension of the ModelGrid class for interfacing with the NCAR ensemble.

Parameters:

  • member (str) – Name of the ensemble member

  • run_date (datetime.datetime object) – Date of the initial step of the ensemble run

  • start_date (datetime.datetime object) – First time step extracted.

  • end_date (datetime.datetime object) – Last time step extracted.

  • path (str) – Path to model output files.

  • single_step (boolean (default=False)) – Whether variable information is stored with each time step in a separate file or one file containing all timesteps.

hagelslag.data.SSEFModelGrid module

class hagelslag.data.SSEFModelGrid.SSEFModelGrid(member, run_date, variable, start_date, end_date, path, single_step=False)

Bases: ModelGrid

Extension of ModelGrid to the CAPS Storm-Scale Ensemble Forecast system.

Parameters:

  • member (str) – Name of the ensemble member

  • run_date (datetime.datetime object) – Date of the initial step of the ensemble run

  • start_date (datetime.datetime object) – First time step extracted.

  • end_date (datetime.datetime object) – Last time step extracted.

  • path (str) – Path to model output files.

  • single_step (boolean (default=False)) – Whether variable information is stored with each time step in a separate file or one file containing all timesteps.

hagelslag.data.WRFModelGrid module

class hagelslag.data.WRFModelGrid.WRFModelGrid(forecast_date, variable, domain, path, nocolons=False)

Bases: object

Load data from raw WRF model output files.

get_global_attributes()
load_full_grid()
load_time_var(time_var='XTIME')

Module contents