SPAREICE

class typhon.retrieval.spareice.SPAREICE(file=None, collocator=None, processes=10, verbose=0, sea_mask_file=None, elevation_file=None)[source]

Retrieval of IWP from passive radiometers

Examples: .. code-block:: python

import pandas as pd from typhon.retrieval import SPAREICE

# Create a SPARE-ICE object with the standard weights spareice = SPAREICE()

# Print the required input fields print(spareice.inputs)

# If you want to know the input fields for the each component, IWP # regressor and ice cloud classifier, you can get them like this: print(spareice.iwp.inputs) # Inputs from IWP regressor print(spareice.ice_cloud.inputs) # Inputs from ice cloud classifier

# If you have yur own input data, you can use retrieve() to run # SPARE-ICE on it. data = pd.DataFrame(…) retrieved = spareice.retrieve(data)

# If your data directly comes from collocations between MHS and AVHRR, # you can use :convert_collocated_data() to make it SPARE-ICE # compatible. collocations = Collocator().collocate(mhs_data, avhrr_data, …) standardized_data = self.standardize_collocations(collocations) retrieved = spareice.retrieve(standardized_data)

__init__(file=None, collocator=None, processes=10, verbose=0, sea_mask_file=None, elevation_file=None)[source]

Initialize a SPAREICE object

Parameters:

  • file – A JSON file with the coefficients of SPAREICE. If not given, the standard configuration will be loaded.

  • collocator – SPARE-ICE requires a collocator when it should be generated from filesets. You can pass your own Collocator object here if you want.

  • processes – Number of processes to parallelize the training or collocation search. 10 is the default. Best value depends on your machine.

  • verbose (int) – Control GridSearchCV verbosity. The higher the

  • value

  • printed. (the more debug messages are)

Methods

__init__([file, collocator, processes, ...])

Initialize a SPAREICE object

load(filename)

Load SPARE-ICE from a json file

report(output_dir, experiment, data)

Test the performance of SPARE-ICE and plot it

retrieve(data[, as_log10])

Retrieve SPARE-ICE for the input variables

retrieve_from_collocations(inputs, output[, ...])

Retrieve SPARE-ICE from collocations between MHS and AVHRR

save(filename)

Save SPARE-ICE to a json file

score(data)

Calculate the score of SPARE-ICE on testing data

standardize_collocations(data[, fields, ...])

Convert collocation fields to standard SPARE-ICE fields.

train(data[, iwp_inputs, ice_cloud_inputs, ...])

Train SPARE-ICE with data

Attributes

ice_cloud

Return the ice cloud classifier of SPARE-ICE

inputs

Return the input fields of the current configuration

iwp

Return the IWP regressor of SPARE-ICE